CATALYST DETERIORATION DETECTING SYSTEM OF INTERNAL

COMBUSTION ENGINE

BACKGROUND OF THE INVENTION 1. Field of the Invention

[0001] The invention relates to a catalyst deterioration detecting system that detects deterioration of a catalyst disposed in an exhaust passage of an internal combustion engine.

2. Description of the Related Art

[0002] Generally, a catalyst is disposed in an exhaust passage of an internal combustion engine so as to clean exhaust gas or engine emissions. The catalyst, for example, a three-way catalyst, has an O2 storage function of absorbing and releasing oxygen depending on the air-fuel ratio of exhaust gas flowing into the catalyst. More specifically, the catalyst absorbs excessive oxygen present in the exhaust gas when the air-fuel ratio of the exhaust gas flowing into the catalyst becomes larger than the stoichiometric air-fuel ratio, namely, becomes lean, and releases the absorbed oxygen when the air-fuel ratio becomes smaller than the stoichiometric ratio, namely, becomes rich. Accordingly, even if the air-fuel ratio of a fuel-air mixture supplied to the engine swings to the rich side or lean side relative to the stoichiometric ratio depending on operating conditions during normal operation of the engine, the surface of the catalyst is substantially kept at the stoichiometric ratio, and the three-way catalyst having the O2 storage function is able to convert NOx, HC and CO into harmless substances at the same time. More specifically, when the fuel-air mixture becomes lean, excessive oxygen is adsorbed onto and held by the catalyst, so that NOx is reduced. When the mixture becomes rich, the oxygen adsorbed and stored in the catalyst is released from the catalyst, so that HC and CO are oxidized.

[0003] In a conventional system, an air-fuel ratio sensor for detecting the

exhaust air-fuel ratio is disposed in an exhaust passage upstream of the catalyst. In operation, the amount of fuel supplied to the engine is increased when the exhaust air-fuel ratio becomes lean, and is reduced when the exhaust air-fuel ratio becomes rich, so that the air-fuel ratio is controlled to or around the stoichiometric air-fuel ratio. Thus, even if the air-fuel ratio fluctuates between the rich side and the lean side, the amounts of NOx, HC and CO can be reduced at the same time.

[0004] In the meantime, the catalytic conversion efficiency with which the exhaust gas is cleaned up is reduced as the three-way catalyst deteriorates. There is a correlation between the degree of deterioration of the three-way catalyst and the degree of reduction of the O2 storage function since both of the catalytic action and the O2 storage function involve chemical reaction through the medium of a noble metal. It is thus possible to detect deterioration of the catalyst by detecting reduction of the O2 storage function.

[0005] A system for detecting catalyst deterioration according to the above principle is disclosed in, for example, Japanese Patent Application Publication No. 5-133264 (JP"A"5- 133264). In this system, a first air-fuel ratio sensor is disposed in an exhaust passage upstream of a three-way catalyst, and a second air-fuel ratio sensor is disposed in an exhaust passage downstream of the three-way catalyst. An air-fuel ratio switching device is also provided for switching the air-fuel ratio detected upstream of the catalyst from a lean-side air-fuel ratio to a rich-side air-fuel ratio or vice versa relative to the stoichiometric air-fuel ratio. If the air-fuel ratio of a fuel-air mixture supplied into the engine cylinders is switched from a lean air-fuel ratio to a rich air-fuel ratio, the air-fuel ratio detected by the second airfuel ratio sensor is switched to a rich air-fuel ratio after a lapse of a certain period of time. The absolute amount of oxygen absorbed by the three-way catalyst is obtained from the product of a deviation of the rich air-fuel ratio from the stoichiometric ratio, the time it takes from switching of the air-fuel ratio of the fuel-air mixture to switching of the air-fuel ratio detected by the second air-fuel ratio sensor, and the intake air amount, and the degree of deterioration of the

three-way catalyst is determined from the absolute amount of the absorbed oxygen.

[0006] As another example of the related art, Japanese Patent Application

Publication No. 5-263686 (JP-A-5-263686) discloses a technology of detecting deterioration of the catalyst by using the locus length of the output of an O2 sensor disposed downstream of the catalyst. As a further example of the related art, Japanese Patent Application Publication No. 5-312025 (JP-A-5-312025) discloses a catalyst deterioration diagnosis system in which the average air-fuel ratio is forced to be shifted to the rich side, and the degree of deterioration of the catalyst is determined from the amount of shift of the air-fuel ratio at which the output of an O2 sensor downstream of the catalyst ceases to be reversed to the opposite side. Japanese Patent Application Publication No. 10-47141 (JP-A-10-47141) discloses a technology of determining deterioration of the catalyst by monitoring variations in the output of an O2 sensor downstream of the catalyst while varying the air-fuel ratio upstream of the catalyst. Japanese Patent Application Publication No. 2005-180201 (JP-A-2005"18020l) discloses a catalyst deterioration diagnosis system in which the air-fuel ratio is initially shifted to the rich side until the amount of oxygen stored in the catalyst becomes substantially zero. Japanese Patent Application Publication No. 2001-304032 discloses a catalyst deterioration detecting system in which, during determination of deterioration of a catalyst in a hybrid vehicle, the output of the engine is set to within a predetermined range, and the output of an electric motor is changed in response to changes in the total power needed by the vehicle as a whole.

[0007] In the method as disclosed in JP-A-5-133264, or so-called Cmax method, in which the degree of deterioration of the catalyst is determined from the absolute amount of oxygen absorbed by the catalyst, exhaust gas that has not been cleaned or treated by the catalyst may be discharged to the downstream of the catalyst (so-called "breakthrough" may occur) before the catalyst absorbs oxygen to the limit of its own O2 storage capability. If the "breakthrough" of gas occurs, the output of the sensor downstream of the catalyst may be switched or reversed more

quickly, and the calculated oxygen absorption amount may become smaller than a true value. As a result, it may be erroneously determined that the catalyst is deteriorated. This problem is more likely to occur particularly when the flow rate of exhaust gas supplied to the catalyst is large. This is because the speed of reaction of the catalyst does not catch up with the speed of supply of the exhaust gas.

[0008] In the known Cmax method, there is a fundamental problem that it is difficult to use the utmost O2 storage capability possessed by the catalyst. Namely, in the known Cmax method, exhaust gas having a constant lean air-fuel ratio is continuously supplied to the catalyst when the catalyst is caused to absorb oxygen. In this way, however, only an upstream portion of the catalyst that is close to its surface can be used for absorbing oxygen. Accordingly, the above-mentioned "breakthrough" of the gas is likely to occur, which makes it difficult to detect catalyst deterioration with high accuracy. It is also difficult to precisely determine the degree of deterioration of a catalyst that is deteriorated to some extent and has a reduced O2 storage capability. It is also difficult to distinguish between a catalyst that has already been deteriorated and a catalyst that is about to be deteriorated.

SUMMARY OF THE INVENTION

[0009] The invention was developed in view of the actual situation as described above, and provides a catalyst determination detecting system of an internal combustion engine, which is able to recognize even a small difference in the degree of deterioration of the catalyst with high accuracy, by utilizing the utmost O2 storage capability possessed by the catalyst.

[OOIO] According to one aspect of the invention, a catalyst deterioration detecting system of an internal combustion engine including a catalyst that is disposed in an exhaust passage of the engine and has an O2 storage function of absorbing and releasing oxygen depending on whether an exhaust air-fuel ratio is

larger or smaller than the stoichiometric air-fuel ratio is provided which includes an active air-fuel ratio control unit that oscillates the exhaust air-fuel ratio between a lean side and a rich side relative to a central air-fuel ratio in a predetermined determination period. In this system, when the catalyst is caused to absorb oxygen, the exhaust air-fuel ratio is oscillated such that the exhaust air-fuel ratio shifted to the rich side is equal to or smaller than the stoichiometric air-fuel ratio, and such that a degree of leanness per cycle of oscillation is larger than a degree of richness. When the catalyst is caused to release oxygen, the exhaust air-fuel ratio is oscillated such that the exhaust air-fuel ratio shifted to the lean side is equal to or larger than the stoichiometric air-fuel ratio, and such that the degree of richness per cycle of oscillation is larger than the degree of leanness.

[OOll] Here, the degree of leanness per cycle of oscillation may be represented by a value obtained by integrating a difference between the exhaust air-fuel ratio and the stoichiometric air-fuel ratio, over the length of a lean-side period in which the exhaust air-fuel ratio is shifted to the lean side relative to the central air-fuel ratio. Similarly, the degree of richness per cycle of oscillation may be represented by a value obtained by integrating a difference between the exhaust air-fuel ratio and the stoichiometric ratio, over the length of a rich-side period in which the exhaust air-fuel ratio is shifted to the rich side relative to the central air-fuel ratio.

[0012] In the catalyst deterioration detecting system of the internal combustion engine as described above, the active air-fuel ratio control unit may set the central air-fuel ratio to a value that is larger than the stoichiometric air-fuel ratio when the catalyst is caused to absorb oxygen, and may set the central air-fuel ratio to a value that is smaller than the stoichiometric air-fuel ratio when the catalyst is caused to release oxygen.

[0013] In the catalyst deterioration detecting system of the internal combustion engine as described above, the active air-fuel ratio control unit may change the central air-fuel ratio in the determination period.

[0014] In the catalyst deterioration detecting system of the internal combustion engine as described above, the active air-fuel ratio control unit may gradually change the central air-fuel ratio such that the central air-fuel ratio passes the stoichiometric air-fuel ratio during the change thereof.

5 [0015] In the catalyst deterioration detecting system of the internal combustion engine as described above, an intake air amount detector that detects an intake air amount may be further provided, and the active air-fuel ratio control unit may increase the rate of change of the central air-fuel ratio as the intake air amount increases, and may reduce the rate of change of the central air-fuel ratio as .0 the intake air amount decreases.

[0016] In the catalyst deterioration detecting system of the internal combustion engine as described above, the active air-fuel ratio control unit may change the central air-fuel ratio by a predetermined amount each time the exhaust air-fuel ratio is reversed from one of the lean side and the rich side relative to the L5 central air-fuel ratio to the other side or vice versa a predetermined number of times.

[0017] In the catalyst deterioration detecting system of the internal combustion engine as described above, the active air-fuel ratio control unit may shorten the period of the oscillation of the exhaust air-fuel ratio as the intake air

20 amount increases, and may extend the period of the oscillation as the intake air amount decreases.

[0018] In the catalyst deterioration detecting system of the internal combustion engine as described above, the active air-fuel ratio control unit may reverse the exhaust air-fuel ratio from one of the lean side and rich side relative to 5 the central air-fuel ratio to the other side each time a total value of the intake air amount reaches a predetermined value.

[0019] The catalyst deterioration detecting system of the internal combustion engine as described above may further include a post-catalyst sensor that detects an exhaust air-fuel ratio downstream of the catalyst, and a

locust-length deterioration determining unit that calculates a locus length of an output of the post-catalyst sensor in the determination period, and makes a determination on deterioration of the catalyst based on the locus length.

[0020] In the catalyst deterioration detecting system of the internal combustion engine as described above, the active air-fuel ratio control unit may oscillate the exhaust air-fuel ratio such that a lean-side period per cycle of oscillation is longer than a rich-side period when the catalyst is caused to absorb oxygen, and may oscillate the exhaust air-fuel ratio such that the rich-side period per cycle of oscillation is longer than the lean-side period when the catalyst is caused to release oxygen.

[002l] In the catalyst deterioration detecting system of the internal combustion engine as described above, an intake air amount detector that detects an intake air amount may be further provided, and the active air-fuel ratio control unit may set the rich-side period and the lean-side period to be closer in length to each other as the intake air amount detected by the intake air amount detector increases.

[0022] In the catalyst deterioration detecting system of the internal combustion engine as described above, an intake air amount detector that detects an intake air amount may be further provided, and the active air-fuel ratio control may increase the frequency of the oscillation as the intake air amount detected by the intake air amount detector increases.

[0023] The catalyst deterioration detecting system of the internal combustion engine as described above may further include an oxygen amount calculating unit that calculates an amount of oxygen absorbed into or released from the catalyst in the determination period, and an oxygen-amount deterioration determining unit that makes a determination on deterioration of the catalyst based on the amount of oxygen calculated by the oxygen amount calculating unit.

[0024] The catalyst deterioration detecting system of the internal combustion engine as described above may be applied to a hybrid vehicle including

the engine and a motor- generator as power sources. This system may further include a fuel cut request time control unit that inhibits fuel cut control and switches the motor generator to a generator mode in which the motor generator operates as a generator when a fuel cut request signal for requesting fuel cut

5 control is produced in the determination period.

[0025] With the catalyst deterioration detecting system of the internal combustion engine constructed as described above, when the catalyst is caused to absorb oxygen, lean gas can be intermittently supplied to the catalyst while the lean/rich balance of exhaust gas supplied to the catalyst is biased toward the lean

0 side. When the catalyst is caused to release oxygen, rich gas can be intermittently supplied to the catalyst while the lean/rich balance of exhaust gas supplied to the catalyst is biased toward the rich side. It is thus possible to cause the catalyst to absorb and release oxygen while using the utmost oxygen storage capability possessed by the catalyst. Also, the true oxygen storage capability of the catalyst

.5 can be measured with high accuracy, by slowly changing the O2 storage condition of the catalyst and using downstream and deeper portions of the catalyst for absorption and release of oxygen. This makes it possible to precisely determine the degree of deterioration of a catalyst that has already deteriorated to some extent, and also makes it possible to distinguish with high accuracy between a

JO catalyst that has been deteriorated and a catalyst that is about to be deteriorated. [0026] Thus, the catalyst deterioration detecting system of the internal combustion engine according to the invention is able to recognize even a small difference in the degree of deterioration of the catalyst with high accuracy, by causing the catalyst to exhibit the utmost O2 storage capability thereof.

Ih

BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The features, advantages, and technical and industrial significance of this invention will be better understood by reading the following detailed description of preferred embodiments of the invention, when considered in

connection with the accompanying drawings, in which:

FIG. 1 is a schematic view showing the basic construction of an internal combustion engine to which embodiments of the invention are applied;

FIG. 2 is a schematic cross-sectional view showing the structure of a 5 catalyst;

FIG. 3 is a graph useful for explaining the degree of leanness and the degree of richness;

FIG. 4 is a time chart showing changes in values associated with a catalyst deterioration detecting system according to a first embodiment of the invention; 0 FIG. 5 is a schematic view useful for explaining a mechanism of absorption and release of oxygen into and from the catalyst;

FIG. 6 is a schematic view useful for explaining a mechanism of absorption and release of oxygen into and from the catalyst;

FIG. 7 is a schematic view useful for explaining a mechanism of absorption 5 and release of oxygen into and from the catalyst;

FIG. 8 is a graph showing a test result, the upper stage of which shows changes in the pre "catalyst air-fuel ratio, and the lower stage of which shows changes in the output value of a post-catalyst sensor;

FIG. 9 is a graph showing a test result, i.e., the result of comparison in !0 terms of the locus length between a small- deterioration catalyst and a large-deterioration catalyst;

FIG. 10 is a graph showing a test result, i.e., the result of comparison in terms of the locus length between a small- deterioration catalyst and a large -deterioration catalyst when the lean-side amplitude and the rich-side .5 amplitude are changed;

FIG. 11 is a graph showing a test result, i.e., the result of comparison in terms of the locus length between a small- deterioration catalyst and a large-deterioration catalyst when the intake air amount is changed;

FIG. 12 is a graph showing a test result, i.e., the result of comparison in

terms of the locus length between a small-deterioration catalyst and a large -deterioration catalyst when the frequency of oscillation is changed;

FIG. 13 is a time chart showing changes in values associated with a catalyst deterioration detecting system of a first modified example of the first embodiment!

FIG. 14 is a time chart useful for explaining active air-fuel ratio control performed by a catalyst deterioration detecting system according to a second modified example of the first embodiment;

FIG. 15 is a graph useful for explaining a method of reversing a target air-fuel ratio in the second modified example, ^"

FIGS. 16A and 16B are flowcharts of a main routine executed when a deterioration detecting process according to a third case of the second modified example is performed;

FIG. 17 is a flowchart of a subroutine executed during execution of the main routine, ^'

FIG. 18 is a time chart useful for explaining active air-fuel ratio control performed by a catalyst deterioration detecting system of another modified example of the first embodiment;

FIG. 19 is a time chart useful for explaining active air-fuel ratio control performed by a catalyst deterioration detecting system of a further modified example of the first embodiment, ^"

FIG. 20 is a time chart showing changes in values associated with a catalyst deterioration detecting system according to a second embodiment of the invention; FIGS. 21A and 21B are flowcharts of a main routine executed when a deterioration detecting process according to the second embodiment is performed;

FIG. 22 is a flowchart of a subroutine executed during execution of the main routine;

FIG. 23 is a flowchart of a subroutine executed during execution of the

main routine;

FIG. 24 is a map that defines the relationship between the intake air amount and a target oxygen amount correction factor, ^" and

FIG. 25 is a map that defines the relationship between the intake air 5 amount and a target oxygen amount.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0028] In the following description and the accompanying drawings, the present invention will be described in more detail with reference to exemplary .0 embodiments.

[0029] Initially, the basic construction of an internal combustion engine having a catalyst deterioration detecting system according to each of the exemplary embodiments will be described with reference to FIG. 1. As shown in FIG. 1, the engine 1 produces power by burning a mixture of fuel and air within a L5 combustion chamber 3 formed in a cylinder block 2, and causing a piston 4 to reciprocate in the combustion chamber 3. The engine 1 is a multi- cylinder engine (only one cylinder of which is illustrated in FIG. l) for a vehicle, more specifically, a spark ignition type gasoline engine.

[0030] In a cylinder head of the engine 1, an intake valve Vi for opening 20 and closing an intake port and an exhaust valve Ve for opening and closing an exhaust port are provided for each cylinder. The intake valve Vi and exhaust valve Ve for each cylinder are opened and closed by respective camshafts (not shown). An ignition plug 7 operable to ignite the fuel-air mixture in the combustion chamber 3 is mounted in the cylinder head for each cylinder, such that 5 the ignition plug 7 is exposed to a top portion of the combustion chamber 3. An injector (fuel injection valve) 12 is also mounted in the cylinder head for each cylinder, and is operable to directly inject fuel into the combustion chamber 3. The piston 4 is a so-called chambered piston having a recess 4a formed in the top face thereof. In the engine 1, while air is drawn into the combustion chamber 3,

the fuel is directly injected from the injector 12 toward the recess 4a of the piston 4. As a result, a layer of the fuel-air mixture, which is separated from a surrounding air layer, is formed (stratified) in the vicinity of the ignition plug 7, so that stable stratified charge combustion is carried out. [0031] The intake port of each cylinder is connected to a surge tank 8 as an intake-air collective chamber, via a branch pipe for each cylinder, and an intake pipe 13 that provides an intake-air collective passage is connected to the upstream end of the surge tank 8. An air cleaner 9 is provided at the upstream end of the intake pipe 13. An air flow meter 5 for detecting the intake air amount or air mass flow and an electronically controlled throttle valve 10 are mounted in the intake pipe 13 such that the air flow meter 5 is located upstream of the throttle valve 10. The intake ports of the respective cylinders, surge tank 8 and the intake pipe 13 form an intake passage of the engine 1.

[0032] On the other hand, the exhaust port of each cylinder is connected to an exhaust pipe 6 as an exhaust collective passage, via a branch pipe for each cylinder, and a catalyst 11 in the form of a three-way catalyst having an O2 storage function is mounted in the exhaust pipe 6. The exhaust ports of the respective cylinders, branch pipes and the exhaust pipe 6 form an intake passage of the engine 1. Pre-catalyst sensor and post-catalyst sensor 17, 18 for detecting the exhaust air-fuel ratio are installed on the upstream side and downstream side of the catalyst 11, respectively. The pre-catalyst sensor 17, which is a so-called wide-range air-fuel ratio sensor, is able to continuously detect the air-fuel ratio over a relatively wide range, and produce an output in the form of a current signal proportional to the detected air-fuel ratio. The post-catalyst sensor 18, which is a so-called O2 sensor, is characterized in that its output voltage changes sharply when the air-fuel ratio exceeds or falls below the stoichiometric air-fuel ratio.

[0033] The above-mentioned ignition plug 7, throttle valve 10, injector 12 and other components are electrically connected to an electronic control unit (hereinafter referred to as "ECU") 20 that serves as a control device. The ECU 20

includes CPU, ROM, RAM, input and output ports, and a storage device, all of which are not illustrated. Also, the above-mentioned air flow meter 5, pre-catalyst sensor 17, post-catalyst sensor 18, a crank angle sensor 14 for detecting the crank angle of the engine 1, an acceleration stroke sensor 15 for detecting the accelerator pedal position (or the amount of depression of the accelerator pedal), and various other sensors are electrically connected to the ECU 20 via respective A/D converters (not shown), or the like, as shown in FIG. 1. The ECU 20 controls the ignition plug 7, throttle valve 10, injector 12 and other components so as to provide desired outputs, based on the values detected by the above-indicated various sensors, thereby to control the ignition timing, the fuel injection amount, the fuel injection timing and the throttle opening, for example. The throttle opening is normally controlled to a degree of opening responsive to the amount of depression of the accelerator pedal.

[0034] The catalyst 11 is adapted to treat NOx, HC and CO at the same time when the air-fuel ratio A/F (hereinafter abbreviated to AF) of exhaust gas flowing into the catalyst 11 is equal to the stoichiometric air-fuel ratio AFs (which is, for example, 14. 6). To enable the catalyst 11 to perform this function, the ECU 20 controls the air-fuel ratio during normal operation of the engine so that the exhaust air-fuel ratio detected upstream of the catalyst, which may be called "pre-catalyst air-fuel ratio AFfr", becomes equal to the stoichiometric ratio AFs. More specifically, the ECU 20 sets a target air-fuel ratio AFt to the stoichiometric ratio AFs, and controls the amount of fuel injected from the injector 12 so that the pre-catalyst air-fuel ratio AFfr detected by the pre-catalyst sensor 17 coincides with the target air-fuel ratio AFt. In this manner, the air-fuel ratio of the exhaust gas supplied to the catalyst 11 is kept at or near the stoichiometric air-fuel ratio, and the catalyst 11 exhibits the maximum exhaust purification performance.

[0035] In the following, the catalyst 11 will be described in more detail. As shown in FIG. 2, the catalyst 11 has a carrier substrate 33, a coating material 31 with which a surface of the carrier substrate 33 is coated, and catalyst

components 32 in the form of particles disposed on the coating material 31. In the catalyst 11, a large number of catalyst components 32 are scattered on the coating material 31, and are exposed to the interior of the catalyst 11. The catalyst components 32 are formed mainly of a noble metal, such as Pt or Pd, and provide

5 active sites for reaction with exhaust gas components, such as NOx, HC and CO. On the other hand, the coating material 31 serves as a promoter for promoting reactions at the interfaces between the exhaust gas and the catalyst components 32, and contains an oxygen storage component capable of absorbing and releasing oxygen depending on the air-fuel ratio of surrounding gas. The oxygen storage

0 component is, for example, cerium dioxide CeO2. If a gas surrounding the catalyst components 32 and the coating material 31 has a richer (i.e., smaller) air-fuel ratio than the stoichiometric air-fuel ratio, for example, oxygen that has been stored in the oxygen storage component present around the catalyst components 32 is released, whereby unburned components, such as HC and CO, are oxidized by the

5 released oxygen, to be thus converted into harmless substances. Conversely, if a gas surrounding the catalyst components 32 and the coating material 31 has a leaner (i.e., larger) air-fuel ratio than the stoichiometric ratio AFs, the oxygen storage component present around the catalyst components 32 absorbs oxygen from the surrounding gas, so that NOx is reduced and converted into a harmless

!0 substance.

[0036] Owing the above -de scribed function of absorbing and releasing oxygen, the catalyst 11 is able to treat three exhaust components, i.e., NOx, HC and CO, at the same time even if the pre-catalyst air-fuel ratio AF varies to some extent relative to the stoichiometric ratio AFs during normal air-fuel ratio control.

Iδ Thus, under normal air-fuel ratio control, it is possible to perform exhaust gas purification by minutely oscillating the pre-catalyst air-fuel ratio AFfr with respect to the stoichiometric ratio AFs as the center of oscillation, and thus causing the catalyst 11 to repeatedly absorbing and releasing oxygen.

[0037] When the catalyst 11 is in a condition of a new product, a large

number of catalyst components in the form of minute particles are uniformly scattered on the coating material 31, as described above, and the probability of contact between the exhaust gas and the catalyst components 32 is kept at a high level. As the catalyst 11 deteriorates, however, a part of the catalyst components

5 32 may disappear, and/or two or more catalyst components 32 may be burned by heat of the exhaust and stuck together into a sintered condition (as indicated by broken lines in FIG. 2). In such cases, the probability of contact between the exhaust gas and the catalyst components 32 may be reduced, which would cause a reduction of the catalytic conversion efficiency. In addition, the amount of the

.0 coating material 31, or the amount of the oxygen storage component, present around the catalyst components 32 may be reduced, resulting in a reduction of the oxygen storage capability itself.

[0038] As is understood from the above description, there is a correlation between the degree of deterioration of the catalyst 11 and the degree of reduction of

L5 the oxygen storage capability. In the present embodiment, therefore, the degree of deterioration of the catalyst 11 is detected on the basis of the oxygen storage capability of the catalyst 11. Here, the oxygen storage capability of the catalyst 11 is represented by the magnitude of the oxygen storage capacity (the unit of which is gram) that is the maximum amount of oxygen that can be stored in the catalyst 11.

20 [0039] Next, detection of deterioration of the catalyst according to the present embodiment will be explained. In the present embodiment, the ECU 20 performs active air-fuel ratio control when detecting deterioration of the catalyst 11. When the active air-fuel ratio control is executed, the exhaust air-fuel ratio, i.e., the air-fuel ratio of the exhaust gas flowing into the catalyst 11, is oscillated to 5 the lean side and rich side relative to a certain central air-fuel ratio in a certain determination period. The amplitude of the oscillation is made larger that of normal air-fuel ratio control, and is equal to, for example, 0.5 as expressed in terms of the air-fuel ratio. The above-mentioned determination period means a period of time required for making a single determination on deterioration of the catalyst.

[0040] The detection of deterioration of the catalyst 11 is carried out when at least the catalyst 11 and the pre-catalyst and post-catalyst sensors 17, 18 are within a certain activation temperature range, preferably, when the engine 1 is in a steadystate operating condition. In the present embodiment, the temperature of

5 the catalyst 11 is estimated based on engine operating conditions, using a certain map or function, though the catalyst temperature may be directly detected. The detection of deterioration of the catalyst 11 is carried out, for example, once for each operation of the engine or each trip of the vehicle. When it is determined at least twice in a row that the catalyst 11 is in a deteriorated condition, the

.0 deterioration is confirmed, and a warning device is activated.

[0041] In the meantime, the conventional catalyst deterioration detecting system has a problem that it is difficult for the system to use the utmost oxygen storage capability possessed by the catalyst. To solve this problem, an improvement as described below is made according to the present embodiment to L5 the method of oscillating the exhaust air-fuel ratio under the active air-fuel ratio control.

[0042] In the active air-fuel ratio control of the present embodiment, when the catalyst 11 is caused to absorb oxygen, the exhaust air-fuel ratio is oscillated such that the exhaust air-fuel ratio shifted to the rich side becomes equal to or 20 smaller than the stoichiometric air-fuel ratio, and such that the degree of leanness per cycle of oscillation is larger than the degree of richness. When the catalyst 11 is caused to release oxygen, the exhaust air-fuel ratio is oscillated such that the exhaust air-fuel ratio shifted to the lean side becomes equal to or larger than the stoichiometric ratio, and such that the degree of richness per cycle of oscillation is 5 larger than the degree of leanness. Thus, when the catalyst 11 is caused to absorb oxygen, lean gas is intermittently supplied to the catalyst 11, namely, supply of the lean gas is periodically stopped for a moment, while the lean/rich balance of the exhaust gas supplied to the catalyst is biased toward the lean side. When the catalyst 11 is caused to release oxygen, rich gas is intermittently supplied to the

catalyst 11, namely, supply of the rich gas is periodically stopped for a moment, while the lean/rich balance of the exhaust gas supplied to the catalyst is biased toward the rich side. In this manner, it is possible to cause the catalyst to absorb or release oxygen by using the utmost oxygen storage capability possessed by the catalyst.

[0043] Referring now to FIG. 3, the degree of leanness and degree of richness per cycle of oscillation may be represented by the areas of respective regions denoted by Kl and Kr, for example. Namely, the degree of leanness Kl may be expressed as a value obtained by integrating a difference DAFl (= AFl — AFs) between the exhaust air-fuel ratio (i.e., lean-side air-fuel ratio) AFl and the stoichiometric ratio AFs in a lean-side period Tl in which the exhaust air-fuel ratio AF is shifted to the lean side relative to the central air-fuel ratio AFc, over the lean-side period Tl. The degree of richness Kr may be expressed as a value obtained by integrating a difference DAFr (= AFs — AFr) between the exhaust air-fuel ratio (i.e., rich-side air-fuel ratio) AFr and the stoichiometric ratio AFs in a rich-side period Tr in which the exhaust air-fuel ratio AF is shifted to the rich side relative to the central air-fuel ratio AFc, over the rich-side period Tr.

[0044] The example shown in FIG. 3 illustrates the case where the catalyst 11 is caused to absorb oxygen, and the degree of leanness Kl is made larger than the degree of richness Kr. In one cycle of oscillation as shown in FIG. 3, the lean-side amplitude relative to the central air-fuel ratio AFc is expressed as δAF1 = AFl — AFc, and the rich-side amplitude relative to the central air-fuel ratio AFc is expressed as δAFr = AFc — AFr. Both of the amplitudes δAF1, δAFr take positive values. [0045] Next, some embodiments of the catalyst deterioration detecting system that performs the active air-fuel ratio control as described above will be described.

[0046] FIG. 4 shows changes in values associated with the catalyst deterioration detecting system according to a first embodiment of the invention.

In FIG. 4, the graph (A) shows changes in the air-fuel ratio (pre-catalyst air-fuel ratio) AFfr of the exhaust gas supplied to the catalyst 11, more specifically, the output value Vfr of the pre-catalyst sensor 17. Since the exhaust air-fuel ratio AFfr for the catalyst 11 is changed in accordance with changes in the target air-fuel

5 ratio AFt set by the ECU 20, the target air-fuel ratio AFt and the exhaust air-fuel ratio AFfr are substantially equivalent to each other, and it can be said that the graph (A) shows changes in the target air-fuel ratio AFt. The graph (B) shows changes in the exhaust air-fuel ratio detected downstream of the catalyst 11, more specifically, the output value Vrr of the post-catalyst sensor 18. In the graph (B),

0 the one-dot chain line corresponds to the case where the catalyst 11 is deteriorated by a small degree (the case of a small- deterioration catalyst), and the solid line corresponds to the case where the catalyst 11 is deteriorated by a large degree (the case of a large-deterioration catalyst), while the broken line corresponds to the case where the catalyst 11 is a new product. The graph (C) indicates changes in the

5 locus length Vrrsum of the output value Vrr of the post-catalyst sensor 18. In the graph (C), the one-dot chain line corresponds to the case of a small- deterioration catalyst, and the solid line corresponds to the case of a large -deterioration catalyst.

[0047] As shown in the graph (A) in FIG. 4, the active air-fuel ratio control is continuously performed in one determination period Tj, in which the target

!0 air-fuel ratio AFt is oscillated at short intervals, and the exhaust air-fuel ratio upstream of the catalyst is also oscillated at short intervals in accordance with the target air-fuel ratio AFt. Then, the exhaust air-fuel ratio downstream of the catalyst oscillates in different ways depending on the degree of deterioration of the catalyst, in response to the oscillation of the exhaust air-fuel ratio upstream of the

.5 catalyst (see (B) in FIG. 4). The ECU 20 adds up the amount of change of the output value Vrr of the post-catalyst sensor 18 for each infinitesimal time period during the determination period Tj, to thus calculate the locus length Vrrsum. The ECU 20 then compares the final locus length Vrrsum with a predetermined deterioration determination value. If the locus length Vrrsum is equal to or larger

than the predetermined deterioration determination value, as in the case of the large-deterioration catalyst shown in the graph (C), it is determined that the catalyst is deteriorated. If the locus length Vrrsum is smaller than the deterioration determination value, as in the case of the small- deterioration catalyst shown in the graph (C), on the other hand, it is determined that the catalyst is normal.

[0048] As the degree of deterioration of the catalyst increases, the exhaust air-fuel ratio downstream of the catalyst is more likely to fluctuate or vary in accordance with variations in the exhaust air-fuel ratio upstream of the catalyst, and the frequency at which the output value Vrr of the post-catalyst sensor 18 is reversed from the lean side to the rich side or from the rich side to the lean side is also increased, resulting in an increase in the locus length Vrrsum. Thus, deterioration of the catalyst can be detected by comparing the locus length Vrrsum with the deterioration determination value. [0049] In the active air-fuel ratio control of the first embodiment, the central air-fuel ratio AFc of the oscillation of the air-fuel ratio upstream of the catalyst is gradually changed from a certain lean value to a certain rich value (this change may be called "sweep"), while passing the stoichiometric air-fuel ratio in the middle of the change. In the example illustrated in FIG. 4, the frequency, period and amplitude of the oscillation of the air-fuel ratio are predetermined, fixed values. The lean-side amplitude and rich-side amplitude relative to the central air-fuel ratio AFc are also equal to each other. Although the waveform of the oscillation is slightly rounded and some variations in the amplitude are observed in the graph (B) that indicates the actual measurement values of the post-catalyst sensor 18, the waveform of the oscillation of the target air-fuel ratio AFt itself is rectangular, and its amplitude is constant. In this example, the central air-fuel ratio AFc is changed at a constant speed per unit time. The active air-fuel ratio control as heretofore explained is performed through feedforward control according to a predetermined program.

[0050] In an early period of the determination period Tj, the central air-fuel ratio AFc is shifted to a value (on the lean side) that is larger than the stoichiometric air-fuel ratio AFs, so that the degree of leanness per cycle of oscillation becomes larger than the degree of richness, and the lean/rich balance is biased toward the lean side. In this period as a whole, the catalyst is caused to absorb oxygen. Also, the rich-side air-fuel ratio is made equal to or smaller than the stoichiometric ratio AFs (see a portion of the graph (A) defined by a broken-line circle), and oxygen is released from the catalyst while the air-fuel ratio falls below the stoichiometric ratio AFs. Thus, absorption of oxj _^ gen into the catalyst takes place intermittently. In other words, the catalyst repeatedly undergoes alternate absorption of a relatively large amount of oxygen and release of a relatively small amount of oxygen.

[0051] In a middle period of the determination period Tj, as the central air-fuel ratio AFc approaches the stoichiometric ratio AFs, the degree of leanness and the degree of richness per cycle of oscillation come to be close to each other, thus establishing a better lean/rich balance. In this period as a whole, the catalyst almost evenly undergoes absorption of oxygen and release of oxygen.

[0052] In a late period of the determination period Tj, the central air-fuel ratio AFc is shifted to a value (on the rich side) that is smaller than the stoichiometric ratio AFs, so that the degree of richness per cycle of oscillation becomes larger than the degree of leanness, and the lean/rich balance is biased toward the rich side. In this period as a whole, oxygen is released from the catalyst. Also, the lean-side air-fuel ratio is made equal to or larger than the stoichiometric ratio AFs, and oxygen is absorbed into the catalyst while the air-fuel ratio exceeds the stoichiometric ratio AFs. Thus, release of oxygen from the catalyst takes place intermittently. In other words, the catalyst repeatedly undergoes alternate release of a relatively large amount of oxygen and absorption of a relatively small amount of oxygen.

[0053] In a preparation period Tp prior to the determination period Tj, the

exhaust air-fuel ratio is kept larger (on the lean side) than the stoichiometric air-fuel ratio AFs, so that the catalyst is brought into a condition where oxygen is absorbed in the catalyst to the full extent of the oxygen storage capability. As a result, lean gas that flows through the catalyst without having oxygen absorbed by the catalyst hits on the post-catalyst sensor 18, and the output value Vrr of the post-catalyst sensor 18 becomes a lean value.

[0054] With the active air-fuel ratio control as described above, the lean/rich balance of the exhaust air-fuel ratio shifts from the lean side to the rich side with the passage of time in the determination period Tj. The smaller the oxygen storage capability of the catalyst is, namely, the more deteriorated the catalyst is, the earlier the output value Vrr of the post-catalyst sensor 18 starts oscillating and switching between the lean side and the rich side. Also, the frequency at which the output value Vrr is reversed to the rich side starts increasing at an early stage. In the case of a new catalyst as indicated by the broken line in the graph (B) in FIG. 4, the output value Vrr of the post-catalyst sensor 18 is reversed to the rich side only once at the final stage of the determination period Tj.

[0055] In conventional active air-fuel ratio control using, for example, a Cmax method, the air-fuel ratio of exhaust gas supplied to the catalyst is set at a constant value, and lean or rich gas is continuously supplied to the catalyst. In this method, however, only the upstream portion and surface of the catalyst are used for absorption and release of oxygen, and breakthrough of the lean or rich gas (i.e., passage of the gas through the catalyst without involving any reactions) takes place at an early stage, whereby the oxygen storage capability is likely to be estimated to be smaller than a true value. In the case of the active air-fuel ratio control of the present embodiment, on the other hand, lean or rich gas is intermittently (i.e., not continuously) supplied to the catalyst, so that the catalyst undergoes slow changes of state. As a result, the downstream and deep portions of the catalyst can be also used for absorption and release of oxygen, and the true

oxygen storage capability can be measured with high accuracy. This also makes it possible to precisely determine the degree of deterioration of the catalyst that has been deteriorated by some degree, and to distinguish, with high accuracy, between a catalyst that is fully deteriorated and a catalyst that is about to deteriorate. [0056] Referring next to FIG. 5 through FIG. 7, the mechanism of oxygen absorption and release that take place in the catalyst will be explained. In each of FIG. 5 - FIG. 7, changes in the air-fuel ratio of exhaust gas supplied to the catalyst are illustrated in the upper left section of the figure, and a partial cross-section of the catalyst, in which a hatched area represents a portion used for oxygen absorption and release, is illustrated in the lower right section of the figure. As also shown in FIG. 2, the catalyst includes the carrier substrate 33 and the coating material 31. The hollow arrow in each of FIGS. 5 — 7 indicates the direction of flow of exhaust gas.

[0057] Referring initially to FIG. 5, if the exhaust air-fuel ratio is oscillated with large variations δF in the air-fuel ratio from the lean side to the rich side or vice versa, an upstream end portion (left end portion in FIG. 5) of the coating material 31 ranging from its surface to a deep position can be used for absorption and release of oxygen. However, where the lean-side period Tl and rich-side period Tr are long (namely, where the period of the oscillation is long), only a surface portion of the coating material 31 can be used for absorption and release of oxygen, and a deep portion of the coating material 31 which is located downstream of the upper end portion cannot be used for absorption and release of oxygen.

[0058] If, on the other hand, the lean-side period Tl and rich-side period Tr are shortened (namely, the period of the oscillation is shortened) with the same variations δF in the air-fuel ratio, as shown in FIG. 6, a deep portion of the coating material 31 that can be used for absorption and release of oxygen is extended further to the downstream side.

[0059] If the lean-side period Tl and rich-side period Tr are shortened with the same variations δF in the air-fuel ratio, and the central air-fuel ratio is

gradually changed (undergoes a sweep), as shown in FIG. 7, a deep portion of the coating material 31 that can be used for absorption and release of oxygen is extended to the downstream side even further than that of FIG. 6, as indicated by one-dot chain lines in FIG. 7. By employing the method of oscillating the air-fuel ratio as shown in FIG. 4, therefore, deterioration of the catalyst can be detected with high accuracy, using the utmost oxygen storage capability of the catalyst.

[0060] Referring next to FIG. 8 through FIG. 12, the results of tests concerning the first embodiment will be explained. FIG. 8 illustrates the pre-catalyst air-fuel ratio AFfr (in the upper graph) and the output value Vrr of the post-catalyst sensor (in the lower graph), as in the graphs (A) and (B) of FIG. 4. The pre-catalyst air-fuel ratio AFfr is obtained by converting the output value Vfr of the pre-catalyst sensor into the air-fuel ratio. In the active air-fuel ratio control performed in this test, the central air-fuel ratio AFc was changed from 15.1 to 14.1 at a constant speed per unit time. The lean-side amplitude δAF1 and rich-side amplitude δAFr were equally set to 0.5. The period of the oscillation was Is, namely, the frequency of the oscillation was IHz. With regard to catalysts to be tested, a large -deterioration catalyst B and a small- deterioration catalyst A, having a 3g difference of the oxygen storage capacity, were used. The test was conducted under a condition that the flow rate (air mass flow) GA of intake air drawn into the engine (which will also be called "intake air amount") was 20g/s.

[0061] FIG. 9 shows the results of comparison in terms of the locus length Vrrsum between the case of the small-deterioration catalyst A and the case of the large-deterioration catalyst B. As shown in FIG. 9, an apparently large difference in the locus length Vrrsum was observed between the catalyst A and the catalyst B. It was thus confirmed that the catalyst deterioration detecting system of the first embodiment is advantageous in its ability to recognize a slight difference in the degree of deterioration of catalyst.

[0062] FIG. 10 shows the results in the case where the lean-side amplitude δAF1 and rich-side amplitude δAFr were changed from 0.5 to 0.3 and 0.2. It was

confirmed that the difference in the locus length Vrrsum between the small-deterioration catalyst A and the large -deterioration catalyst B becomes larger as the lean-side amplitude δAF1 and rich-side amplitude δAr become larger, as seen in FIG. 10. [0063] FIG. 11 shows the results in the case where the intake air amount

GA was changed from 20 g/s to 10 g/s and 5 g/s. It was confirmed that the difference in the locus length Vrrsum becomes larger as the intake air amount GA increases, as seen in FIG. 11.

[0064] FIG. 12 shows the results in the case where the frequency of the oscillation of the air-fuel ratio was changed from 1 Hz to 0.75 Hz and 0.5 Hz. It was confirmed that the difference in the locus length Vrrsum becomes larger as the oscillation frequency becomes higher, as seen in FIG. 12.

[0065] Next, a first modified example of the first embodiment will be described. If fuel cut control for stopping supply of fuel to the engine is performed during execution of the active air-fuel ratio control, the active air-fuel ratio control is interrupted since the exhaust air-fuel ratio becomes extremely lean due to the fuel cut control. Depending on the use or application of the engine, the fuel cut control may be performed at a relatively high frequency during engine operation.

In this case, the active air-fuel ratio control is interrupted at an increased frequency, resulting in a reduced frequency of catalyst deterioration detection.

[0066] In view of the above problem, the ECU 20 performs control as described below, so as to reduce the frequency of interruption of the active air-fuel ratio control due to the fuel cut control, and assure an increased frequency of catalyst deterioration detection. In the example as described below, the engine 1 is installed on a hybrid vehicle, namely, the hybrid vehicle is provided with the engine 1 and a motor- generator as power sources.

[0067] FIG. 13 shows changes in values or parameters during execution of control for deterioration detection in the above case. In FIG. 13, (A) indicates the vehicle speed, (B) indicates the total torque as a total value of the engine torque

and the motor torque, (C) indicates the engine torque, (D) indicates the motor torque, (E) indicates a fuel cut (FC) request signal, (F) indicates the prexatalyst sensor output VTr, (G) indicates the post-catalyst sensor output Vrr, and (H) indicates the locus length Vrrsum.

5 [0068] The ECU 20 produces a fuel cut request signal as an internal signal in the form of a flag, or the like, namely, sets the fuel cut request signal to ON, when certain fuel cut conditions are satisfied. The fuel cut conditions may be, for example, two conditions that (l) the accelerator pedal position AC detected by the acceleration stroke sensor 15 indicates that the accelerator pedal is substantially

0 fully released, and that (2) the engine speed NE calculated from the output of the crank angle sensor 14 is equal to or higher than a predetermined speed NEl that is a little higher than the idle speed. If the fuel cut request signal is produced, the ECU 20 normally stops applying current to the injectors 12, so as to execute fuel cut control.

.5 [0069] During execution of the active air-fuel ratio control, however, the fuel cut control is not carried out even if the fuel cut request signal is set to ON, as shown in FIG. 13. The engine continues to be operated, and the active air-fuel ratio control and catalyst deterioration detection continue to be performed. In the meantime, the motor- generator is switched to the generator mode in order to

-0 achieve a deceleration desired by the driver. Upon switching, the motor- generator generates electric power with which a battery is charged, so that the motor- generator produces negative torque due to the resistance that appears during power generation. In this manner, the total torque can be reduced, and the desired deceleration can be achieved. In FIG. 13, the broken line in the graph (G)

25 indicates changes in the post-catalyst sensor output Vrr in the case where the fuel cut control is carried out when the fuel cut request signal is set to ON.

[0070] Thus, the active air-fuel ratio control and catalyst deterioration detection are not interrupted even if a fuel cut request signal is produced during execution of the active air-fuel control, and therefore, the frequency of the catalyst

deterioration detection can be increased.

[0071] It is to be understood that the control of the motor- generator at the time of production of the fuel cut request signal according to the first modified example may be applied to all of the examples that will be described later. [0072] Next, a second modified example of the first embodiment will be described. In the active air-fuel ratio control as described above, the period and frequency of the air-fuel ratio oscillation are constant, and the central air-fuel ratio AFc is changed at a constant speed or rate per unit time. In the second modified example, on the other hand, the period and frequency of the air-fuel ratio oscillation and the rate of change of the central air-fuel ratio AFc are changed depending on the intake air amount GA.

[0073] In FIG. 14, the graph (A) shows changes in the intake air amount GA, and the graphs (B), (C) and (D) illustrate difference modes or patterns of the air-fuel ratio oscillation under the active air-fuel ratio control. In the active air-fuel ratio control as shown in each of the graphs (B), (C) and (D), the central air-fuel ratio AFc is changed from the rich side to the lean side, contrary to that of the first embodiment as described above. Thus, the direction of change of the central air-fuel ratio AFc may be either from the rich side to the lean side or from the lean side to the rich side. In the graphs (B), (C) and (D), changes in the target air-fuel ratio AFt are illustrated.

[0074] The graph (D) illustrates a pattern of air-fuel ratio oscillation (a first case) similar to that of the graph (A) in FIG. 4, in which the period and frequency of the air-fuel ratio oscillation and the rate of change of the central air-fuel ratio AFc are constant irrespective of the intake air amount GA. [0075] On the other hand, the graph (C) illustrates a pattern of air-fuel ratio oscillation (a second case) in which the period and frequency of the air-fuel ratio oscillation are changed depending on the intake air amount GA. The period of the air-fuel ratio oscillation is shortened as the intake air amount GA increases, and is extended or increased as the intake air amount GA decreases. Accordingly,

the frequency of the air-fuel ratio oscillation becomes higher as the intake air amount GA increases, and becomes lower as the intake air amount GA decreases.

[0076] The graph (B) illustrates a pattern of air-fuel ratio oscillation (a third case) in which the period and frequency of the airfuel ratio oscillation are changed depending on the intake air amount GA, as in the second case of (C), and the rate of change of the central air-fuel ratio AFc of the air-fuel ratio oscillation is also changed depending on the intake air amount GA. The rate of change of the central air-fuel ratio AFc is increased as the intake air amount GA increases, and is reduced as the intake air amount GA decreases. [0077] The period and frequency of the oscillation and the rate of change of the central air-fuel ratio AFc are changed depending on the intake air amount in the manners as described above, for the reasons as follows.

[0078] When the air-fuel ratio is oscillated under the active air-fuel ratio control, it is desirable that the amount of exhaust gas (lean gas) supplied to the catalyst in the lean-side period Tl in which the air-fuel ratio is controlled to the lean side relative to the central air-fuel ratio AFc is equal to that of exhaust gas (rich gas) supplied in the rich-side period Tr in which the air-fuel ratio is controlled to the rich side. In the meantime, the intake air amount GA varies depending on the amount of depression of the accelerator pedal by the driver, and the amount of exhaust gas also changes depending on the intake air amount GA. Thus, in the method in which the period and frequency of the air-fuel ratio oscillation are constant, the amount of exhaust gas supplied to the catalyst in the lean-side period Tl is not always equal to that of exhaust gas supplied in the rich-side period Tr.

[0079] In the method in which the period and frequency are constant, as the intake air amount GA decreases, the amounts of lean gas and rich gas supplied to the catalyst are reduced, and the amounts of lean gas and rich gas discharged to the downstream of the catalyst are also reduced. As a result, the variations in the output value Vrr of the post-catalyst sensor and the frequency of reversal (i.e., switching from lean to rich or vice versa) of the output value Vrr are reduced, and

the locus length. Vrrsum calculated from the output value Vrr becomes somewhat smaller than a value obtained when the intake air amount GA is larger, resulting in a possibility of an erroneous determination that the catalyst is normal. Namely, even with regard to the same catalyst, the result of calculation and the result of 5 detection may vary depending on the intake air amount GA, resulting in reduced detection accuracy.

[0080] In the second and third cases as shown in the graphs (C) and (B) in

FIG. 14, therefore, the total air amount TGA as a total value of the intake air amount in each of the lean-side period Tl and the rich-side period Tr is calculated,

0 as shown in FIG. 15. More specifically, the ECU 20 successively adds up the value of the intake air amount GA detected by the air flow meter 5 at minute time intervals δT, from the time of reversal (tl) of the target air-fuel ratio AFt to the lean side. At a point in time (t2) when the value of the total air amount TGA reaches a predetermined threshold value TGAs, the target air-fuel ratio AFt is

5 reversed to the rich side.

[008l] According to this method, the total intake air amount in the lean-side period Tl and the total intake air amount in the rich-side period Tr can be made equal to each other, and therefore, the amount of exhaust gas supplied to the catalyst in the lean-side period Tl and the amount of exhaust gas supplied in the

!0 rich-side period Tr can be made equal to each other.

[0082] With the above control executed, the period of the air-fuel ratio oscillation is shortened as the intake air amount GA increases, and is extended as the intake air amount GA decreases, as described above. Since the period of the air-fuel ratio oscillation is extended as the intake air amount GA decreases, in .5 particular, the amounts of lean gas and rich gas supplied to the catalyst when the intake air amount GA is small can be made larger than those of the case where the period is constant. As a result, the amounts of lean gas and rich gas discharged to the downstream of the catalyst are increased, and the variations in the post-catalyst sensor output value Vrr and the frequency of reversal thereof are

increased, so that the locus length Vrrsum is increased, and the ECU 20 is prevented from making an erroneous determination that the catalyst is normal. It is also possible to reduce variations in the locus length Vrrsum due to variations in the air amount, and thus improve the detection accuracy. [0083] In the meantime, it was found that the following problem occurs even where the period of the air-fuel ratio oscillation is changed depending on the intake air amount GA in the manner as described above. That is, when the intake air amount GA is reduced, the period of the air-fuel ratio oscillation is extended or increased, and the number of cycles of air-fuel ratio oscillation per unit time is reduced. If the period of the air-fuel ratio oscillation is extended when the central air-fuel ratio AFc is changed at a constant rate per unit time, an amount of shift (or a skip amount) of the central air-fuel ratio AFc as a difference between the central air-fuel ratio AFc at the beginning of a certain cycle of oscillation and the central air-fuel ratio AFc at the beginning of the next cycle of oscillation is increased. In this connection, it was found that, if the central air-fuel ratio AFc becomes close to the stoichiometric air-fuel ratio AFs while the intake air amount GA is small, the variations and the frequency of reversal in the post-catalyst sensor output value Vrr are reduced due to the increased skip amount of the central air-fuel ratio AFc. As a result, the locus length Vrrsum of the post-catalyst sensor output value Vrr calculated over this period is reduced as compared with the case where the intake air amount GA is large, and the final locus length Vrrsum calculated at the end of the determination period becomes more or less smaller than a value that should be obtained, resulting in a possibility of an erroneous determination that the catalyst is normal. [00840 In order to solve the above problem, in the third case as shown in the graph (B) of FIG. 14, the central air-fuel ratio AFc is changed by a predetermined skip amount δAFc each time the target air-fuel ratio AFt (i.e., the exhaust air-fuel ratio AFfr) undergoes one cycle of oscillation (that involves two reversals). Namely, the central air-fuel ratio AFc is changed at a fixed rate per

unit number of reversals, rather than per unit time. With this control executed, the skip amount of the central air-fuel ratio AFc during its sweep can be kept constant irrespective of the length in time of the oscillation period. If this control is performed along with the control of the second case, as described above, the oscillation period is shortened and the rate of change of the central air-fuel ratio AFc per unit time is increased as the intake air amount GA increases, and the oscillation period is extended and the rate of change of the central air-fuel ratio AFc per unit time is reduced as the intake air amount GA decreases.

[0085] In the third case as described above, when the central air-fuel ratio AFc becomes close to the stoichiometric air-fuel ratio while the intake air amount GA is small, the rate of change of the central air-fuel ratio AFc per unit time can be reduced as compared with the case where the central air-fuel ratio AFc is changed at a constant rate per unit time. As a result, the number of cycles of air-fuel ratio oscillation per unit time in this period is increased, and the variations and the frequency of reversal in the post-catalyst sensor output value Vrr are increased, so that the locus length Vrrsum is increased, and the ECU 20 is prevented from making an erroneous determination that the catalyst is normal. It is also possible to reduce variations in the locus length Vrrsum due to variations in the air amount, and thus improve the detection accuracy. [0086] As another method of changing the central air-fuel ratio AFc, the central air-fuel ratio AFc may be changed by a predetermined step amount δAFc each time the target air-fuel ratio AFt (i.e., the exhaust air-fuel ratio AFfr) switches from the rich side to the lean side or vice versa relative to the central air-fuel ratio AFc, namely, for each reversal in the oscillation. [0087] Next, a catalyst deterioration detecting process according to the third case of the second modified example of the first embodiment will be described with reference to FIGS. 16A, 16B and FIG. 17. FIGS. 16A and 16B are flowcharts of a main routine for executing the deterioration detecting process, and FIG. 17 is a flowchart of a subroutine executed in step S114 of the main routine. The ECU 20

repeatedly executes the routines of FIGS. 16A, 16B and FIG. 17 at certain minute time intervals (e.g., 16 msec).

[0088] As shown in FIGS. 16A and 16B, it is initially determined in step SlOl whether a condition for executing deterioration detection is met. The

5 executing condition is met when all of the conditions suitable for executing deterioration detection, including, for example, (l) the coolant temperature is equal to or higher than a predetermined value, (2) the pre-catalyst sensor 17 and the post-catalyst sensor 18 are in active states, and (3) the engine is not at idle, are satisfied.

.0 [0089] When it is determined in step SlOl that the deterioration detection executing condition is not met, the ECU 20 proceeds to step S119 to clear the values of the total number of reversals and total air amount as described later to zero, and set the target air-fuel ratio AFt to a normal value, so that normal air-fuel ratio control is carried out. Then, the present routine is finished.

L5 [0090] If it is determined in step SlOl that the deterioration detection executing condition is met, on the other hand, the ECU 20 proceeds to step S102 to determine whether an initial condition process completion flag (which will be described later) is ON. The ECU 20 proceeds to step S114 when this flag is not ON, and proceeds to step S103 when the same flag is ON.

20 [0091] In step S 114, a preparation process for deterioration detection as shown in FIG. 17 is performed. The preparation process will be hereinafter described. As shown in FIG. 17, it is initially determined in step S201whether a target air-fuel ratio set flag is ON. If this flag is not ON, the ECU 20 proceeds to step S202 to determine whether the current post-catalyst sensor output value Vrr 5 is equal to or larger than a lean/rich determination threshold value Vrrs (see FIG. 4).

[0092] If the post-catalyst sensor output value Vrr is equal to or larger than the lean/rich determination threshold value Vrrs, it is determined that the current post-catalyst air-fuel ratio AFrr is richer than the stoichiometic air-fuel

ratio AFs, and the target air-fuel ratio AFt is set to a certain value AFtI that is leaner than the stoichiometric ratio in step S203. If the post:catalyst sensor output value Vrr is smaller than the lean/rich determination threshold value Vrrs, on the other hand, it is determined that the current post-catalyst air-fuel ratio 5 AFrr is leaner than the stoichiometric ratio AFs, and the target air-fuel ratio AFt is set to a certain value AFtr that is richer than the stoichiometric ratio in step S204. Thus, the target air-fuel ratio AFt is set to the side (lean or rich side) opposite to that of the post-catalyst sensor output value Vrr.

[0093] After execution of steps S203 and S204, the ECU 20 proceeds to 0 step S205 to set the target air-fuel ratio set flag to ON, and the present routine is finished.

[0094] If it is determined in step S201 that the target air-fuel ratio set flag is ON, it is determined in step S206 whether there is a history of reversal of the post-catalyst sensor output Vrr, namely, whether the post-catalyst sensor output 5 Vrr has passed the lean/rich determination threshold value Vrrs. If there is no history of reversal of the post-catalyst sensor output, it is determined in step S207 whether the post-catalyst sensor output has been reversed, namely, has switched from the lean side to the rich side or vice versa relative to the threshold value Vrrs. If the post-catalyst sensor output has been reversed, it is determined in step S208 !0 that there is a history of reversal, and the present routine is finished. If the post-catalyst sensor output has not been reversed, the present routine is finished without executing step S208.

[0095] If it is determined in step S206 that there is a history of reversal, on the other hand, the ECU 20 proceeds to step S209 to perform summation of the .5 intake air amount GA in the manner as described above (see FIG. 15). It is then determined in step S210 whether the total air amount TGA as a result of the summation is equal to or larger than a predetermined threshold value TGAs. If the total air amount TGA is smaller than the threshold value TGAs, the present routine is finished. If the total air amount TGA is equal to or larger than the

threshold value TGAs, the above-mentioned initial condition process completion flag is set to ON in step S211, and the total air amount TGA is reset to zero in step S212. Then, the present routine is finished.

[0096] According to the preparation process as shown in FIG. 17, a negative decision (NO) is obtained in step S210 when the routine of FIG. 17 is executed for the first time, and it is determined in step S202 whether the current post-catalyst sensor output is rich or lean. Then, the target air-fuel ratio is set to the side (lean or rich) opposite to that of the post-catalyst sensor output. Namely, the target air-fuel ratio is set to a lean value if the sensor output is rich, and is set to a rich value if the sensor output is lean.

[0097] If the target air-fuel ratio is set to a lean value, for example, a lean gas whose air-fuel ratio is leaner (or larger) than the stoichiometric ratio is supplied to the catalyst in the second and subsequent cycles of the routine of FIG. 17, so that the catalyst absorbs oxygen from the lean gas. Once the amount of oxygen stored in the catalyst reaches the full storage capacity of the catalyst, the lean gas starts leaking into the downstream of the catalyst, and the post-catalyst sensor output is reversed from rich to lean (YES in step S207). While oxygen has already been stored in the catalyst up to the full storage capacity at this time, further lean gas continues to be supplied to the catalyst (step S209) so as to make sure that the catalyst is in the full oxygen storage condition. When the total air amount TGA as a sum of the intake air amounts from the time of reversal of the post-catalyst sensor output reaches the threshold value TGAs (YES in step S210), the preparation process for the subsequent active air-fuel ratio control is finished.

[0098] Returning to FIGS. 16A and 16B, if it is determined in step S102 that the initial condition process completion flag is ON, the locus length Vrrsum of the post-catalyst sensor output is calculated in step S103. More specifically, an amount of change δVrr of the post-catalyst sensor output value from the last cycle to the current cycle of the routine is calculated by subtracting the post-catalyst sensor output value Vrrold of the last cycle from the post-catalyst sensor output

value Vrr of the current cycle, and the locus length Vrrsum of the current cycle is calculated by adding the amount of change δVrr of the post-catalyst sensor output value to the locus length Vrrsumold calculated in the last cycle.

[0099] It is then determined in step S104 whether the locus length Vrrsum calculated in step S103 is equal to or larger than a predetermined deterioration determination value. If the locus length Vrrsum is equal to or larger than the deterioration determination value, the ECU 20 proceeds to step S117 to determine that the catalyst is deteriorated, and a catalyst deterioration detection completion flag (i.e., a flag indicative of completion of catalyst deterioration detection) is set to ON, and step S119 as explained above is executed. Then, the present routine is finished.

[OIOO] If the locus length Vrrsum is smaller than the deterioration determination value, on the other hand, the target air-fuel ratio AFt for the active air-fuel ratio control is set in steps S105 through S113. Of these steps S105 - S113, steps S105 to S107 are concerned with setting of the central air-fuel ratio AFc, and steps S108 to S112 are concerned with setting of the amplitude of the oscillation.

[0101] Initially, it is determined in step S 105 whether the central air-fuel ratio AFc has reached a predetermined final target value.

[0102] If the central air-fuel ratio AFc has not reached the final target value, the ECU 20 proceeds to step S106 to determine whether the total number of reversals is equal to or larger than a predetermined value. The total number of reversals means the total number of times that the target air-fuel ratio AFt is reversed or switched from one side (lean or rich) to the other side relative to the central air-fuel ratio AFc. In this example, the central air-fuel ratio AFc is changed stepwise when the target air-fuel ratio AFt is reversed twice, namely, undergoes one cycle of oscillation, and therefore, the above-indicated predetermined value is set to 2. It is to be understood that, if the predetermined value is set to 1, the central air-fuel ration AFc can be changed for each reversal, and that, if the predetermined value is set to 4, 6, ..., the central air-fuel ratio AFc

can be changed every two cycles, thee cycles, ... of the oscillation.

[0103] If the total number of reversals is equal to or larger than the predetermined value, the central air-fuel ratio AFc is changed in step S107. Namely, a new central air-fuel ratio AFc is calculated by adding a certain skip amount δAFc to the central air-fuel ratio AFc used in the last cycle of the routine. The direction of change of the central air-fuel ratio AFc is determined in advance, and the skip amount δAFc takes a negative value if the central air-fuel ratio AFc is changed toward the rich side, and the skip amount δAFc takes a positive value if the central air-fuel ratio AFc is changed toward the lean side. [0104] If the total number of reversals has not reached the predetermined value, on the other hand, step S 107 is skipped. Before the central air-fuel ratio AFc is changed for the first time, the central air-fuel ratio AFc is set to a predetermined initial value.

[0105] Subsequently, it is determined in step S 108 whether the total air amount TGA as explained above referring to FIG. 15 becomes equal to or larger than a predetermined threshold value TGAs. If it is determined that the total air amount TGA is equal to or larger than the threshold value TGAs, the ECU 20 proceeds to step S109 to determine whether the current amplitude δAF is a rich-side amplitude δAFr. If the amplitude δAF is a rich-side amplitude δAFr, a lean-side amplitude δAF1 on the opposite side is set as a value of the amplitude

δAF in step SIlO. If the amplitude δAF is not a rich-side amplitude δAFr (i.e., if the amplitude δAF is a lean-side amplitude δAF1), on the other hand, a rich-side amplitude δAFr on the opposite side is set as the amplitude δAF in step Sill. In this manner, the target air-fuel ratio AFt is reversed. The magnitudes of the lean-side amplitude δAF1 and rich-side amplitude δAFr are determined in advance, and are equal to each other in this example.

[0106] In the following step S112, the total air amount TGA is cleared to zero, and the number of reversals is incremented. The ECU 20 then proceeds to step S113.

[0107] If it is determined in step S 108 that the total air amount TGA is smaller than the threshold value TGAs, on the other hand, the amplitude δAF is not set but kept constant, and the ECU 20 directly proceeds to step S113.

[0108] In step S113, the target air-fuel ratio AFt is calculated by adding or subtracting the amplitude δAF set in step SlIO or Sill to or from the central air-fuel ratio AFc calculated in step S107, and is set to the calculated value. When the lean-side amplitude δAF1 is set, the amplitude δAF1 is added to the central air-fuel ratio AFc. When the rich-side amplitude δAFr is set, the amplitude δAFr is subtracted from the central air-fuel ratio AFc. After execution of step S113, the present routine is finished.

[0109] If it is determined in step S105 that the central air-fuel ratio AFc has reached the final target value, the ECU 20 proceeds to step S115 to determine whether the locus length Vrrsum is equal to or larger than a predetermined deterioration determination value. If the locus length Vrrsum is equal to or larger than the deterioration determination value, it is determined in step S 117 that the catalyst is deteriorated. If the locus length Vrrsum is smaller than the deterioration determination value, it is determined in step S 116 that the catalyst is normal (or is not deteriorated). Then, steps S 118 and S 119 are executed, and the present routine is finished. [OHO] According to the deterioration detecting process as shown in FIGS.

16A and 16B, once the preparation process of step S114 is finished, an affirmative decision (YES) is made in step S 102, and the active air-fuel ratio control, which involves oscillation of the target air-fuel ratio AFt (the air-fuel ratio of exhaust gas supplied to the catalyst) and change of the central air-fuel ratio AFc, is started and performed in steps S105 through S113. For example, the central air-fuel ratio AFc is initially set to an initial lean value that is leaner (i.e., larger) than the stoichiometric air-fuel ratio AFs, and the amplitude δAF is set to a lean-side amplitude δAF1 as an initial value. In this condition, if the total air amount TGA as a result of summation of the intake air amount GA reaches the threshold value

TGAs (YES in step S 108), the amplitude δAF is set to the opposite rich- side amplitude δAFr (step Sill), and the number of reversals is incremented (step S112), while the target air-fuel ratio AFt is set to a rich-side air-fuel ratio AFr that is richer than the central air-fuel ratio AFc (step S113). Subsequently, when the target air-fuel ratio AFt is reversed to a lean-side air-fuel ratio AFl that is leaner than the central air-fuel ratio AFc through a similar process, the total number of reversals becomes equal to 2 (step S112), and one cycle of oscillation is finished. Then, an affirmative decision (YES) is made in step S106, and the central air-fuel ratio AFc is changed toward the rich side by the skip amount δAFc (step S 107). [OHl] After repeated execution of the above -de scribed process, the central air-fuel ratio AFc eventually reaches the final target value (YES in step S 105). The active air-fuel ratio control is finished at this point in time, and the locus length Vrrsum is compared with the deterioration determination value (step S 115) and it is determined that the catalyst is normal (S116) or that the catalyst is deteriorated (S 117).

[0112] Other modified examples of the first embodiment will be now described. In active air-fuel ratio control as shown in FIG. 18, the central air-fuel ratio AFc of the air-fuel ratio oscillation is set to and kept at a constant value that is displaced from the stoichiometric ratio AFs to the lean side. In the air-fuel ratio oscillation, the lean-side and rich-side periods Tl, Tr are set to an equal, constant value, and the lean-side and rich-side amplitudes δAF1, δAFr are set to an equal, constant value. The rich-side air-fuel ratio AFr reached by the target air-fuel ratio when it is shifted to the rich side relative to the central air-fuel ratio AFc is equal to the stoichiometric air-fuel ratio AFs. It is also possible to perform active air-fuel ratio control with the central air-fuel ratio AFc displaced to the rich side, using a pattern of oscillation obtained by flipping the pattern as shown in FIG. 18 with respect to the line indicative of the stoichiometric ratio AFs.

[0113] With the active air-fuel ratio control as described above, too, lean gas can be intermittently supplied to the catalyst while the lean/rich balance is

biased toward the lean side, so as to enable the catalyst to absorb oxygen using the utmost oxygen storage capability possessed by the catalyst.

[0114] Another example of active air-fuel ratio control is illustrated in FIG. 19, in which the central air-fuel ratio AFc is changed stepwise every three cycles of oscillation. This active air-fuel ratio control provides substantially the same effect or advantage as that as described above.

[0115] Next, a catalyst deterioration detecting system according to a second embodiment of the invention will be described. FIG. 20 shows changes in values associated with the catalyst deterioration detecting system of the second embodiment. In FIG. 20, the graph (A) shows the ON/OFF state of a deterioration detection execution flag, and the graph (B) shows changes in the target air-fuel ratio AFt. The graph (C) shows changes in the exhaust air-fuel ratio detected downstream of the catalyst, more specifically, the output value Vrr of the post-catalyst sensor 18, and the graph (D) shows changes in the amount of oxygen absorbed by the catalyst, while the graph (E) shows changes in the amount of oxygen released from the catalyst.

[0116] The catalyst deterioration detection according to the second embodiment generally relates to an improvement of the Cmax method as described above. While the active air-fuel ratio control is continuously carried out in one determination period Tj (see the graph (B) in FIG. 20), the target air-fuel ratio AFt is oscillated under the active air-fuel ratio control of the second embodiment, whereas the target air-fuel ratio AFt is held constant under the conventional active air-fuel ratio control. In response to the oscillation of the target air-fuel ratio AFt, the exhaust air-fuel ratio AFfr upstream of the catalyst is also oscillated. [0117] In the example as shown in FIG. 20, after the first preparation period TpI, the first active air-fuel ratio control is performed in the first determination period TjI in which the catalyst is caused to absorb catalyst. Subsequently, after the passage of the second preparation period Tp2, the second active air-fuel ratio control is performed in the second determination period Tj2 in

which oxygen is released from the catalyst.

[0118] In the first active air-fuel ratio control, the target air-fuel ratio AFt is oscillated with the degree of leanness made larger than the degree of richness, and a relatively large amount of lean gas and a relatively small amount of rich gas are alternately and repeatedly supplied to the catalyst, so that the amount of oxygen absorbed by the catalyst (which will also be called "absorbed OSA") is eventually increased at the end of the determination period TjI while repeatedly increasing and decreasing during this period. When the catalyst absorbs oxygen up to the full oxygen storage capacity of its own, the lean gas flows through the catalyst (without being treated) to the downstream of the catalyst, and the output value Vrr of the post-catalyst sensor 18 is reversed to the lean side. When the output value Vrr becomes equal to a lean/rich determination value Vrrs, the first determination period TjI and the first active air-fuel ratio control are finished.

[0119] Subsequently, the second active air-fuel ratio control is performed after the passage of the second preparation period Tp2. In the second active air-fuel ratio control, the target air-fuel ratio AFt is oscillated with the degree of richness made larger than the degree of leanness, contrary to the first active air-fuel ratio control, and a relatively large amount of rich gas and a relatively small amount of lean gas are alternately and repeatedly supplied to the catalyst, so that the amount of oxygen released from the catalyst (which will also be called

"released OSA") is eventually increased at the end of the determination period Tj2 while repeatedly increasing and decreasing during this period. If the oxygen stored in the catalyst is fully released from the catalyst, the rich gas flows through the catalyst (without being treated) to the downstream of the catalyst, and the output value Vrr of the post-catalyst sensor 18 is reversed to the rich side. When the output value Vrr becomes equal to the lean/rich determination value Vrrs, the second determination period Tj2 and the second active air-fuel ratio control are finished.

[0120] Then, an average oxygen amount (which will also be called "average

OSA") as an average value of the final absorbed oxygen amount calculated in the first active air-fuel ratio control and the final released oxygen amount calculated in the second active air-fuel control is calculated, and the value of the average oxygen amount is compared with a predetermined deterioration determination threshold value. If the average oxygen amount is larger than the deterioration determination threshold value, it is determined that the catalyst is normal. If the average oxygen amount is equal to or smaller than the deterioration determination threshold value, it is determined that the catalyst is deteriorated. While deterioration of the catalyst may be determined by comparing only one of the absorbed oxygen amount and the released oxygen amount with the deterioration determination threshold value, the deterioration determination is preferably made by comparing the average value of these amounts with the threshold value, so as to eliminate an influence of a deviation of the pre-catalyst air-fuel ratio from the stoichiometric ratio, and thus further improve the reliability. Needless to say, the amount of data used for deterioration detection may be increased so as to further enhance the reliability.

[0121] The absorbed oxygen amount is calculated in the following manner. Initially, an absorbed oxygen amount δOSA is calculated at predetermined minute time intervals δt according to the following equation (l). [0122] AOSA = AAF x Q x K = (AFfr - AFs) x Qx K (l)

[0123] In the above equation (l), Q is the fuel injection amount, and an excessive air amount can be calculated by multiplying a difference δAF in the air-fuel ratio by the fuel injection amount Q. K is a proportion (about 0.23) of oxygen contained in air. When the exhaust air-fuel ratio AFfr upstream of the catalyst is larger than the stoichiometric air-fuel ratio AFs, the air-fuel ratio difference δAF (= AFfr — AFs) takes a positive value, and the absorbed oxygen amount δOSA also takes a positive value. When the exhaust air-fuel ratio AFfr upstream of the catalyst is smaller than the stoichiometric ratio AFs, on the other hand, the air-fuel ratio difference δAF takes a negative value, and the absorbed

oxygen amount δOSA also takes a negative value. The absorbed oxygen amount

δOSA for each minute time interval is successively added up from the start of the active air-fuel ratio control to the end thereof (namely, over the first determination period Tjl), so that the absorbed oxygen amount OSA as a final value is calculated. [0124] On the other hand, the released oxygen amount is calculated in the following manner. Initially, a released oxygen amount δOSA is calculated at predetermined minute time intervals according to the following equation (l)'.

[0125] A OSA = δAF x Q x K = (AFs - AFfr) x Qx K (l)'

[0126] When the exhaust air-fuel ratio AFfr upstream of the catalyst is smaller than the stoichiometric ratio AFs, the air-fuel ratio difference δAF (= AFs —

AFfr) takes a positive value, and the released oxygen amount δOSA also takes a positive value. When the exhaust air-fuel ratio AFfr upstream of the catalyst is larger than the stoichiometric ratio AFs, on the other hand, the air-fuel ratio difference δAF takes a negative value, and the released oxygen amount δOSA also takes a negative value. The released oxygen amount δOSA for each minute time interval is successively added up from the start of the active air-fuel ratio control to the end thereof (namely, over the second determination period Tj2), so that the released oxygen amount OSA as a final value is calculated.

[0127] In the first preparation period TpI, a preparation process is performed in which oxygen is fully released from the catalyst in preparation for subsequent oxygen absorption. More specifically, as shown in FIG. 20, the target air-fuel ratio AFt (exhaust air-fuel ratio AFfr) is set to a constant value that is richer than the stoichiometric ratio AFs, and oxygen is fully released from the catalyst within at least the preparation period TpI, so that rich gas is caused to flow through the catalyst to the downstream of the catalyst, and the output value Vrr of the post-catalyst sensor 18 eventually becomes a rich-side value. Similarly, in the second preparation period Tp2, a preparation process is performed in which the catalyst is caused to fully absorb oxygen in preparation for subsequent oxygen release. More specifically, the target air-fuel ratio AFt (exhaust air-fuel ratio

AFfr) is set to a constant value that is leaner than the stoichiometric ratio AFs, and the catalyst is caused to fully absorb oxygen within at least the preparation period TP2, so that lean gas is caused to flow through the catalyst to the downstream of the catalyst, and the output value Vrr of the post-catalyst sensor 18 eventually becomes a lean-side value.

[0128] The first preparation period TpI is started at the same time that the deterioration detection execution flag is set to ON, and the deterioration detection execution flag is cleared to OFF at the same time that the second determination period Tj2 is terminated. [0129] In the active air-fuel ratio control of the second embodiment, the central air-fuel ratio AFc of the air-fuel ratio oscillation is made equal to the stoichiometric air-fuel ratio AFs, and the lean-side and rich-side amplitudes δAF1, ZiAFr in one cycle of oscillation are set to an equal, constant value that is determined in advance, as shown in the graph (B) of FIG. 20. For example, the central air-fuel ratio AFc, which is equal to the stoichiometric ratio AFs, is equal to 14.6, and the lean-side amplitude δAF1 and rich-side amplitude δAFr are equal to 0.5. Also, the lean-side air-fuel ratio AFl is constantly equal to 15.1, and the rich-side air-fuel ratio AFr is constantly equal to 14.1.

[0130] However, the lean-side period Tl and the rich-side period Tr in one cycle of oscillation have different lengths in time. More specifically, when the first active air-fuel ratio control is performed in which the catalyst is caused to absorb oxygen, the lean-side period Tl is made longer than the rich-side period Tr. When the second active air-fuel ratio control is performed in which oxygen is released from the catalyst, the rich-side period Tr is made longer than the lean-side period Tl. The active air-fuel ratio control of this embodiment is also performed through feedforward control according to a predetermined program.

[0131] Since the lean-side period Tl is made longer than the rich-side period Tr in the first determination period TjI, as described above, the degree of leanness per cycle of oscillation is made larger than the degree of richness, and the

lean/rich balance is biased toward the lean side. Consequently, a certain amount of oxygen is absorbed into the catalyst in the first determination period TjI as a whole. Meanwhile, the rich-side air-fuel ratio AFr is set to a value equal to or smaller than the stoichiometric ratio AFs, and oxygen is released from the catalyst while the target air-fuel ratio AFt is shifted to the rich-side air-fuel ratio AFr. Thus, absorption of oxygen into the catalyst takes place intermittently, and absorption of a relatively large amount of oxygen and release of a relatively small amount of oxygen are alternately repeated.

[0132] Since the rich-side period Tr is made longer than the lean-side period Tl in the second determination period Tj2, the degree of richness per cycle of oscillation is made larger than the degree of leanness, and the lean/rich balance is biased toward the rich side. Consequently, a certain amount of oxygen is released from the catalyst in the second determination period Tj2 as a whole. Meanwhile, the lean-side air-fuel ratio AFl is set to a value equal to or larger than the stoichiometric ratio AFs, and oxygen is absorbed into the catalyst while the target air-fuel ratio AFt is shifted to the lean-side air-fuel ratio AFl. Thus, release of oxygen from the catalyst takes place intermittently, and release of a relatively large amount of oxygen and absorption of a relatively small amount of oxygen are alternately repeated. [0133] Since lean or rich gas is intermittently (i.e., not continuously) supplied to the catalyst, rather than continuously supplied as in the conventional Cmax method, the catalyst undergoes slow changes of state, and the downstream and deep portions of the catalyst are also used for absorption and release of oxygen, so that the true oxygen storage capability can be measured with high accuracy. This also makes it possible to precisely determine the degree of deterioration of the catalyst that has been deteriorated by some degree, and to distinguish, with high accuracy, between a catalyst that is fully deteriorated and a catalyst that is about to deteriorate.

[0134] Next, a catalyst deterioration detecting process according to the

second embodiment will be described with reference to FIGS. 21A and 21B through FIG. 23. FIGS. 21A and 21B are flowcharts of a main routine according to which the deterioration detecting process is implemented, and FIG. 22 is a flowchart of a subroutine executed in step S308 of the main routine, while FIG. 23 is a flowchart of a subroutine executed in step S321 of the main routine. The ECU 20 repeatedly executes these routines at certain minute time intervals (e.g., 16 msec).

[0135] As shown in FIGS. 21A and 21B, it is initially determined in step S301 whether a condition for executing deterioration detection is met, as in the above-mentioned step SlOl of the flowchart of FIG. 16A. If it is determined that the deterioration detection executing condition is not met, the present routine is finished. If it is determined that the deterioration detection executing condition is met, on the other hand, the ECU 20 proceeds to step S302 to set a deterioration detection execution flag to ON (see the graph (A) in FIG. 20).

[0136] After execution of step S302, the ECU 20 proceeds to step S303 to determine whether an initial condition process completion flag (i.e., a flag indicative of completion of a process for establishing an initial condition) is ON. If this flag is not ON, the ECU 20 proceeds to step S308. If this flag is ON, the ECU 20 proceeds to step 304.

[0137] In step S308, a preparation process for deterioration detection as shown in FIG. 22 is performed. The preparation process will be hereinafter described. As shown in FIG. 22, it is initially determined in step S401 whether the target air-fuel ratio (target AF) set flag is ON. If this flag is not ON, the ECU 20 proceeds to step S405 to determine whether the current output value Vrr of the post-catalyst sensor is equal to or larger than a lean/rich determination threshold value Vrrs (see the graph (C) in FIG. 20).

[0138] If the post-catalyst sensor output value Vrr is equal to or larger than the lean/rich determination threshold value Vrrs, it is determined that the current post-catalyst air-fuel ratio is richer than the stoichiometric air-fuel ratio, and the target air-fuel ratio AFt is set to a lean-side air-fuel ratio AFl that is leaner

than the stoichiometric ratio in step S406. If the post-catalyst sensor output value Vrr is smaller than the lean/rich determination threshold value Vrrs, on the other hand, it is determined that the current post-catalyst air-fuel ratio is leaner than the stoichiometric ratio, and the target air-fuel ratio AFt is set to a rich-side air-fuel ratio AFr that is richer than the stoichiometric ratio in step S407. Thus, the target air-fuel ratio AFt is set to the side (lean or rich) opposite to that of the post-catalyst sensor output value Vrr.

[0139] After execution of step S406 or S407, the ECU 20 proceeds to step S408 to set the target air-fuel ratio set flag to ON, and the routine of FIG. 22 is finished.

[0140] If it is determined in step S401 that the target air-fuel ratio set flag is ON, on the other hand, summation of the initial oxygen amount OSAst is performed in the following manner. When the target air-fuel ratio AFt is set to a lean-side air-fuel ratio AFl, the absorbed oxygen amount δOSA of the current cycle of the routine is calculated according to the above -indicated equation (l), and the initial oxygen amount OSAst of this cycle is calculated by adding the absorbed oxygen amount δOSA of this cycle to the initial oxygen amount OSAstold obtained in the last cycle of the routine. When the target air-fuel ratio AFt is set to a rich-side air-fuel ratio AFr, on the other hand, the released oxygen amount δOSA of the current cycle of the routine is calculated according to the above-indicated equation (l)', and the initial oxygen amount OSAst of this cycle is calculated by adding the released oxygen amount δOSA of this cycle to the initial oxygen amount OSAstold obtained in the last cycle of the routine.

[0141] If the summation of the initial oxygen amount OSAst is performed in the above manner, it is determined whether the initial oxygen amount OSAst is equal to or larger than a predetermined value OSAsts. The predetermined value OSAsts is set to a value at which oxygen can be fully absorbed into or released from the catalyst. If the initial oxygen amount OSAst is not equal to or larger than the predetermined value OSAst, the present routine is finished. If the initial oxygen

amount OSAst is equal to or larger than the predetermined value OSAsts, the initial condition process completion flag is set to ON in step S404, and the present routine is finished.

[0142] When the preparation process of FIG. 22 is executed for the first time, a negative decision (NO) is made in step S401, it is determined in step S405 whether the current post-catalyst sensor output is rich or lean. Then, the target air-fuel ratio is set to the side (lean or rich) opposite to that of the post-catalyst sensor output. Namely, the target air-fuel ratio is set to a lean value if the sensor output is rich, and is set to a rich value if the sensor output is lean. [0143] If the target air-fuel ratio is set to a lean value, for example, lean gas whose air-fuel ratio is leaner than the stoichiometric ratio is continuously supplied to the catalyst in the second and subsequent cycles of the routine of FIG.

22. During this period, the amount of oxygen absorbed into the catalyst is summed in step S402. When the total value of the absorbed oxygen amount reaches the predetermined value OSAsts, the catalyst is in a condition in which the catalyst fully absorbs oxygen, namely, oxygen is stored in the catalyst to the full storage capacity. Thus, the preparation process for subsequent active air-fuel ratio control is substantially finished.

[0144] Returning to FIGS. 21A and 21B, it is determined in step S304 whether this step is executed for the first time. If it is determined that step S304 is executed for the first time, steps S305, S306, S307 are executed. These steps

S305, S306, S307 are similar to steps S405, S406, S407 of the routine of FIG. 22, respectively. With these steps S305 - S307 executed, the target air-fuel ratio is set to the side (lean or rich) opposite to that of the current post-catalyst sensor output, and the active air-fuel ratio control is started. After execution of steps

S305 — S307, the present routine is finished.

[0145] If it is determined in step S304 that this step is not executed for the first time (namely, the current cycle of the routine is the second or subsequent cycle), the ECU 20 proceeds to step S309 to determine whether the post-catalyst

sensor output has been reversed, namely, whether the post-catalyst sensor output Vrr has changed from a value larger than the lean/rich determination threshold value Vrrs to a value smaller than the threshold value Vrrs, or vice versa.

[0146] If it is determined in step S309 that the post-catalyst sensor output has not been reversed, the ECU 20 proceeds to step S320 to perform summation of the amount OSA of oxygen absorbed into or released from the catalyst. This summation is performed in substantially the same manner as the summation of the initial oxygen amount OSAst in step S402. More specifically, where the target air-fuel ratio AFt is set to a lean-side air-fuel ratio AFl, the absorbed oxygen amount δOSA of the current cycle of the routine is calculated according to the above-indicated equation (l), and the oxygen amount OSA of this cycle is calculated by adding the absorbed oxygen amount δOSA of this cycle to the oxygen amount OSAoId obtained in the last cycle of the routine. Where the target air-fuel ratio AFt is set to a rich-side air-fuel ratio AFr, on the other hand, the released oxygen amount δOSA of this cycle is calculated according to the above -indicated equation (l)\ and the oxygen amount OSA of this cycle is calculated by adding the released oxygen amount δOSA of this cycle to the oxygen amount OSAoId of the last cycle.

[0147] Next, the ECU 20 proceeds to step S321 in which a target oxygen amount (which will also be called "target OSA") is set according to the routine as shown in FIG. 23. Here, a process of setting the target oxygen amount will be described.

[0148] As shown in FIG. 23, step S501 is initially executed to calculate a target oxygen amount correction factor according to a certain map (or function) as shown in FIG. 24, based on the value of the intake air amount GA detected by the air flow meter 5. It is understood from this map that the intake air amount GA and the target oxygen amount correction factor have a predetermined relationship, and that the target oxygen amount correction factor, which is always smaller than 1, approaches 1 as the intake air amount GA increases.

[0149] Once setting of the target oxygen amount correction factor is

finished, it is determined in step S502 whether the target air-fuel ratio AFt obtained in the first cycle of the routine of FIGS. 21A and 21B, i.e., the target air-fuel ratio AFt set in step S306 or S307, is a rich-side air-fuel ratio AFr. Namely, it is determined whether the currently executed active air-fuel ratio control is intended to initially supply lean gas to the catalyst so as to cause the catalyst to absorb oxygen, or initially supply rich gas to the catalyst so as to cause the catalyst to release oxygen therefrom.

[0150] If the target air-fuel ratio AFt obtained in the first cycle is a rich-side air-fuel ratio AFr, the ECU 20 proceeds to step S503 to calculate a rich-side target oxygen amount according to a certain map (or function) as shown in FIG. 25, based on the intake air amount GA detected by the air flow meter 5. It is understood from this map that the intake air amount GA and the target oxygen amount have a predetermined relationship, and that the target oxygen amount tends to decrease as the intake air amount GA increases. Accordingly, the period of oscillation of the active air-fuel ratio control becomes shorter as the intake air amount GA increases, as will be understood later. Also, a lean-side target oxygen amount is calculated by multiplying the rich-side target oxygen amount by the target oxygen amount correction factor obtained in step S501. Since the target oxygen amount correction factor is smaller than 1, the lean-side target oxygen amount is smaller than the rich-side target oxygen amount. Accordingly, the rich-side period Tr in one cycle of oscillation of the active air-fuel ratio control becomes longer than the lean-side period Tl, as will be understood later.

[0151] If it is determined in step S502 that the target air-fuel ratio AFt obtained in the first cycle of the routine of FIGS. 21A and 21B is not a rich-side air-fuel ratio AFr (namely, it is a lean-side air-fuel ratio AFl), the ECU 20 proceeds to step S504 to calculate a lean-side target oxygen amount according to the map as shown in FIG. 25, based on the detected value of the intake air amount GA, in the same manner as in step S503. Also, a rich-side target oxygen amount is calculated by multiplying the lean-side target oxygen amount by the target oxygen

amount correction factor obtained in step S501. Thus, the rich-side target oxygen amount becomes smaller than the lean-side target oxygen amount, and the lean-side period Tl in one cycle of oscillation of the active air-fuel ratio control becomes longer than the rich-side period Tr. After execution of step S503 or S504, the present routine is finished.

[0152] Returning to FIGS. 21A and 21B, step S321 is followed by step S322 in which it is determined whether the oxygen amount OSA obtained through summation in step S320 is equal to or larger than the target oxygen amount calculated in step S321. If the oxygen amount OSA is smaller than the target oxygen amount, the present routine is finished.

[0153] If the oxygen amount OSA is equal to or larger than the target oxygen amount, on the other hand, it is determined in step S 323 whether the current target air-fuel ratio AFt is a rich- side air-fuel ratio AFr. If the current target air-fuel ratio AFt is a rich-side air-fuel ratio AFr, the ECU 20 proceeds to step S324 to set the target air-fuel ratio AFt to a lean-side air-fuel ratio AFl opposite to the current target air-fuel ratio AFr. If the current target air-fuel ratio AFt is not a rich-side air-fuel ratio AFr (namely, it is a lean-side air-fuel ratio AFl), on the other hand, the ECU 20 proceeds to step S325 to set the target air-fuel ratio AFt to a rich-side air-fuel ratio AFr opposite to the current target air-fuel ratio AFl. Thus, the target air-fuel ratio AFt is reversed or switched from lean to rich or vice versa each time the oxygen amount OSA reaches the target oxygen amount. After execution of step S324 or S325, the present routine is finished.

[0154] In the meantime, if it is determined in step S309 that the post-catalyst sensor output has been reversed, the ECU 20 proceeds to step S310 to increment a counter for counting the number of reversals of the post-catalyst sensor output. The ECU 20 then proceeds to step S 311 to determine whether the post-catalyst sensor output Vrr that has been reversed is equal to or larger than the lean/rich determination threshold value Vrrs.

[0155] If the post-catalyst sensor output value Vrr that has been reversed

is equal to or larger than the lean/rich determination threshold value Vrrs, it is assumed in step S 312 that rich gas was mainly supplied to the catalyst before the reversal so that oxygen was released from the catalyst. In this case, the total oxygen amount OSA obtained in step S320 is stored as a released oxygen amount OSAh in step S320. If the post-catalyst sensor output value Vrr that has been reversed is smaller than the lean/rich determination threshold value Vrrs, on the other hand, it is assumed in step S313 that lean gas was mainly supplied to the catalyst before the reversal so that the catalyst absorbs oxygen. In this case, the total oxygen amount OSA obtained in step S320 is stored as an absorbed oxygen amount OSAk in step S320.

[0156] Subsequently, it is determined in step S314 whether the counter for the number of reversals of the post-catalyst sensor output is equal to a predetermined value. This predetermined value is preferably an even number, for example, 2. If the counter for the number of reversals of the post-catalyst sensor output is not equal to the predetermined value, the present routine is finished.

[0157] If the counter for the number of reversals of the post-catalyst sensor output is equal to the predetermined value, an average oxygen amount OSAave is calculated in step S315 according to the following equation^ OSAave = (OSAh + OSAk) / 2. In step S316, the average oxygen amount OSAave is compared with a predetermined deterioration determination threshold value. If the average oxygen amount OSAave is larger than the deterioration determination threshold value, it is determined that the catalyst is normal. If the average oxygen amount OSAave is equal to or smaller than the deterioration determination threshold value, it is determined that the catalyst is deteriorated. Thereafter, the deterioration detection execution flag is cleared or set to OFF in step S319, and the present routine is finished.

[0158] The deterioration detecting process as described above will be further explained when it is applied to the specific example as shown in FIG. 20. If the deterioration detection execution flag is set to ON (step S302), the

preparation process (step S308) is initially carried out, and the preparation period TpI is started. In the preparation process, the target air-fuel ratio AFt is set to the side (rich side in FIG. 20) opposite to that (lean side in FIG. 20) of the post-catalyst sensor output detected when the flag is set to ON (steps S405 — S407). As a result, oxygen is released from the catalyst. The released oxygen amount δOSA is successively added up (step S402) to obtain a total value of the initial oxygen amount OSAst. The oxygen is already completely released from the catalyst by the time when the total value reaches the threshold value OSAsts (YES in step S403). Upon completion of the release of oxygen, rich gas flows through the catalyst to the downstream thereof, and the post-catalyst sensor output is reversed to the rich side. Thus, when the total value of the initial oxygen amount OSAst reaches the threshold value OSAsts, the preparation process and the preparation period TpI are finished. By performing the preparation process as described above, it is possible to reduce variations in the oxygen storage capacity for each measurement.

[0159] Once the preparation process is finished, an affirmative decision (YES) is made in step S303, and the first active air-fuel ratio control and first determination period TjI are started. Initially, the target air-fuel ratio AFt is set to the side (lean in FIG. 20) opposite to that (rich in FIG. 20) of the post-catalyst sensor output detected at this point in time (steps S305 — S307). As a result, absorption of oxygen into the catalyst is initiated. The absorbed oxygen amount δOSA is added up (step S320) for summation of the absorbed oxygen amount OSA during absorption of oxygen, and, when the absorbed oxygen amount OSA reaches the lean-side target oxygen amount OSAl (YES in step S322), the target air-fuel ratio AFt is reversed to the rich side (step S325). As a result, release of oxygen from the catalyst is initiated. The released oxygen amount δOSA is added up (step S320) for summation of the released oxygen amount OSA during release of oxygen, and, when the released oxygen amount OSA reaches the rich-side target oxygen amount OSAr (YES in step S322), the target air-fuel ratio AFt is reversed

to the lean side (step S 324).

[0160] Since the lean-side target oxygen amount OSAl is larger than the rich-side target oxygen amount OSAr (step S504), the lean-side period Tl is made longer than the rich-side period Tr. Accordingly, the degree of leanness per cycle of oscillation becomes larger than the degree of richness, and absorption of a relatively large amount of oxygen and release of a relatively small amount of oxygen take place.

[0161] It is understood from the map shown in FIG. 24 that, as the intake air amount GA increases, the target oxygen amount correction factor approaches 1, and the rich-side period Tr is made closer to the lean-side period Tl, so that the degree of richness per cycle of oscillation is made closer to the degree of leanness. This is because, as the intake air amount GA increases, the flow rate of gas supplied to the catalyst increases, and the gas is more likely to flow through the catalyst without being treated by the catalyst due to an insufficient reaction speed of the catalyst. If the degree of leanness per cycle of oscillation is controlled to be more equivalent to the degree of richness as the intake air amount GA increases, as in the present embodiment, the overall absorption or release of oxygen into or from the catalyst takes place more slowly, and breakthrough of the gas can be prevented, namely, the gas is prevented from flowing through the catalyst without being treated by the catalyst. It is thus possible to prevent the oxygen storage capacity of the catalyst from being estimated to be smaller than a true value due to the breakthrough of the gas, thus assuring improved detection accuracy, and is also possible to prevent the system from making an erroneous determination on catalyst deterioration. [0162] As is also understood from the map as shown in FIG. 25, the target oxygen amount is reduced as the intake air amount GA increases, in view of the breakthrough of the gas as described above. With this arrangement, as the intake air amount GA increases, the period of time required to reach the target oxygen amount is reduced, and the lean-side period Tl and the rich-side period Tr are

shortened. As a result, the period of the oscillation can be shortened (namely, the frequency can be increased), so that breakthrough of the gas and an erroneous deterioration determination caused by the breakthrough of the gas can be prevented, and the detection accuracy can be improved.

5 [0163] With the target air-fuel ratio repeatedly reversed to the lean side and rich side as described above, the absorbed oxygen amount of the catalyst gradually increases. When the output of the post-catalyst sensor is reversed to the lean side (YES in step S309, time trl in FIG. 20), the absorbed oxygen amount OSAk obtained at this point in time is stored (step S313). Then, the first active

.0 air-fuel ratio control and first determination period TjI are finished.

[0164] Next, the second preparation process and second preparation period Tp2 are initiated. In the preparation period Tp2, the target air-fuel ratio AFt is kept at a lean-side air-fuel ratio AFl, and the catalyst is caused to absorb oxygen. At this point in time, however, the catalyst is already saturated with oxygen, and

L5 therefore, oxygen absorption takes place just for the sake of guarantee. In the meantime, summation of the initial oxygen amount OSAst of the catalyst is successively performed, and the second preparation period Tp2 is terminated when the total value reaches the threshold value OSAsts. The second preparation process is performed according to a routine (not shown) that is different from the

20 routine shown in FIG. 22.

[0165] Concurrently with the termination of the second preparation period Tp2, the second active air-fuel ratio control and second determination period Tj2 are started. Initially, the target air-fuel ratio AFt is set to the side (rich in FIG. 20) opposite to that (lean in FIG. 20) of the post-catalyst sensor output detected at 5 this point in time (steps S305 — S307). As a result, release of oxygen from the catalyst is started. During the period of release of oxygen, summation of the released oxygen amount OSA is conducted (step S320). When the released oxygen amount OSA reaches the rich-side target oxygen amount OSAr (YES in step S322), the target air-fuel ratio AFt is reversed to the lean side (step S324). As a result,

absorption of oxygen into the catalyst is started. During the period of absorption of oxygen, summation of the absorbed oxygen amount OSA is conducted (step S320). When the absorbed oxygen amount OSA reaches the lean-side target oxygen amount OSAl (YES in step S322), the target air-fuel ratio AFt is reversed to the rich side (step S325).

[0166] In the second active air-fuel ratio control, the rich-side target oxygen amount OSAr is larger than the lean-side target oxygen amount OSAl (step S503), contrary to the first active air-fuel ratio control. Accordingly, the rich-side period Tr becomes longer than the lean-side period Tl, and the degree of richness per cycle of oscillation becomes larger than the degree of leanness. Thus, release of a relatively large amount of oxygen and absorption of a relatively small amount of oxygen take place.

[0167] As the intake air amount GA increases, the lean-side period Tl is made closer to the rich-side period Tr, and the degree of leanness per cycle of oscillation is made closer to the degree of richness, as in the case of the first active air-fuel ratio control. Also, as the intake air amount GA increases, the lean-side period Tl and rich-side period Tr are shortened, and the period of oscillation is shortened, as in the case of the first active air-fuel ratio control.

[0168] With the target air-fuel ratio repeatedly reversed to the rich side and lean side as described above, the released oxygen amount of the catalyst gradually increases. When the output of the post-catalyst sensor is reversed to the rich side (YES in step S309, time tlr in FIG. 20), the released oxygen amount OSAh reached at this point in time is stored (step S312). Then, the second active air-fuel ratio control and second determination period Tj2 are terminated. [0169] Subsequently, the average oxygen amount OSAave as an average value of the stored absorbed oxygen amount OSAk and released oxygen amount OSAh is calculated, and the average oxygen amount OSAave is compared with the deterioration determination threshold value so as to determine whether the catalyst is normal or deteriorated (steps S315 — S318).

[0170] While the preferred embodiments of the invention have been described in detail, various other embodiments of the invention may be contemplated. For example, while the above-described internal combustion engine is of the direct injection type adapted to inject the fuel directly into the cylinders, the invention may be equally applied to an internal combustion engine of an intake port (intake passage) injection type adapted to inject the fuel into the intake port or passage, or of a dual injection type serving as both the direct injection type and the intake port injection type.

[0171] In the illustrated embodiments, the ECU 20 and the injector 12 provide the above-mentioned active air-fuel ratio control unit, and the ECU 20 provides the locus-length deterioration determining unit, oxygen amount calculating unit, oxygen- amount deterioration determining unit, and the fuel cut request time control unit. Also, the air flow meter 5 provides the intake air amount detector. [0172] It is to be understood that the invention is not limited to the above -de scribed embodiments, but all modified examples, applications and equivalents that are covered by the concept or principle of the invention as defined in the appended claims are included in the present invention. Thus, the invention should not be limitedly interpreted, but may be applied to other technologies that pertain to or belong to the scope of the invention.