The invention relates to an electrical actuator comprising a first actuator body, a second acmator body which is pivotable with respect to the first acmator body through an angle of rotation about an axis of rotation, electrical energizing means for exerting an electromagnetic torque on the second actuator body, and an electrical control unit for controlling the angle of rotation of the second acmator body, said control unit comprising an electrical input for receiving an electrical signal corresponding to a required angle of rotation of the second actuator body and an electrical output for supplying an electrical signal corresponding to a required electrical current through the energizing means. The invention further relates to a throttle device for use in an air inlet of an internal-combustion engine, which throttle device comprises a throttle-valve housing, an air passage which is connectable to the air inlet, a throttle-valve which is journalled in the throttle-valve housing so as to be pivotable in the air passage, and an electrical acmator for pivoting the throttle-valve.

An electrical acmator of the kind mentioned in the opening paragraph is disclosed in chapter 15.2 entitled "Linear Position Control" of the book "Control of Electrical Drives" by W. Leonhard, ISBN 3-540-13650-9 Springer-Verlag Berlin Heidelberg New York Tokyo. The control unit of the known acmator has a so-called cascade control strucmre comprising a system of several superimposed control loops for the electromagnetic torque, the angular acceleration, the speed of rotation and the angle of rotation of the actuator. The control loop for the electromagnetic torque comprises a control member with an electrical input for receiving an electrical signal corresponding to a required angular acceleration and an electrical output for supplying the electrical signal corresponding to the required electrical current through the energizing means. The cascade control strucmre provides a namral control sequence which corresponds to the strucmre and operation of the acmator. As a result, the control unit has a transparent structure and can be designed and optimized step by step. A drawback of the known actuator is that the response of the control unit of the known actuator to alterations of the signal corresponding to the required angle of

rotation is relatively slow if the dependence of the electromagnetic, magnetostatic or mechanical characteristics of the acmator on the angle of rotation of the second acmator or on the current through the energizing means is strongly non-linear. Due to the non-linearity of said characteristics, the number of iterative calculations which have to be made by the control unit before a required angle of rotation is achieved is relatively high.

It is an object of the invention to provide an electrical acmator of the kind mentioned in the opening paragraph which has a control unit with a cascade control strucmre, wherein the response time of the control unit to alterations of the signal corresponding to the required angle of rotation is improved.

According to the invention, the electrical actuator is characterized in that the control unit comprises a first control member with an electrical input for receiving the signal corresponding to the required angle of rotation and an electrical output for supplying an electrical signal corresponding to a required electromagnetic torque on the second acmator body, and a second control member with an electrical input for receiving the signal corresponding to the required electromagnetic torque and an electrical output for supplying the signal corresponding to the required current. Owing to the use of said first and second control members, the control unit comprises a refined control sequence wherein the signal corresponding to the required angle of rotation is first converted into a signal corresponding to the required electromagnetic torque by the first control member and subsequently the signal corresponding to the required electromagnetic torque is converted into the signal corresponding to the required current through the energizing means by the second control member. In this way, the first control member allows for a specific calculation of the required electromagnetic torque, taking into account the mechanical and magnetostatic properties of the acmator, while the second control member allows for a specific calculation of the required current taking into account the electromagnetic properties of the acmator. Since the mechanical and magnetostatic properties of the acmator on the one hand and the electromagnetic properties of the acmator on the other hand are taken into account separately, knowledge about these properties of the acmator is taken into account in a relatively specific and detailed manner, so that the calculations of the first and second control members are relatively accurate and the co-operation between the first and second control members is very effective. In this way the number of iterative calculations which have to be made by the control members before a required angle of rotation is achieved is limited. A particular embodiment of an electrical acmator according to the invention is characterized in that the first control member comprises an electrical adder with

an electrical output for supplying the signal corresponding to the required electromagnetic torque, the adder comprising a first electrical input for receiving a feed-forward control signal determined by the signal corresponding to the required angle of rotation, and a second electrical input for receiving a feedback control signal determined by the signal corresponding to the required angle of rotation and by an electrical signal which is supplied by an angle-of-rotation sensor and which corresponds to a measured angle of rotation of the second acmator body. By adding said feed-forward and feedback control signals, a fast and accurate calculation and control of the required electromagnetic torque is achieved.

A further embodiment of an electrical actuator according to the invention is characterized in that the first control member comprises a profile generator having a first electrical input for receiving the signal corresponding to the required angle of rotation, a second electrical input for receiving the signal corresponding to the measured angle of rotation, and an electrical output for supplying an electrical reference signal which corresponds to an angle-of-rotation-versus-time profile generated by the profile generator, the feed-forward control signal being proportional to a required angular acceleration of the second acmator body corresponding to the angle-of-rotation-versus-time profile. The angle- of-rotation-versus-time profile generated by the profile generator extends from the measured angle of rotation to the required angle of rotation. In this way, an instantaneous, discontinuous alteration of the signal corresponding to the required angle of rotation is converted by the profile generator into a profile of the reference signal which is feasible in view of the controllability and the dynamical properties of the electrical actuator.

A special embodiment of an electrical acmator according to the invention is characterized in that the first control member comprises a comparator having a first electrical input for receiving the signal corresponding to the measured angle of rotation, a second electrical input for receiving the reference signal, and an electrical output for supplying a differential signal which is proportional to a difference between the signal corresponding to the measured angle of rotation and the reference signal, the first control member further comprising a regulator with an electrical input for receiving the differential signal and an electrical output for supplying the feedback control signal. Said regulator determines the feedback control signal in such a way that said differential signal is equalized to zero, so that the measured angle of rotation accurately changes according to the angle-of- rotation-versus-time profile generated by the profile generator. By making the regulator to control said differential signal instead of a signal which is proportional to a difference between the signal corresponding to the measured angle of rotation and the signal corresponding to the required angle of rotation, it is achieved that so-called wind-up effects and dynamic overshoot of the regulator are avoided.

A particular embodiment of an electrical acmator according to the invention is characterized in that the adder comprises a third electrical input for receiving an electrical signal corresponding to a load torque which is exerted on the second acmator body and which is dependent on the angle of rotation of the second acmator body and substantially independent of the current through the energizing means. The load torque is exerted on the second actuator body, for example, by a mechanical, magnetic or pneumatic device such as a mechanical spring which is deformed when the second acmator body is pivoted, and is used, for example, to restore the second acmator body to a rest position when the current through the energizing means is zero. The addition of the electrical signal corresponding to the load torque to the feed-forward and feedback control signals mentioned above has the result that, the feed-forward control signal need not be calculated so as to include a component of the electromagnetic torque necessary for compensating said load torque. In this way, the response time and accuracy of the control unit are further improved.

A further embodiment of an electrical acmator according to the invention is characterized in that the load torque is a magnetostatic torque which is exerted on the second acmator body by the first actuator body. Said magnetostatic torque is a magnetic torque which is exerted by the first actuator body on the second actuator body independently of the current through the energizing means. The magnetostatic torque is dependent on the angle of rotation of the second acmator body and constitutes a restoring torque urging the second acmator body towards a rest position. Since the magnetostatic torque is determined by the strucmre of the first and second acmator bodies, the signal corresponding to the load torque can be calculated as a function of the angle of rotation of the second acmator body.

A special embodiment of an electrical acmator according to the invention is characterized in that the first control member comprises an electrical memory with an electrical input for receiving the signal corresponding to the measured angle of rotation and an electrical output for supplying the signal corresponding to the load torque, the memory being provided with a tabular relation between the load torque and the angle of rotation. Since the relation between the load torque and the angle of rotation of the second acmator body is stored in said memory of the first control member in a tabular form, the value of the load torque is relatively accurate and is read out in a simple way without substantial delay. In this way, the response time and accuracy of the control unit are further improved.

A further embodiment of an electrical acmator according to the invention is characterized in that the first control member comprises a disturbance observer for calculating a loading torque exerted on the second acmator body on the basis of a mathematical model of the electrical actuator, the adder comprising a third electrical input for receiving an electrical output signal of the disturbance observer corresponding to a value

of the loading torque calculated by the dismrbance observer. During operation, the angle of rotation of the second acmator body is influenced or disturbed by internal disturbing loading torques such as friction and stiction torques of the bearings of the electrical acmator and a magnetostatic torque exerted on the second acmator body by the first acmator body, and by external disturbing loading torques exerted on the second acmator body. Such disturbing loading torques are not directly measurable or measurable with great difficulty only. The dismrbance observer calculates the total disturbing loading torque on the basis of a mathematical model of the electrical actuator, i.e. on the basis of a set of first-order differential equations describing the physics of the electrical acmator. The addition of the output signal of the dismrbance observer to the feed-forward and feedback control signals has the result that the feedback control signal need not be calculated by the regulator of the first control member so as to include a component of the required electromagnetic torque necessary for compensating the disturbing loading torque. In this way, the required convergence time of the feedback control loop and, consequently, the response time of the control unit are strongly improved.

A particular embodiment of an electrical acmator according to the invention is characterized in that the first control member comprises an electrical limiter for limiting the signal corresponding to the required electromagnetic torque if said signal exceeds a predetermined limit value. In this way, the value of the required electromagnetic torque to be generated by the energizing means is limited to a value which is feasible in view of the mechanical, electrical and thermal properties of the acmator, so that an electromagnetic overtorque, which could lead to damage or malfunction of the acmator, is avoided.

A further embodiment of the electrical acmator according to the invention is characterized in that the second control member comprises an electrical memory with a first electrical input for receiving the signal corresponding to the required electromagnetic torque, a second electrical input for receiving the signal corresponding to the measured angle of rotation, and an electrical output for supplying the signal corresponding to the required current, the memory being provided with a tabular relation between the electromagnetic torque, the angle of rotation and the current. Since the electromagnetic torque is determined by the strucmre of the first and second acmator bodies and the energizing means, the signal corresponding to the required electromagnetic torque can be calculated as a function of the angle of rotation of the second acmator body and the electrical current through the energizing means. Since the relation between the electromagnetic torque, the angle of rotation of the second actuator body and the current through the energizing means is stored in said memory of the second control member in a tabular form, the value of the current is relatively accurate and is read out in a simple way without substantial delay. In this way, the response

time and accuracy of the control unit are further improved.

A special embodiment of an electrical acmator according to the invention is characterized in that the control unit comprises a comparator having a first electrical input for receiving the signal corresponding to the required current, a second electrical input for receiving an electrical signal which is supplied by an electrical current sensor and which corresponds to a measured current through the energizing means, and an electrical output for supplying a differential signal which is proportional to a difference between the signal corresponding to the required current and the signal corresponding to the measured current, the control unit further comprising a regulator with an electrical input for receiving said differential signal and an electrical output for supplying an electrical signal corresponding to an electrical current supplied to the energizing means. Said comparator, current sensor and regulator belong to a current-control loop of the control unit. Said regulator determines the signal corresponding to the electrical current supplied to the energizing means in such a way, that said differential signal is equalized to zero, so that the measured current through the energizing means accurately equals the required current determined by the control unit.

A further embodiment of an electrical actuator according to the invention is characterized in that the dismrbance observer has an electrical input for receiving the signal corresponding to the measured current through the energizing means, the dismrbance observer calculating the angle of rotation, an angular velocity of the second acmator body, and the loading torque on the basis of three state equations for the electrical acmator. The signal conesponding to the measured cuπent is supplied by the cunent sensor which is used in the cunent-control loop of the control unit. Since the control unit comprises a cunent- control loop, the value of the cunent through the energizing means of the electrical acmator is imposed by the cunent-control loop and not by an electrical voltage imposed on the energizing means. In this manner, the value of the cunent through the energizing means is prescribed by the cunent-control loop, so that the mathematical model of the acmator underlying the dismrbance observer can dispense with a usual differential equation for the cunent as a function of an imposed voltage. Since for these reasons the mathematical model comprises only three state equations, the dismrbance observer is relatively simple and suitable for on-line computation.

A still further embodiment of an electrical actuator according to the invention is characterized in that the dismrbance observer comprises a further electrical input for receiving the signal conesponding to the measured angle of rotation, a comparator for determining a deviation between the measured angle of rotation and the calculated angle of rotation, and an adder for conecting the calculated angle of rotation, the calculated angular velocity, and the calculated loading torque by a value proportional to said deviation. In this

embodiment, inaccuracies of the values of the angle of rotation, the angular velocity, and the loading torque calculated by the dismrbance observer and caused by inaccuracies of the mathematical model underlying the dismrbance observer, are conected by a feedback loop. The conected angle of rotation is the sum of the calculated angle of rotation and the product of said deviation and a first weighing factor, the conected angular velocity is the sum of the calculated angular velocity and the product of said deviation and a second weighing factor, and the conected loading torque is the sum of the calculated loading torque and the product of said deviation and a third weighing factor, the first, second and third weighing factors being determined by means of a so-called pole-placement method. A particular embodiment of an electrical acmator according to the invention is characterized in that the control unit comprises an electrical limiter for limiting the signal conesponding to the cunent if said signal exceeds a predetermined limit value. In this way, the value of the electrical cunent through the energizing means is limited to a value which is feasible in view of the thermal properties of the energizing means, so that an overcunent, which could lead to overheating of the energizing means and the acmator, is avoided.

A throttle device of the kind mentioned in the opening paragraph is characterized in that the electrical actuator applied therein is an electrical acmator according to the invention. The throttle device is used in an air inlet of an internal-combustion engine of a vehicle and is adjustable, for example, by means of an accelerator pedal. The accelerator pedal is not mechanically coupled to the throttle- valve of the throttle device, but the electrical acmator is provided with an electrical input for receiving an electrical signal conesponding to a required angle of rotation of the throttle-valve in the air passage of the throttle device, said electrical signal being supplied, for example, by an electronic motor- management system which also controls the fuel-injection and ignition systems of the internal-combustion engine. The angle of rotation of the throttle- valve in the air passage of the throttle device is adjusted by the motor-management system not only as a function of the accelerator-pedal position, but also as a function of, for example, the r.p.m. of the engine, the inlet-air pressure and temperamre, and the engine temperamre. In this way, ti e performance, the fuel consumption and the composition of the exhaust-gases of the internal- combustion engine are improved. Since the throttle-valve of the throttle device is actuated by an electrical acmator according to the invention, the angle of rotation of the throttle-valve required by the motor-management system is achieved in a very accurate manner and the response time which is necessary for carrying out alterations of the required angle of rotation is strongly limited.

The invention will be explained in more detail below with reference to the drawing, in which

Figure 1 diagrammatically shows a throttle device according to the invention, used in an air intake of an internal-combustion engine, Figure 2a is a cross-section of an electrical acmator according to the invention, applied in the throttle device of Fig. 1, in a non-energized condition,

Figure 2b shows the electrical acmator of Fig. 2a in an energized condition,

Figure 3 schematically shows a control unit of the electrical acmator of Fig. 2a,

Figure 4a shows an angle-of-rotation-versus-time profile of the electrical actuator required by a motor-management system of the engine,

Figure 4b shows an angle-of-rotation-versus-time profile generated by a profile generator of the control unit of Fig. 3, Figure 4c shows an angular-acceleration-versus-time profile conesponding to the angle-of-rotation-versus-time profile of Fig. 4b,

Figure 5 schematically shows an alternative control unit of the electrical acmator of figure 2a, and

Figure 6 schematically shows a dismrbance observer of the alternative control unit of figure 5.

The throttle device shown in Fig. 1 comprises a throttle-valve housing 1 with a mbular air passage 3 and a flange 5 by means of which the throttle device can be connected to an air inlet or manifold of an internal-combustion engine not shown in the drawing. The throttle device further comprises a disc-shaped throttle valve 7 which is mounted on a shaft 9 extending diametrically through the air passage 3. The shaft 9 is pivotably journalled in the flange 5 of the throttle- valve housing 1, so that the throttle valve 7 is pivotable in the air passage 3. When the throttle valve 7 is pivoted, the aperture of the air passage 3 and the air flow to the combustion chambers of the internal-combustion engine are altered.

The throttle valve 7 is pivotable in the air passage 3 by means of an electrical actuator 11 comprising a first acmator body 13 which is mounted in an acmator housing 15 of the throttle-valve housing 1 and a second actuator body 17 which is mounted on the shaft 9. As Figs. 2a and 2b show, the second actuator body 17 comprises a cylindrical permanent-magnet rotor body 19 which is diametrically magnetized and has a north pole N

and a south pole S. The first acmator body 13 comprises a U-shaped stator body 21 made of a material having a high magnetic permeability, such as sintered iron, or of magnetic-steel laminations. The U-shaped stator body 21 comprises two limbs 23, 25 which are interconnected by a base 27. The electrical acmator 11 further comprises an energizing means 29 having an electrical coil 31 which is supported by the base 27. The limbs 23, 25 of the stator body 21 are each provided with a pole shoe 33, 35, while the pole shoes 33, 35 each have a curved surface 37, 39. As Figs. 2a and 2b show, the curved surfaces 37, 39 of the pole shoes 33, 35 surround the permanent-magnet rotor body 19, the surface 37 defining an air gap 41 between the rotor body 19 and the pole shoe 33 and the surface 39 defining an air gap 43 between the rotor body 19 and the pole shoe 35. Furthermore, a first gap 45 and a second gap 47 are present between the pole shoes 33, 35, while a first slot 49 is centrally provided in the surface 37 of the pole shoe 33, and a second slot 51 is centrally provided in the surface 39 of the pole shoe 35. In this way, the surface 37 is divided into a first surface portion 53 and a second surface portion 55, and the surface 39 is divided into a first surface portion 57 and a second surface portion 59, while the air gap 41 is divided into a first air- gap portion 61 and a second air-gap portion 63, and the air gap 43 is divided into a first air- gap portion 65 and a second air-gap portion 67. As Figs. 2a and 2b show, the width of the diametrically opposed air-gap portions 61, 67 is smaller than the width of the diametrically opposed air-gap portions 63, 65. Since the width of the air-gap portions 61, 67 is smaller than the width of the air-gap portions 63, 65, a magnetostatic torque T _MS is exerted by the first acmator body 13 on the second acmator body 17, urging the second actuator body 17 into a rest position shown in Fig. 2a when the electrical coil 31 is not energized. To increase the magnetostatic torque T _MS , permanent auxiliary magnets 69, which are indicated in Figs. 2a and 2b with broken lines, may alternatively be mounted in the first surface portion 53 of the pole shoe 33 and in the second surface portion 59 of the pole shoe 35. When the electrical coil 31 is energized, an electromagnetic torque T _EM is exerted on the second acmator body 17, and the second acmator body 17 is pivoted from the rest position shown in Fig. 2a towards a position shown in Fig. 2b which is characterized by an angle of rotation of the second acmator body 17 relative to the rest position. Leaving the external forces on the throttle-valve 7 out of consideration, the electromagnetic torque T _EM equals the magnetostatic torque T _MS in the position shown in Fig. 2b. When the cunent through the coil 31 is switched off, the second acmator body 17 and the throttle valve 7 will remm to their rest position again under the influence of the magnetostatic torque T _MS . The value of the angle of rotation φ in the position shown in Fig. 2b is determined by the value of the electrical cunent through the electrical coil 31 and is adjustable by adjusting the cunent through the

coil 31 in a manner which is described below.

It is noted that the rest position of the electrical acmator 11 shown in Fig. 2a does not conespond exactly to the position occupied by the second acmator body 17 and the throttle valve 7 when the electrical coil 31 is not energized. As Fig. 1 shows, the throttle device also comprises a mechanical stop 71, and the second acmator body 17 comprises a cam 73 which rests against the stop 71 when the coil 31 is not energized. The position of the second actuator body 17 in which the cam 73 rests against the stop 71 differs slightly from the position of the second acmator body 17 shown in figure 2a, so that the cam 73 rests against the stop 71 under the influence of a magnetostatic torque T _{MS 0} . As Fig. 1 shows, this position conesponds to a so-called limp-home position of the throttle valve 7 in the air passage 3 which differs slightly from a so-called idling position of the throttle valve 7 in which the aperture of the air passage 3 is minimal. In the limp-home position of the throttle valve 7, which occurs, for example, when the electrical-energy supply of the throttle device fails, the aperture of the air passage 3 allows for a small air flow towards the combustion chambers of the internal-combustion engine, so that an emergency operation of the engine is still possible. The stop 71 is mechanically adjustable, so that the air flow through the air passage 3 in the limp-home position of the throttle valve 7 is adjustable. In all other positions of the throttle valve 7, including the idling and full-throttle positions, in which the aperture of the air passage 3 is minimal and maximal respectively, an electrical cunent is supplied through the coil 31.

As Fig. 1 shows, the electrical acmator 11 further comprises an electrical control unit 75 by means of which the angle of rotation φ of the throttle valve 7 is controlled. The control unit 75 is diagrammatically shown in Fig. 3 and comprises an electrical input 77 for receiving an electrical signal u _φ which conesponds to a required angle of rotation φ of the second acmator body 17 and the throttle valve 7, and an electrical output 79 for supplying an electrical signal U which determines an electrical cunent through the energizing means 29 of the acmator 11. The signal u _φ is supplied by an electronic motor- management system of the internal-combustion engine, which system is not shown in the drawing. The motor-management system determines the value of the signal _φ not only as a function of the position of an accelerator pedal operated by a driver, but also as a function of other parameters such as, for example, the r.p.m. of the engine, the pressure and temperamre of the inlet-air, and the engine temperamre. Furthermore, the motor-management system controls the idling speed of the engine during and after a cold start of the engine, so that usual air-bypass systems are not necessary. The motor-management system also controls the fuel-injection and ignition devices of the engine. In this way, the operation of the fuel- injection, ignition and throttle devices of the engine are attuned to each other, so that the

performance, the fuel consumption and the composition of the exhaust-gases of the engine are improved.

As Fig. 3 further shows, the control unit 75 comprises a first control member 81 and a second control member 83. The first control member 81 comprises the electrical input 77 of the control unit 75 and an electrical output 85 for supplying an electrical signal u _EM which conesponds to a required electromagnetic torque T _EM to be exerted on the second acmator body 17. The second control member 83 comprises an electrical input 87 for receiving the signal u _EM from the first control member 81 and an electrical output 88 for supplying an electrical signal u, which conesponds to a required electrical cunent through the energizing means 29.

As Fig. 3 shows, the first control member 81 comprises a profile generator 89 with a first electrical input 91 for receiving the signal u _ψ and a second electrical input 93 for receiving an electrical signal u,^ which conesponds to a measured angle of rotation of the second actuator body 17 and the throttle valve 7. The signal u^ is supplied by an angle-of-rotation sensor 95 of the throttle device via a usual high-frequency filter 97. As Fig. 1 shows, the angle-of-rotation sensor 95 is mounted on the throttle-valve housing 1 near an end of the shaft 9 which is remote from the electrical actuator 11. The profile generator 89 generates an angle-of-rotation-versus-time profile which extends from a measured acmal angle of rotation φ _M to the required angle of rotation φ _R . Fig. 4a shows an example of an angle-of-rotation-versus-time profile required by the motor-management system where the required angle of rotation discontinuously alters from φ _M to φ _R at a point in time t„. Such a profile cannot be realized by the electrical actuator 11 because the necessary electromagnetic torque is infinitely high. Fig. 4b shows an angle-of-rotation-versus-time profile generated by the profile generator 89 where the angle of rotation smoothly runs from φ _M to φ _R between the points in time to and t,. Fig. 4c shows an angular-acceleration- versus-time profile which conesponds to the angle-of-rotation-versus-time profile of Fig. 4b. The profile generator 89 comprises a first electrical output 99 for supplying a feed-forward control signal u _FF which is the product of an angular acceleration required according to the angular-acceleration-versus- time profile and a moment of inertia of the pivotable parts of the throttle device. The signal Upp therefore conesponds to an electromagnetic-torque component necessary for realizing said angular acceleration. The profile generator 89 further comprises a second electrical output 101 for supplying an electrical reference signal u _ΨR which conesponds to the angle-of- rotation-versus-time profile generated by the profile generator 89. In this way, an instantaneous, discontinuous alteration of the signal which is supplied by the motor- management system, is convertible by the profile generator 89 into profiles of the feed¬ forward control signal Upp and the reference signal u _φR which are feasible not only in view of

the dynamic properties of the electrical actuator 11 but also in view of the controllability of the acmator 11.

As Fig. 3 further shows, the first control member 81 comprises a comparator 103 with a first electrical input 105 for receiving the signal u^ and a second electrical input 107 for receiving the reference signal u _φR . The comparator 103 comprises an electrical output 109 which supplies a differential signal u _w which is proportional to a difference between the signals u^ and u _φK . The differential signal u _D is supplied to an electrical input 111 of a PID-regulator 113 which fiirther comprises an electrical output 115 for supplying a feedback control signal Upn- The feed-forward control signal Upp and the feedback control signal u _ra are supplied to a first electrical input 117 and to a second electrical input 119, respectively, of an electrical adder 121 of the first control member 81. As Fig. 3 shows, the adder 121 further comprises a third electrical input 123 for receiving an electrical signal u _MS which conesponds to an estimated magnetostatic torque T _MS exerted by the first acmator body 13 on the second acmator body 17. The value of the magnetostatic torque T _MS , which urges the second actuator body 17 and the throttle- valve 7 towards the limp-home position as discussed above, depends on the angle of rotation φ and is substantially independent of the cuπent through the energizing means 31. The relation between T _MS and φ is determined by the strucmre and composition of the first and second actuator bodies 13, 17. Said relation is calculated or measured and is stored in a tabular form in an electrical memory 125 of the first control member 81, said memory comprising an electrical input 127 for receiving the signal u _w and an electrical output 129 for supplying the signal u _MS to the adder 121. By storing the relation between the magnetostatic torque and the angle of rotation in a tabular form in said memory 125, it is achieved that the value of the magnetostatic torque is read out in an accurate and relatively simple manner without substantial delay. In this way, the supply of the signal u _MS to the adder 121 does not increase the response time of the control unit 75.

The adder 121 comprises an electrical output 131 for supplying the electrical signal u _EM conesponding to the required electromagnetic torque to be exerted on the second acmator body 17. The signal u _EM is the mathematical sum of the signals Upp, u _ra and u _MS . In this way, the required electromagnetic torque T _EM is the sum of the electromagnetic-torque component which is necessary for realizing the required angular acceleration of the throttle valve 7, the estimated magnetostatic torque T _MS , and a feedback electromagnetic-torque component represented by the signal Upr,. The PID-regulator 113 determines the signals u _FB and u _EM in such a manner that the differential signal u _Dφ is equalized to zero, so that the measured angle of rotation of the throttle valve 7 changes accurately in accordance with the angle-of-rotation-versus-time profile generated by the

profile generator 89. Since the comparator 103 does not, as is usual, determine a difference between the signals u _φφ and u _φ , but determines the difference between the signals u _φ and u„ _R , the control of the signals u _FB and u _EM by the PID-regulator 113 is very stable, so that usual wind-up effects and dynamic overshoot of the PID-regulator 113 do not occur. Furthermore, the control of the signals u _ra and u _EM by the PID-regulator 113 is very fast as a result of the use of the adder 121. Since the signals Upp and u _MS are added to the signal u _FB , the PID- regulator 113 need not calculate the electromagnetic-torque component necessary for realizing the required angular acceleration of the throttle valve 7 and the electromagnetic- torque component necessary for compensating the magnetostatic torque T _MS . The calculation of these electromagnetic-torque components by a PID-controller in a feedback control loop would demand several controller sampling times, the more so as the relation between the magnetostatic torque T _MS and the angle of rotation φ is strongly non-linear, so that the response time of the control unit 75 would deteriorate and the chance of instabilities of the PID-controller would increase. With the adder 121, the PID-regulator 113 need only calculate a number of electromagnetic-torque components which are small relative to the electromagnetic-torque components mentioned before, such as a component compensating for air-flow forces and a component compensating for mechanical-friction forces. In this way, the response time and accuracy of the control unit 75 are improved.

As Fig. 3 shows, the first control member 81 further comprises an electrical limiter 133 for limiting the signal u _EM when the signal \X__ _M exceeds a predetermined limit value. Said limit value of the signal u _EM is determined in such a way that the electromagnetic torque exerted on the second acmator body 17 and on the throttle valve 7 never exceeds a predetermined maximum torque value. In this way, mechanical damage or malfunctioning of the electrical acmator 11 as well as overheating of the energizing means 29 are avoided. When the signal u _EM supplied by the adder 121 exceeds said predetermined limit value, the value of the signal u^ is adjusted to said limit value by the limiter 133.

As Fig. 3 further shows, the second control member 83 of the control unit 75 comprises an electrical memory 135 with a first electrical input 137 for receiving the signal u _EM from the input 87 of the second control member 83, a second electrical input 139 for receiving the signal u _w from the angle-of-rotation sensor 95, and an electrical output 141 for supplying the electrical signal U _j which conesponds to an electrical cunent through the electrical coil 31 of the energizing means 29 necessary for achieving the required electromagnetic torque T _EM . The value of the electromagnetic torque T _EM is dependent on the angle of rotation φ of the second acmator body 17 and on the value of the electrical cunent through the coil 31. The relation between the electromagnetic torque T _BM , the angle of rotation φ and the cunent through the coil 31 depends on the strucmre and composition of

the first and second acmator bodies 13, 17 and the energizing means 29. Said relation is calculated or measured and is stored in a tabular form in the memory 135. In this way, the value of the cunent necessary for achieving a required electromagnetic torque at the measured angle of rotation is read out from the memory 135 in an accurate and simple manner without substantial delay. It is noted that a calculation of the required cunent by a usual calculator would demand a substantial amount of time, the more so as the relation between the electromagnetic torque, the angle of rotation and the cunent is strongly non¬ linear. With the use of memory 135, the short response time of the control unit 75 obtained by the PID-regulator 113 in combination with the adder 121 is not deteriorated by the second control member 83.

The control unit 75 further comprises a comparator 143 having a first electrical input 145 for receiving the signal u, from the output 88 of the second control member 83, a second electrical input 147 for receiving an electrical signal u _π which conesponds to a measured electrical cunent through the energizing means 29, and an electrical output 149 for supplying a differential signal u _DI which is proportional to a difference between the signals u, and u _u . The signal u _π is supplied by an electrical cunent sensor 151 via a usual high-frequency filter 153. The cunent sensor 151 measures the electrical cunent which is supplied to the energizing means 29 by a power end stage 155 of the electrical acmator 11. In Fig. 3, the cunent sensor 151 and the power end stage 155 are shown diagrammatically only. Furthermore, the control unit 75 comprises a Pi-regulator 157 with an electrical input 159 for receiving the differential signal u _D ι and an electrical output 161 for supplying an electrical signal u'j which conesponds to the electrical cunent to be supplied to the energizing means 29 by the power end stage 155. The Pi-regulator 157 determines the signal u', in such a way that the differential signal u _D1 is equalized to zero, so that the measured cunent supplied by the power end stage 155 to the energizing means 29 equals the required cunent deteπnined by the second control member 83.

As Fig. 3 further shows, the power end stage 155 of the electrical acmator 11 is fed by a constant electrical voltage of, for example, a battery. The power end stage 155 comprises four NPN-transistors, i.e. two upper transistors 163, 165 and two lower transistors 167, 169, and two electrical inverters 171, 173. The transistors 163, 165, 167, 169 and the invertors 171, 173 are interconnected in a usual bridge configuration. The transistors 163, 165, 167, 169 are driven in a usual manner by a pulse width modulator 175 of the control unit 75, which comprises a first electrical input 177 for receiving the signal u', supplied by the Pi-regulator 157 and a second electrical input 179 for receiving the signal u _π supplied by the cunent sensor 151. A first electrical output 181 of the pulse width modulator 175 is connected to the base of lower transistor 167 and via the invertor 171 to the base of

upper transistor 163, while a second electrical output 183 of the pulse width modulator 175 is connected to the base of lower transistor 169 and via the invertor 173 to the base of upper transistor 165. The signal u', is converted by the pulse width modulator 175 into mutually complementary pulsatory drive signals Uc and -Uc at the first and second electrical outputs 181, 183, respectively, of the pulse width modulator 175. In dependence on the polarity of the drive signals Uc and -Uc, the lower transistor 167 and the upper transistor 165 are opened whereby an electrical cunent in the energizing means 29 is admitted in one direction, or the lower transistor 169 and the upper transistor 163 are opened whereby an electrical cunent in the energizing means 29 is admitted in the opposite direction. The pulse width modulator 175 further comprises an electrical limiter for limiting the pulse width of the drive signals u _c and -u _c when the signal u _π supplied by the cunent sensor 151 exceeds a predetermined limit value. In this way, the pulse width of the electrical cunent through the coil 31 is limited to a value which is feasible in view of the thermal properties of the energizing means 29. An overcunent in the coil 31, which could lead to overheating of the energizing means 29 and the electrical acmator 11, is avoided in this way.

The control unit 75 described before has a so-called cascade control strucmre according to which the signal conesponding to a required angle of rotation is first converted into a signal conesponding to a required angular acceleration, the signal conesponding to the required angular acceleration being subsequently converted into a signal U _EM coπesponding to a required electromagnetic torque, and the signal u _EM coπesponding to the required electromagnetic torque being finally converted into a signal u, coπesponding to a required electrical cuπent through the energizing means 29. As described above, this refined cascade control structure with the first and second control members 81, 83 allows for a specific calculation of the required electromagnetic torque T _EM , taking into account the mechanical and magnetostatic properties of the electrical acmator 11, and for a specific calculation of the required cuπent, taking into account the electromagnetic properties of the acmator 11. This refined cascade control strucmre leads to a response time of the control unit 75 which is short relative to common and usual control strucmres according to which the required cuπent is calculated in an iterative manner by a feedback control loop without or with fewer intermediate control steps. Said common and usual control strucmres would require a high number of iterative calculations and therefore would lead to a long response time, particularly because the relation between the required cunent and the angle of rotation is strongly non-linear.

In the control unit 75 described before, the third input 123 of the adder 121 receives an electrical signal u _MS conesponding to an estimated magnetostatic torque T _MS exerted by the first actuator body 13 on the second actuator body 17. The magnetostatic

torque T _MS is an internal loading torque influencing or disturbing the angle of rotation of the second acmator body 17 and the throttle valve 7. The angle of rotation of the second acmator body 17 is also dismrbed by other internal disturbing loading torques such as friction and stiction torques of the bearings of the electrical acmator 11. The angle of rotation of the second actuator body 17 is also dismrbed by external disturbing loading torques exerted on the second acmator body 17 and the throttle valve 7 such as a torque caused by air-flow forces exerted on the throttle valve 7 by the air flowing through the air passage 3. Figure 5 shows an alternative control unit 185 of the electrical acmator 11 in which the electrical memory 125 of the control unit 75 is replaced by a so-called dismrbance observer 187 for calculating the total disturbing loading torque exerted on the second actuator body 17 and the throttle valve 7 on the basis of a mathematical model of the throttle device and the electrical acmator 11. The dismrbance observer 187, which will be described in more detail hereinafter, comprises an electrical output 189 for supplying an electrical signal Uc _LT conesponding to a value of the loading torque calculated by the dismrbance observer 187. Said signal Uc _LT is supplied to the third input 123 of the adder 121. By the use of the dismrbance observer 187, a direct measurement of the disturbing loading torque, which is very difficult or even impossible, is avoided. Furthermore, the PID-regulator 113 need not calculate the electromagnetic-torque component necessary for compensating the total disturbmg loading torque exerted on the second acmator body 17 and the throttle valve 7. With the dismrbance observer 187, the PID-regulator 113 need only calculate a relatively small deviation between the calculated loading torque and a loading torque acmally influencing the throttle valve 7 and the second acmator body 17. In this way, the response time and accuracy of the control unit 185 are further improved.

As mentioned before, the dismrbance observer 187 is used for calculating the loading torque exerted on the second acmator body 17 and the throttle valve 7 on the basis of a mathematical model of the throttle device and the electrical acmator 11, so that a difficult and unreliable measurement of the loading torque is avoided. The mathematical model underlying the dismrbance observer 187 is based upon a set of three first-order differential equations which read as follows:

J.dω/dt = k(φ).I _ACT - T _L0AD [1] ω = dφ/dt [2] dT _L0AD dt = 0 [3]

Equation [1] is an equation of motion of the throttle valve 7 and the second acmator body 17, wherein J is the moment of inertia of the pivotable parts of the throttle device, ω is the

angular velocity of the pivotable parts of the throttle device, k(φ).I _ACT is the electromagnetic torque T _EM exerted on the second acmator body 17, k(φ) being a factor which is dependent on the angle of rotation φ and I _ACT being the cunent through the energizing means 29, and T _OAD i ^{s e} loading torque exerted on the throttle valve 7 and the second acmator body 17. Equation [2] describes the relation between the angular velocity ω and the angle of rotation φ of the throttle valve 7. Equation [3] comprises a simplifying assumption for the loading torque, namely that the loading torque is constant.

Since the value of the cunent I _ACT through the energizing means 29 is determined by the Pl-regulator 157 of the control unit 75 and not by the electrical voltage by which the power end stage 155 of the electrical actuator 11 is fed, the mathematical model underlying the dismrbance observer 187 can dispense with a fourth differential equation describing a relation between the cunent through the energizing means 29 and the voltage imposed on the energizing means 29. As Fig. 5 shows, the dismrbance observer 187 has a first electrical input 191 for receiving the signal u _π supplied by the cunent sensor 151 and conesponding to the measured electrical cunent through the energizing means 29. The disturbance observer 187 calculates the angle of rotation φ, the angular velocity ω, and the loading torque on the basis of the input signal u _π and the three differential equations [1], [2] and [3] mentioned above. Since the mathematical model underlying the disturbance observer 187 comprises only three first-order differential equations, the dismrbance observer 187 is relatively simple and suitable for on-line computation.

In a matrix form, the set of equations [1], [2] and [3] reads as follows:

Φ).I ACT

Furthermore, the dismrbance observer 187 is based upon the following discretizations:

φ _k+1 = φ _k + T.ω _k + T ² /2J.k(φ).I _ACT - T ² /2J.T _LOAD)k ; Α ₊ i = ω _k + T/J.k(φ).I _ACT - T/J.T _{L0AD k} ;

wherein φ _k+ , and ω _k+1 are the values of the angle of rotation and the angular velocity calculated by the dismrbance observer 187 at a point of time k+1, wherein φ _k , ω _k , and T _LOAD, ,. are the values of the angle of rotation, the angular velocity, and the loading torque calculated by the disturbance observer 187 at a point of time k, and wherein T is a time

interval between the points of time k and k+1. With these discretizations, the set of equations [1], [2] and [3] in matrix form reads as follows:

x _k+ ι = Φ.x _k + H.k(φ).I _ACT ;

wherein

l

The vectors x _k and _k+ ι are the state vectors for the points of time k and k+1, the matrix Φ is the system matrix, and the matrix H is the input matrix.

The equations [1], [2] and [3] are implemented in the dismrbance observer 187 in the form of a computer program. Fig. 6 diagrammatically shows the disturbance observer 187 in the form of a number of function blocks representing the computer program. As mentioned before, the dismrbance observer 187 comprises a first electrical input 191 for receiving the signal u _π conesponding to the measured cunent I _ACT through the energizing means 29. Furthermore, the dismrbance observer 125 comprises a second electrical input 193 for receiving the signal u^ conesponding to the measured angle of rotation. The signal u,^ is used by the dismrbance observer 187 in a manner to be described hereinafter. As Fig. 6 further shows, the disturbance observer 187 comprises a first function block 195 for multiplying the value I _ACT by a constant factor K representing an average value of the factor k(φ). Alternatively, the function block 195 may contain a relation between k(φ) and φ, for example, in a tabular form, in which case the function block 195 comprises an input 197 for receiving the input signal u . In Fig. 6, the altemative input 197 is shown with a broken line. The dismrbance observer 187 further comprises a second function block 199 for multiplying the input matrix H by the value K.I _ACT or by the value k(Φ).l _A cτ» a ¹¹ output of the function block 199 representing the vector H.k(φ).I _ACT . The disturbance observer 187 further comprises a third function block 201 for adding up the

vector H.k(φ).I _ACT and a vector X _CORR to be described hereinafter, an output of the third function block 201 representing the new state vector x _k+1 . Furthermore, the dismrbance observer 187 comprises a fourth function block 203 for supplying the component T _{L0AD k+} , of the new state vector x _k+1 to the output 189 of the dismrbance observer 187. Furthermore, the fourth function block 203 leads the state vector x _k+1 to a fifth function block 205 which multiplies the state vector x _k+1 by the system matrix Φ. An output of the fifth function block 205 represents the value Φ.x _k .

As described before, the dismrbance observer 187 calculates the values of the angle of rotation φ, the angular velocity ω, and the loading torque T _L0AD on the basis of the set of equations [1], [2] and [3]. Since the value of the angle of rotation φ is also measured by the angle-of-rotation sensor 95, the measured value of the angle of rotation can be used to conect inaccuracies of the mathematical model underlying the disturbance observer 187 and inaccuracies of the discretizations of the equations [1], [2] and [3]. For this purpose, the dismrbance observer 187 comprises a sixth function block 207 for comparing the measured value of the angle of rotation represented by the input signal u^ and the calculated value φ _k+1 of the angle of rotation which is supplied by the fourth function block 203. An output value Δφ of the sixth function block 207 conesponds to a deviation between said measured angle of rotation and said calculated angle of rotation and is led to a seventh function block 209 which multiplies a conector matrix L by the value Δφ. The conector matrix L comprises a first weighing factor L,, a second weighing factor Lj, and a third weighing factor I^ for respectively conecting the calculated value of the angle of rotation, the calculated value of the angular velocity, and the calculated value of the loading torque, said weighing factors being determined by means of a so-called pole-placement method which is known and usual per se. An output vector L.Δφ of the seventh function block 209 is led to an eighth function block 211 of the disturbance observer 187 which is used for adding up the output vector L.Δφ of the seventh function block 209 and the vector Φ.x _k . In this manner, an output vector X _CORR of the eighth function block 211 reads as follows:

XCORR =- *- ^x k + L.Δφ ;

with

L = L ₂

Therefore, the new state vector x _k+1 reads as follows:

x _k+1 = Φ,x _k + H.k(φ).I _ACT + L.Δφ .

In the electrical acmator 11 described above, the first acmator body 13 exerts a magnetostatic torque on the second acmator body 17, said magnetostatic torque being dependent on the angle of rotation of the second actuator body 17 relative to the first acmator body 13. It is noted that the invention also relates to other types of electrical actuators with a first acmator body, a second actuator body which is pivotable relative to the first acmator body through a limited angle of rotation, energizing means for exerting an electromagnetic torque on the second acmator body, and a control unit for controlling said angle of rotation. The actuator may, for example, be provided with a mechanical torsion spring for exerting a mechanical spring torque on the second acmator body instead of or in addition to the magnetostatic torque. In such a case, the memory 125 of the first control member 81 is omitted or replaced by a memory in which a relation between said mechanical spring torque and the angle of rotation is stored in a tabular form, said memory supplying an electrical signal conesponding to an estimated mechanical spring torque.

It is further noted that the signal u _EM conesponding to the required electromagnetic torque may also be determined in an alternative way by the first control member 81, while the signal u, conesponding to the required electrical cunent through the energizing means 29 may also be determined in an alternative way by the second control member 83. In the first control member 81, for example, the profile generator 89 may be omitted or a comparator may be used having the signals u _φ and Uψψ as input signals. Furthermore, the memory 125 may be replaced by a calculator containing a mathematical relation between the magnetostatic torque and the angle of rotation. Furthermore, depending on the strucmre and composition of the electrical acmator, a different load torque exerted on the second acmator body in dependence of the angle of rotation may be determined by the memory 125 or calculator instead of the magnetostatic torque. Finally, the memory 135 of the second control member 83 may be replaced by a calculator containing a mathematical relation between the electromagnetic torque, the angle of rotation of the second acmator

body, and the cunent through the energizing means.

It is finally noted that the electrical acmator according to the invention may also be applied in other devices in which the angular position of a shaft should be controlled to a constant or variable reference angle. The electrical acmator may, for example, be used in servo-actuated valves in chemical plants and power stations or in devices for deflecting the control surfaces of an aircraft. The acmator may be used as a so-called prime actuator without a transmission, in which case the acmator directly drives a body which is to be displaced, as in the embodiment of the invention described above, or in combination with a transmission for converting a rotational motion into another rotational motion or into a linear motion, in which case the linear position of a body can be accurately controlled by the electrical actuator.