METHOD AND APPARATUS FOR MACHINE HEALTH INFERENCE BY COMPARING TWO LIKE LOADED COMPONENTS

Technical Field

This invention relates generally to a machine diagnostic system and more particularly to a system for selectively processing operating parameter data to provide data useful for diagnosing and predicting component degradation or failure.

In this specification the term "component" refers to any machine component or system with a measurable operating output shared by another such component or system, such that two or more components operating in parallel have parallel outputs.

Background Art

For service and diagnostic purposes, machines are sometimes equipped with sensors for measuring operating parameters such as engine RPM, oil pressure, water temperature, boost pressure, oil contamination, electric motor current, hydraulic pressure, system voltage, exhaust manifold temperature and the like. In some cases, storage devices are provided to compile a database for later evaluation of machine performance and to aid in diagnosis. Service personnel examine the accrued data to determine the cause(s) of any failure or to aid in diagnosis. Similarly, service personnel can evaluate the stored data to predict future failures and to correct any problems before an actual failure occurs. Such diagnosis and failure prediction are particularly pertinent to over-the-highway trucks and large work machines such as off-highway mining trucks, hydraulic

excavators, track-type tractors, wheel loaders, and the like. Stationary machines such as large stationary engines can also benefit from accurate component diagnosis and failure prediction. These machines represent large capital investments and are capable of substantial productivity when operating properly. It is therefore important to fix or replace degraded components and to predict failures so minor problems can be repaired before they lead to catastrophic failures, and so servicing can be scheduled during periods in which productivity will be least affected.

Systems used in the past often acquire and store data from the machine sensors during different machine operating conditions. For example, some data is required while the engine is idling while other data is acquired while the engine is under full load. This poses a problem for service personnel to compare data acquired under such different circumstances and to observe meaningful trends in the sensed parameters.

It is sometimes advantageous to accumulate parameters only when the machine is in a particular operating condition. This type of information is predominantly used during performance evaluation but may also be used in failure diagnosis and prognosis.

To further aid in diagnostics, it is beneficial to package information in such a way that the analysis is simplified as much as possible. Where a machine includes similar or identical components or systems loaded in parallel during operation, comparisons are sometimes made between two or more of these like-loaded components to diagnose problems or predict failure. One known method is the "hands-on" approach, in which the machine operator or maintenance personnel physically

look, listen or feel for differences between the two components during operation. For example, machine operators have been known to feel for temperature differences between cylinder banks or ports looking for causes of poor fuel mileage.

Another known technique for comparing similarly-loaded components is the "drop-out test", in which individual components are selectively turned off to determine which dropped-out component has the least effect on engine or system output.

Other known methods for assessing the comparative performance of various components include the use of chassis dynamometers for measuring the difference between vehicle wheel or drive shaft performance, exhaust pyrometer and cylinder port gages to indicate differences in exhaust manifold or cylinder temperatures, and other non-historical comparison methods and devices. Drawbacks of known component-comparing methods and devices include the subjective nature of many tests; the fact that many tests cannot be performed under actual operating conditions; the lack of historical component-to- component comparison data accumulated during periods of vehicle operation; and difficulty in usefully comparing the data from multiple components such that degradation or failure of one or more of them can be readily determined.

Disclosure of the Invention The present invention provides indications of historical operating data for two or more similar machine components operated in parallel . The difference between an operating parameter level of a plurality of parallel components is trended. The information is useful for diagnosing degradation or

predicting failure based on the trended difference in performance between the components.

In general, the present invention is achieved with a system for sensing the operating parameters of similar components operated in parallel to produce signals or data indicative of the level of a common operating parameter. Means are provided for determining a difference between the operating parameters of sensed components, and for averaging the difference between them over a trend period. A trend value based on the average of the difference is then generated for that trend period. Trend values are used to generate trends indicating the relative performance of the components. In a further aspect of the invention, trend values are generated within a trap window or trend definition satisfied when one or more machine operating parameters (e.g. engine speed, rack, coolant temperature) are at desired levels. A plurality of trap windows may be defined so that the components can be compared and trends evaluated under different machine operating conditions.

A trap window may be narrowed to increase the accuracy of the trending by trapping or accumulating component readings after a specified trap-lag period has occurred. The trap-lag period can comprise a period in which one or more machine dependency parameter levels are continuously met. The method and apparatus of the present invention may be performed and located on the machine on which components are being measured. Preferably, the system includes both on and off-machine operations to increase the flexibility of failure diagnosis and prediction. For example, trend values can be generated on the machine and transmitted to a remote

station where trends are generated for diagnostic observation and review.

In general, changes in the difference between component operating parameters as determined by the invention will indicate a problem with one of the components. The trended difference between the components can be checked against predefined limits to generate a warning signal to the operator or diagnostic personnel. In one embodiment of the invention the difference between two identical components is trended. In another embodiment of the invention the difference between an individual component and the average of a plurality of components is trended. The plurality of components may include the first component.

These and other features of the present invention will become apparent upon further reading of the specification.

Brief Description of the Drawings

FIG. 1 is a schematic diagram of an engine with a split exhaust manifold, and a system according to the present invention for trending the difference between the split exhaust temperatures to diagnose and predict degradation or failure of components;

FIG. 2 is a graphical representation of trend data generated by the system of FIG. 1;

FIGS. 3 and 3A are schematic flowcharts of a method used in the embodiment of FIG. 1 to generate the trend data of FIG. 2;

FIG. 4 is a schematic diagram of an engine with individual cylinder exhaust port temperature measurements, and a system according to the present

invention for trending the difference between individual cylinders and cylinder bank averages;

FIG. 5 is a graphical representation of trend data generated by the system of FIG. 4; and FIG. 6 is a schematic flowchart of a method used in the embodiment of FIG. 4 to generate the trend data of FIG. 5.

Best Mode for Carrying Out the Invention Referring to Figure 1, an exemplary system for carrying out one application of the present invention is shown for a machine engine comprising an engine block (12) , a number of cylinders (14) with associated exhaust ports (16) , and split exhaust manifolds (18,20) leading to turbocharger turbine inlets at (22) . The machine engine is provided with an electronic engine control (24) of known type for monitoring operating parameters such as engine RPM, rack, and coolant temperature. For purpose of illustration the engine (12) represents a six-cylinder diesel engine for over-the-highway trucks, available from Caterpillar Inc.

Exhaust temperature measurements at each turbocharger turbine inlet from manifold splits (18,20) indicate approximate mean gas temperature from each bank of cylinders. Each manifold split is provided at the turbine inlet with an exhaust temperature probe (26) of know type, for example a thermocouple exhaust probe. The electrical signals from probes (26) are delivered to an RS-485 converter module (28) where the input from thermocouple probes (26) is converted into digital temperature readings.

The system further includes a diagnostic ^■ monitor (30) , in Figure 1 onboard the machine with serial port connections to electronic engine control

(24) and converter module (28) . In the illustrated embodiment of Figure 1 diagnostic monitor (30) comprises a main microprocessor and battery-backed working and data memory (two M-Bytes RAM) , Flash program memory (256 K-Bytes) , a real-time clock chip, and RS-232 ports for telemetry hook-up and programming/user interface. Diagnostic monitor (30) also includes a data input processor and serial ports connected to electronic engine control (24) via SAE J- 1708/J-1587 "ATA" data link and to converter module (28) via RS-485 data link.

A software trending module (32) is stored in the memory of diagnostic monitor (30) .

Diagnostic monitor (30) receives machine operating information such as engine RPM, rack and coolant temperature from electronic engine control (24) . Diagnostic monitor (30) also receives exhaust temperature information for manifold splits (18,20) from converter module (28) . Trending module (32) reads and manipulates this information to generate trend information for the sensed component parameters. Trend information determined by the trending module (32) in diagnostic module (30) is transmitted by RS-232 port connection to a telemetry device (36) and then to a similarly-equipped base workstation computer (37) . Telemetry devices (36) can comprise known transmitter to transceiver apparatus such as a two-way radio transceiver or satellite transceiver, for example. Alternately, or additionally, trend information determined by the trending module in diagnostic monitor (30) may be stored and graphically displayed onboard the machine with a suitable operator display/interface (34) .

In the illustrated embodiment, the trend information received by the base computer from transceiver device (36) is stored and translated by the base computer into graphic representations of the component trends, for example on an hourly, daily or weekly basis. Since the historical trends can be monitored by maintenance personnel off the machine, diagnosis can be conveniently scheduled on a periodic basis. Warnings are generated by the off-board computer whenever high or low limits for the exhaust temperature splits are exceeded. Optionally, warnings can be generated onboard the machine by the trending module software in the diagnostic monitor (30) to trigger an alarm at display (34) and relay a diagnostic code via telemetry device (36) to the base station.

In the split exhaust manifold application illustrated in Figure 1, a change in the average difference between the split exhaust temperatures indicates degradation of a fuel injector, for example. Should one fuel injector fail to deliver fuel, the split of the associated engine bank will be cooler than the temperature measured for the other bank. By comparing the temperature splits and setting both high and low limits on their trended difference, diagnosis of fuel-delivery related problems can be made.

Referring now to Figure 2, a graph illustrates trended data points (38) from the sensed component machine parameter (exhaust temperature) . Such displays of historical trend information can be generated by workstation computer (37) and/or the diagnostic monitor (30) on associated printer or video display apparatus from the trend data produced by trending module (32) . By viewing the trend of the sensed parameter, component degradation and resulting

^■ Si -

failure can be identified early by observing, for example, significant increases or decreases in the sensed average temperature split over a period of time. Figure 2 shows the results of trend analysis for the over-the-highway truck engine of Figure 1 over 2,380 hours of operation. At approximately 143 hours a noticeable increase in the average exhaust temperature difference can be observed, coinciding with the beginning of degradation of a fuel injector to one of the engine banks. The graph of Figure 2 shows that the degradation progressively increased, up to the time the faulty fuel injector was replaced at approximately 250 hours. The trend graph then returned to normal for the remainder of the measured period. In the illustrated embodiment, an off-board warning was triggered at approximately 180 engine hours when a predefined limit on the trended difference between the exhaust manifolds was exceeded. The faulty fuel injector could then be diagnosed and monitored until repair/replacement was determined to be convenient or necessary at the 250-hour mark. Referring now to Figures 3 and 3A, the method employed by the trending module (32) in diagnostic monitor (30) to generate trend information and appropriate warning signals is illustrated in flow chart form.

Each data point plotted on the graph of Figure 2 is an hourly trend point representing the average temperature difference between the exhaust manifold splits as determined by the system of Figure 1 over a predefined trend period of one hour. The trend period for which each data point is determined may be measured either in time such as the illustrated period of one hour, or in counts measured in changes in the operating state of the machine or a system or

component thereof, for example transmission shifts. A trend period of one hour is used in the illustrated embodiment for ease of explanation.

A subset of machine operating parameters for which trend data is to be produced is also predefined, comprising a trending definition or "trap window" which must be satisfied before trend data is produced. The trending definition or trap window for different sensed parameters will vary, and may be a function of one or more machine operating parameters, sometimes referred to as "dependency" parameters. Trend data is gathered and stored in memory as the specified dependency definition in the trap window is met over a specified trend period. For example, in the illustrated embodiment of Figures 1-3, two trending definitions or trap windows are defined by two different combinations of engine RPM, rack and coolant temperature levels or ranges corresponding to idle and loaded operating conditions of the machine. Of course, the machine engine must be running as a prerequisite to the idle and loaded trap windows being satisfied.

Referring still to Figure 3, the system is initially powered up at step 104, for example automatically when the operator turns the machine on. At step 104 the trending module collects sensor data T _X ,T ₂ representing the temperatures of the parallel components (exhaust manifold splits 18,20) as measured by probes (26) . At step 106 the difference T _spUt between the exhaust temperatures of the split manifolds is calculated by subtracting T _x from T ₂ . At step 108 the system determines whether the trap window conditions are currently met using the algorithm illustrated in Figure 3A and described in further detail below. If the trap window conditions have not

been satisfied, the system proceeds directly to step 112 to determine whether the trend period, in the illustrated embodiment one hour, has expired. If not, the system returns to step 104 and the loop is repeated.

If the trap window conditions are met at step 108, the system proceeds first to step 110 where the current temperature differential value T _split is added to any previous sum of temperature differences T _splitsum to "accumulate" or increment the value T _splitsuιn . A count of the differential temperature readings accumulated for the current trend period is also incremented at step 110. The system then proceeds to step 112, where, as described above, it determines whether the current trend period of one hour has expired. If not, the loop is again repeated beginning at step 104 with the collection of component sensor data.

The loop of steps 104,106,108,110,112 continues until the trend period has expired, at which point the system proceeds to step 114 to determine whether enough differential temperature readings have been determined and accumulated for that period to provide an adequate basis for a trended data point for that trend period. If the total number of differential temperature readings T _splitcounter for the last trend period is less than a minimum defined trend count, the system proceeds to step 116 where the trend point for that trend period is tagged invalid and thereafter ignored by the system. If, however, the minimum trend count is satisfied at step 114, the system proceeds to step 118 where a trended data point T- ^piitavg (the average temperature split for the last trend period) is determined by dividing the sum of the

temperature split readings by the total number of counts recorded for the trend period.

At step 120 the system generates and delivers a signal representing the trended data point for the last trend period. This signal is stored and translated by the associated computer apparatus (37) and/or (30) into a plotted trend point on a graph display as shown in Figure 2. At step 122 the system checks the trended data point T _splitavg against predefined high ^* and low temperature split limits to determine whether a warning signal is needed to indicate possible degradation of a measured component. In the illustrated embodiment of Figure 3, if the trended temperature split for the last trend period is below a minimum limit or above a maximum limit, a warning is generated and delivered at step 124. If the average temperature split at step 122 falls between the high and low limits, no warning signal is generated. After either step 122 or 124, the system next proceeds to step 102 where the accumulated sum of the temperature splits and the differential temperature reading counter are reset to zero. The system then repeats the process for the next trend period. In the illustrated embodiment of Figure 1, steps 122 and 124 are performed at base station computer (37) off the machine. Alternately, or additionally, steps 122,124 can be performed onboard the machine by supplementing the diagnostic monitor (30) with suitable software and hardware for comparing the trended data points with predefined limits; displaying the trended data points, for example in graph form as shown in Figure 2; and generating a warning signal, for example a diagnostic code and/or

audiovisual signals such as flashing lights, buzzers or displays.

Referring now to Figure 3A, the subroutine for determining whether trap window conditions have been met for step 108 in Figure 3 is illustrated.

In the illustrated embodiment of Figures 1-3 for the trending of temperature differences between the split exhaust manifolds of an engine, two trap windows are defined: the first trap window for collecting data when the engine is in a loaded condition; and a second trap window for collecting data when the engine is in an idle condition.

The trap window determination illustrated at step 108a in Figure 3A is generic to the specific trap windows for loaded and idle conditions. The specific trap windows in the illustrated embodiment for loaded and idle conditions, respectively, are as follows:

When Engine Speed > MIN_LOADCHECK_ENGINE_SPEED and Engine Speed < MAX_LOADCHECK_ENGINE_SPEED and Fuel Demand (Rack) > MIN_LOADCHECK_RACK and Coolant temperature > MIN_LOADCHECK_COOLANT_TEMPERATURE

When Engine Speed > MIN_IDLECHECK_ENGINE_SPEED and Engine Speed < MAX_ID ECHECK_ENGINE_SPEED and Fuel Demand (Rack) > MIN_IDLECHECK_RACK and Coolant temperature > MIN_IDLECHECK_COOLA T_TEMPERATURE

If the above conditions are met, the trap windows for loaded and idle operating conditions are satisfied and the data is trended. The graph of trended data points in Figure 2 is drawn from data collected within the loaded and idle trap windows over the 2380 operating hours.

If the appropriate trap window is not met, the system proceeds to step 108b wherein a count of time in which the machine operating parameters have been within the trap window is registered as zero. At step 108c the system tags the temperature split

reading as false since the machine is not operating within the trap window. The system then returns to step 108 in Figure 3 and then to step 112.

If the trap window conditions are satisfied at step 108a, the system proceeds to step 108d where the trap time count is incremented. At step 108e a "trap lag" determination is made to further refine the trending by eliminating potentially false component data readings. The trap lag is defined as a period during which the current trap window conditions at step 108a have been continuously met. Data from the machine component sensors is ignored for trending purposes until the trap window conditions have been continuously met for the trap lag period. The trap lag time can be set based on experience or other factors to account for machine component hysteresis as machine operating conditions change. For example, when the machine is first turned on the trap window for idle conditions may be satisfied, but the exhaust manifold temperatures will not reach a steady temperature until the machine has properly idled for a minimum period. Data trended before the trap lag period expires may be inaccurate. Likewise, when switching from idle to a loaded operating condition, some amount of time is needed before the exhaust temperatures T _1# T ₂ reach a steady state for the loaded condition. The trap lags in the illustrated embodiment for loaded and idle conditions, respectively, are as follows:

Engine Parameters have been continuously met for LOADCHECK for greater than MIN_LOADCHECK_LAG_TIME seconds

Engine Parameters have been continuously met for ID ECHECK for greater than MIN_IDLECHECK_LAG_TIME seconds

The trap lag accordingly serves to narrow the trap window to eliminate the effects of component hysteresis as the machine changes operating state or condition, allowing the sensed components to "warm up" before their data is trended. It will be understood that a trap lag can be employed for virtually any set of machine components which are subject to changes of state as the machine is operated.

If the trap time is less than the trap lag at step 108e, the trend data is designated as false at step 108c and the system returns to steps 108 and 112 in Figure 3. If the trap time is greater than the trap lag, indicating that the components have reached a steady state within the current trap window, the system proceeds to step 108f at which the trend data is designated as true. The system then proceeds to step 110 in Figure 3 to increment or accumulate the sum of the temperature split values and to generate a trend data point for the current trend period. It will be apparent to those skilled in the art that the apparatus and method described above for trending the operating temperature of parallel exhaust manifolds can be applied to many types of machine components operated in parallel with common measurable outputs. The presently-illustrated components (exhaust manifolds) and operating parameters (temperature) are but one specific example.

In the illustrated embodiment of Figures 1- 3, diagnosis/prognosis of component degradation and failure is determined for parallel operating components by trending the average difference between a common operating parameter (temperature) . In the alternate and preferred embodiment of Figures 4-6, the usefulness of trend information is enhanced by trending an individual component versus the average of

multiple components. This method improves the sensitivity of individual component diagnosis. Referring now to Figure 4, a further exemplary application of the present invention is illustrated for a stationary generator set engine with cylinder port temperature measurements. In this application the difference between each cylinder port and the average temperature of multiple cylinder ports is trended, representing a preferred method for providing sensitive individual component diagnostics. The generator engine is illustrated comprising two engine banks (212) , each having eight cylinders (214) with individual exhaust ports (216) . The exhaust ports (216) of engine banks (212) feed into parallel turbocharger turbine inlets (222) .

In Figure 4, the parallel components to be monitored (the cylinders) are equipped for individual exhaust port measurements via thermocouple probes (224) located at each exhaust port (216) . Probes (224) in Figure 4 are K-type thermocouple exhaust port probes of a known type.

The temperature output signal from each exhaust port probe (216) is delivered via communication line or ring (217) to a programmable logic controller (PLC) (226) which monitors and controls the function of the engine and associated generator apparatus (not shown) in known manner. PLC (226) in the present invention is analogous to the electronic engine control and converter module structure in the embodiment of Figure 1, and provides a convenient apparatus for gathering dependency parameter and component data from the engine and sensors (224) and converting it to digital form. Individual port measurement information is delivered from PLC (226) to personal computer (228) via RS-232

serial link, where trending module software determines hourly trend information from the individual port measurements. The hourly trend information is then delivered via phone modem (229) to a master command and control workstation computer (230) at a base site for generating trends and monitoring the operation of the machine. Alternately, the trending module and generation of hourly trend information can be located and performed offsite at the master workstation computer.

It will be understood by those skilled in the art that the apparatus illustrated in Figure 4 for receiving, organizing and transmitting information derived from the individual port measurements (226,228,230) could be replaced with the diagnostic monitor and transceiver apparatus of the embodiment of Figure 1 for a mobile machine application. Figure 4 illustrates that the apparatus used for measuring and collecting parameter data for the machine components can take alternate forms, depending on the nature of the machine. As in the embodiment of Figure 1, the apparatus for collecting and trending machine component measurements can be located on or off the machine in various combinations, with communication of data through suitable telemetry devices.

Referring now to Figure 5, a graph illustrates plotted trends for two different trends measured by the system of Figure 4. Upper trend plot (240) represents hourly measurements of the individual port temperature of a single cylinder (e.g., No. 14), in °C corresponding to y-axis scale (242) . Lower trend plot (244) represents hourly measurements of the difference between the individual port temperature of cylinder No. 14 and the associated cylinder bank average temperature. The port minus bank average

trend points of plot (244) correspond to the inner y- axis scale (246) . It should be noted that outer y- axis scale (242) for individual port temperature and inner y-axis scale (246) for port minus bank average temperature represent the same absolute range in oc, i.e., a 250oc difference between the bottom and top end of the scales.

The trend plots in Figure 5 show that trending the port temperature versus bank average temperature removes nearly all of the variation present in the single port temperature trend (240) above it. In the window of time illustrated in Figure 5 no problems were experienced with port No. 14, and the resulting trend plot (244) is a relatively smooth, uniform line of trend points. However, if the fuel injector for port No. 14 had become plugged, for example, the problem could have been more readily detected from the port minus bank average trend (244) than from the single port temperature trend (240) ,* diagnostically-significant problem variations in the temperature of cylinder port No. 14 are more easily distinguished from the smooth, uniform port minus bank average trend with its built-in reference standard of normal operation. Referring now to Figure 6, a method for trending the port minus bank average measurements is illustrated. It will be understood that, like the method of Figure 3, the method illustrated in Figure 6 lends itself to other machine component trending applications, and is not limited to cylinder port or exhaust temperature measurement.

For purpose of illustration the engine is assumed to be running at a fixed RPM except at start¬ up/shut down. Accordingly, the trending definition or

trap window is satisfied when engine speed is greater than the minimum predefined engine speed, or

Engine Speed > MIN_CHECK_ENGINE_SPEED At the initial power-up of the system in Figure 4, sensor data is collected at step 254 for each of the sixteen cylinder ports. In the illustrated port temperature embodiment, the collected data is the temperature of each cylinder port. At step 254 the collected port temperatures are assigned to corresponding system values (T _x , T ₂ , ... T ₁₆ ) by the PC- based trend collection software in PC 228.

At step 256 the average cylinder port temperature for each bank of cylinders is determined. In the illustrated embodiment the average temperature for the bank of odd-numbered cylinders is the sum of (T _lf T ₃ , ... T ₁₅ ) divided by eight, while the average for the bank of even-numbered cylinders is the sume of (T ₂ , T ₄ , ... T ₁₆ ) divided by eight.

At step 258 the port minus bank average value T _{portsplit (j)} for each cylinder (j) is determined by subtracting the corresponding bank average T _avg from the individual port temperature T _j .

At step 260 the system determines whether the trap window of machine operating parameters (engine RPM) has been satisfied. This can be determined with the method illustrated in Figure 3A, using the simplified trap

Engine Speed > MIN_CHECK_ENGINE_SPEED at step 108a. If the trap window conditions are met at step 260, the system proceeds to step 262 where the port minus bank average value T _{portsplit (j)} for the current system loop is added to any previous sum _portsp ii _tsum ( _j ) ° ^f P°rt minus bank average readings for each cylinder (j) . A trend count for each cylinder (j) is also incremented at step 262.

After step 262, or if the trap window conditions are not met at step 260, the system proceeds to step 264 to determine whether the current trend period, for example one hour, has expired. If no, the system returns to step 264 and the loop is repeated to generate the next set of data points for the cylinders in the system. If the trend period is determined to have expired at step 264, the system proceeds first to step 266 where the cylinder value (j) is reset to 1, and then to step 268 where the counter for cylinder (j) is compared to a trend count predetermined by the user to be the minimum necessary to have determined an accurate average port minus bank average value T _{portsplit avg (j)} . If the minimum trend count is not met at step 268, the system proceeds to step

270 where T _{portsplit avg (j)} is tagged invalid and ignored by the system. If the minimum trend count is satisfied at step 268, the system proceeds to step 272 where an average value for the port minus bank average difference is determined by dividing the sum of the individual port minus bank average data points by the associated count.

The system then proceeds to step 274 where a signal representing the average value for the port minus bank average for cylinder (j) is generated.

This signal is stored and translated by the associated computer apparatus (230) and/or (228) (Fig. 4) into a plotted trend point on a graph display as shown in Figure 5. At step 276 the system checks that trend value against predefined high and low limits. If the trend point is determined to be above or below the high or low limits, a warning signal is generated and delivered at step 278. If not, the system proceeds to step 280 where the cylinder being monitored is incremented to the next cylinder. If at step 282 the

incremented cylinder value is greater than the total number of cylinders, here sixteen cylinders, the system returns to step 252 where the trend accumulators are reset to zero to begin the next trend period. If at step 282 the current cylinder number is not greater than the total number of cylinders, the system returns to step 268 to trend the next cylinder in the system.

Industrial Applicability

It will be apparent from the foregoing that the method of trending parallel components can be applied to many different types of parallel loaded components operating in a machine system. By trending historical operating parameter data for components loaded in parallel as illustrated above, variations in component performance indicating possible degradation failure can be more accurately diagnosed than by looking at individual component trends alone. This increases the likelihood of early problem diagnosis, such that repairs can be made before a component fails or has a significant negative effect on machine performance. It also facilitates the convenient scheduling of component replacement or repair around machine operation.

The choice of parameter(s) measured, trap window conditions and definitions, the length of the trend period, the number of components measured, averaged and compared, the types of sensors, and the associated apparatus for receiving the component data and trending that information can vary. The foregoing illustrative embodiments of methods and apparatus for carrying out the present invention are therefore not to be construed as limiting the invention beyond the scope of the appended claims.