WO2007040404A1

WO2007040404A1 - Method, device and system for controlling an electro-pneumatic actuator

Info

Publication number: WO2007040404A1
Application number: PCT/NO2006/000317
Authority: WO
Inventors: Glenn-Ole Kaasa
Original assignee: Kongsberg Automotive As
Priority date: 2005-10-05
Filing date: 2006-09-15
Publication date: 2007-04-12
Also published as: SE529494C2; SE0502195L

Abstract

The invention relates to a method, a control device and a system for controlling an electro-pneumatic actuator in accordance with a reference input. The method comprises the steps of providing the reference input, providing a measurement signal representative of the actuator position, calculating estimates of unmeasured states in the actuator based on the position measurement and a control signal, calculating the control signal based on the reference input and the estimates of the unmeasured states, and supplying the control signal as an input signal to the electro-pneumatic actuator. The estimates are calculated by means of an observer which comprises a non-linear model of the pressure dynamics of the actuator and of the actuator load characteristics. A reference model is included for providing derivatives of a primary reference. The invention is particularly applicable for controlling a clutch actuator in a vehicle.

Description

METHOD, DEVICE AND SYSTEM FOR CONTROLLING AN ELECTRO- PNEUMATIC ACTUATOR

FIELD OF THE INVENTION

The present invention relates generally to the field of control systems for electro- pneumatic actuators. More specifically, the present invention relates to a method, a device, a computer program and a system for controlling an electro-pneumatic actuator, such as a vehicle clutch actuator.

BACKGROUND OF THE INVENTION

Electro-pneumatic actuators are used in a variety of industrial applications in order to move the position of machine elements.

Formerly, pneumatic actuators were commonly used as on/off actuators because of their robustness, their low cost, and the convenience of feeding them with compressed air. Solutions have also been developed wherein the position of the electro -pneumatic actuator is servo-controlled by means of a position sensor. However difficulties have been encountered due to the non-linear nature of the actuator, in particular caused by the compressibility of air or of the gas contained in the actuator. These difficulties give rise to instability such as oscillations and lack of stiffness in the pneumatic actuator.

One particular application of an electro-pneumatic actuator is for clutch actuation in a vehicle. The clutch in a vehicle is arranged to allow engagement or disengagement of the connection between the vehicle engine and the vehicle transmission. In heavy-duty vehicles, such as trucks, it is necessary that the clutch operation is assisted by an actuator. Since compressed air is already available as an energy supply in such vehicles, in particular for braking purposes, it would be convenient and advantageous to use an electro-pneumatic clutch actuator for assisting the clutch operation.

US-5 424 941 relates to a controller which includes a programmed microprocessor, configured for positioning a pneumatic actuator. In order to improve control performance, a differential pressure measurement signal is used as an additional feedback signal in the control loop. Since position feedback is not sufficient, there is a need for additional sensors in the actuator, resulting in higher costs and an additional source of error. SUMMARY OF THE INVENTION

An object of the present invention is to provide a method, a device, a computer program and a system for controlling an electro-pneumatic actuator in accordance with a reference input. Another object of the invention is to provide such solutions particularly adapted for the non- linear nature of the electro-pneumatic actuator.

Still another object of the invention is to provide such solutions which are based solely on position feedback from the actuator, thus avoiding the need for additional sensors in the actuator, which represent additional costs and possible sources of error.

A further object of the invention is to provide such solutions which result in a robust control of the electro-pneumatic actuator, i.e. that it effectively deals with model uncertainties and attenuates possible disturbances, e.g., due to ageing, wear and temperature variations. At least some of the above objects are achieved by a method, a device, a computer program and a system as set forth in the appended, independent claims.

Further advantageous features are set forth in the dependent claims.

BRIEF DESCRIPTION OF THE DRAWINGS

The present invention will be described in detail below in conjunction with the appended drawings in which:

Fig. 1 is a schematic block diagram illustrating a system which operates in accordance with the present invention,

Fig. 2 is a schematic block diagram illustrating the principles of an electro- pneumatic clutch actuator, Fig. 3 is a schematic block diagram illustrating the structure of a control device in accordance with the present invention,

Fig. 4 is a flow chart schematically illustrating a method in accordance with the present invention,

Fig. 5 is a schematic block diagram illustrating an example implementation of an observer,

Fig. 6 is a schematic block diagram illustrating an example implementation of a controller,

Fig. 7 is a schematic block diagram illustrating an example implementation of a reference model, and Fig. 8 is a graph illustrating an exemplary load characteristic. DETAILED DESCRIPTION OF THE INVENTION

Fig. 1 is a schematic block diagram illustrating a system which operates in accordance with the present invention.

The block diagram in fig. 1 illustrates schematically a control device 100 for controlling an electro-pneumatic actuator in accordance with a reference input. In this example embodiment the actuator is an electro-pneumatic clutch actuator which is arranged to actuate a clutch in the drive train of a motor vehicle, such as a truck.

Although a vehicle clutch is specified herein as an application example, the skilled person will readily realize that the principles of the present invention may be applied also in other systems and applications wherein the position of an electro- pneumatic actuator shall be controlled in accordance with a reference input.

In fig. 1, a motor vehicle (not shown) comprises an engine 110, typically an internal combustion engine such as a diesel engine. The engine 110 provides rotary torque on a rotatable engine shaft 112, which is connected to a friction clutch 120. The friction clutch 120 comprises an axially fixed friction plate 122 and an axially movable friction plate 124. The plates are arranged for being fully engaged, partially engaged or not engaged.

A goal of the control task is to bring the degree of engagement between the clutch plates 122, 124 in accordance with a primary reference input 142, provided by a primary reference device 140. In the example illustrated, the primary reference device 140 is a clutch pedal or a clutch by wire (CBW) device, operated by the driver of the vehicle. Alternatively, the primary reference device 140 may be an automated manual transmission (AMT) device. In either case, the primary reference input signal 142 represents the desired position of the axially movable plate 124 relatively to the axially fixed plate 122 in the clutch 120.

The axially movable friction plate 124 supplies rotary torque to the transmission 130, which is further connected to the drive shaft (not shown) of the vehicle.

The movable friction plate 124 is axially positioned by an electro-pneumatic clutch actuator 180. The clutch actuator 180 is described in closer detail with reference to figure 2.

A signal representative of the axial position of the movable friction plate 124 is measured by a position sensor 190. Since the clutch actuator is directly connected to the movable friction plate, the position sensor 190 may be included as a part of the actuator 180. The position sensor provides the measured position signal 192. The measured position signal 192 is fed into the observer 170. A control signal 162 supplied by a controller 160 is also fed into the observer 170. The observer 170 is arranged for calculating an estimate 172 of at least one unmeasured state in the electro-pneumatic actuator 180, based on the measured position signal 192 and the control signal 162.

The observer 170 comprises a model arranged for simulating the dynamics of the electro-pneumatic actuator. The model is preferably non-linear. Advantageously, the non-linear model comprises a model of the pressure dynamics of the electro-pneumatic clutch actuator.

Advantageously, the non-linear model also comprises a model of actuator load characteristics, in particular a clutch spring load characteristic and a friction load characteristic of the clutch.

The operation of the observer is described in closer detail with reference to fig. 5.

The controller 160 is arranged for calculating the above mention control signal 162 based on the reference input 152 and the estimate 172 of at least one unmeasured state. The operation of the controller 160 is described in closer detail with reference to fig. 6.

The control signal output 162 is supplied as input signal to the electro-pneumatic clutch actuator 180.

The reference input to the controller 160 is advantageously connected to the output of a reference model 150 supplied by the primary reference input 142, i.e. the signal provided by the CBW device (clutch pedal) or the AMT device.

The operation of the reference model 150 is described in closer detail with reference to fig. 7.

Fig. 2 is a schematic block diagram illustrating the principles of an electro- pneumatic clutch actuator.

The actuator 180 is controlled by a closed center three-way proportional valve. The control device 100 provides a control signal output 162 to the proportional valve, based on the reference input 142 and the position signal 192 provided by the position sensor 190. Thus, the position 192 of the actuator 180 is controlled by manipulating the control signal 162 to generate the necessary pressure P_A in the appropriate chamber of the actuator.

In this example, a proportional valve has been illustrated for simplicity. The skilled person will readily realize that on/off valves may be employed as well.

Fig. 3 is a schematic block diagram illustrating the structure of a control device in accordance with the present invention.

The control device is advantageously implemented as a functional part of a device known in the automotive industry as an Electronic Control Unit (ECU). An Electronic Control Unit (ECU) is a centralized digital control unit arranged in a vehicle, which may be used for controlling various functions in the vehicle. For instance, the ECU may control engine functions such as fuel injection, or braking functions, such as the operation of anti-locking breaks. Alternatively or in addition, the ECU may control vehicle transmission functions. For instance, in the case of an automatic transmission, the ECU may set parameters associated with automatic transmission modes.

In the present invention, a special ECU may be provided and configured in accordance with the invention for controlling an electro-pneumatic clutch actuator in the vehicle.

The structural hardware design of the control device 100, which may be a ECU, is illustrated in fig. 3.

The control device 100 comprises an internal bus 310, operatively connected to a processing unit 350, in particular a microprocessor. A memory 320 is operatively connected to the bus 310. The memory 320 comprises a random access memory (RAM) portion 330, for storing temporary data during processing, and a nonvolatile memory portion 340 such as a Flash memory portion, for storing program instructions and fixed data.

The reference input 142 is fed to the I/O adapter 360, which is operatively connected to the bus 310. This enables the processing unit 350 to read the reference input 142.

Likewise, the measured position input 192 is fed to the I/O adapter 360.

The output signal 162 provided by the I/O adapter 360 is operatively connected to the control input of the electro-pneumatic actuator 180. Although the signals are shown as separate signal lines for simplicity of illustration, the skilled person will realize that special digital bus technology commonly used in the automotive industry, such as the CAN bus, may advantageously be used for the communication between the ECU and the external components such as the actuator and the position sensor. In accordance with the present invention, the memory 320, and in particular the non-volatile memory portion 340, comprises processor instructions that causes the processing unit 350 to perform a method according to the present invention, as described in detail with particular reference to fig. 4 below.

Advantageously, a shared ECU is used. In this case, the hardware structure of fig. 3 may represent the overall ECU. An operating system included in the memory 320 is provided for low-level control of the hardware and for enabling higher-level computer program modules or portions held in the memory 320 to implement various control functions related to the operation of the vehicle. The present invention may in this case be put into effect by the described shared hardware structure and a computer program portion that performs the method in accordance with the invention, working in conjunction with the operating system.

Alternatively, the control device may be implemented as a separate unit, e.g. as a separate unit similar to the ECU illustrated in fig. 3.

Fig. 4 is a flow chart schematically illustrating a method in accordance with the present invention.

The purpose of the method is controlling an electro-pneumatic actuator, such as a clutch actuator in a vehicle, in accordance with a reference input. The method starts at the initial step 400.

First, in the reference providing step 410, the reference input is provided. This step advantageously comprises the substeps of

- providing a primary reference input, and

- filtering the primary reference input by a reference model. The primary reference input is advantageously supplied by a CBW signal or an AMT signal.

An exemplary reference model used in the reference providing step 410 is further described with reference to fig. 7 below.

Next, in step 420, a measurement signal representative of the actuator position is provided. Advantageously, this step comprises to read a position sensor in the electro-pneumatic actuator.

Next, in step 430, an estimate of at least one unmeasured state in the electro- pneumatic actuator is calculated based on the measurement signal and a control signal. Advantageously, this step comprises operating a non-linear model which simulates the dynamics of the electro-pneumatic actuator. The non-linear model advantageously comprises a model of the pressure dynamics of the electro- pneumatic actuator.

Advantageously, the non-linear model comprises a model of actuator load characteristics, including a spring load characteristic and a friction load characteristic.

An exemplary observer used in the estimate providing step 430 is further described with reference to fig. 5 below.

Next, in step 440, the control signal is calculated based on the reference input and the estimate of an unmeasured state.

An exemplary controller used in the control signal calculating step is further described with reference to fig. 6 below. Further, in step 450, the calculated control signal is supplied as an input signal to the electro-pneumatic actuator.

Fig. 5 is a schematic block diagram illustrating an example implementation of an observer 170. The purpose of the observer 170, illustrated in principle in figure 1 and in closer detail in fig. 5, is to perform the estimate calculating step 430 mentioned with reference to fig. 4 above.

More specifically, the purpose of the observer is to provide estimates of unmeasured states in the electro-pneumatic actuator, based on the position signal 192 supplied by the position sensor 190 and the control signal 162 supplied by the controller 160. More particularly, the estimator calculates estimates of the actuator velocity and actuator pressure.

In order to calculate an estimate of these unmeasured states in the actuator, the observer is adapted for operating a model which simulates the dynamics of the electro-pneumatic actuator. The model is non-linear, and it is arranged for modelling the pressure dynamics of the actuator.

The model is also arranged for modelling actuator load characteristics, in particular a clutch spring load characteristic and a friction load characteristic of the clutch.

The observer may equivalently be described by the following state-space model, i.e. by the following set of differential equations:

y = v + K₁(y -y) ϊ = ^ AfUP- ^AP ₀ -fι(y^,pM-ff(^y,p,v,t)]^{+ κ}i(^y-^y) ⁽D

P

wherein y is the estimated position and v is the estimated velocity of the clutch actuator, while p is estimated pressure in the clutch actuator. In accordance with common notation, the diacritic dot (^") indicates the time derivative.

The load function fι(y,p,v,t) corresponds to the element denoted "non-linear load" in fig. 5. The load function is arranged to provide the load force signal, which in general may be nonlinear and dependent on actuator position;/, actuator pressure p, actuator velocity v, and time t. The most significant variable is the position;/, so in a simplified implementation the load function may only be dependent on the position y. The characteristics of the load function fι(y,p,v,t), or in the simplified case, /, (y) , may routinely be predetermined by experimental tests of the electro- pneumatic actuator.

The friction function f_f (γ,p,v,t) corresponds to the element denoted "non-linear friction" in fig. 5. The friction function is arranged to provide the nonlinear friction force signal, which in general may be non-linear and dependent on actuator position y, actuator pressure p, actuator velocity v, and time t. The most significant variables are the velocity v and the pressure p, so in a simplified implementation the friction function may only be dependent on the velocity v and the pressure/?. The characteristics of the friction function f_f (y,p,v,t) or in the simplified case, f_f(v,p), may routinely be predetermined by experimental tests of the electro- pneumatic actuator.

An example of an experimental test for predetermining /_/ and /_/ ^■ may be explained as follows, with further reference to fig. 8. The horizontal axis indicates actuator position y, while the vertical axis indicates the applied force/ A linear spring may be used as a device for measuring the force. By connecting such a spring to the load, it is possible to determine the force needed to get the load in a given position. By measuring the force at multiple positions, a load characteristic with hysteresis, such as the graph illustrated in fig. 8, will be obtained. An increasing position of the load results in the graph segment denoted/_//,, while decreasing position results in the graph called/_//. The skilled person will realize that the above example is given for illustrative purposes only, and that numerous other possibilities exist for predetermining the functions/_/ and Jy.

When an appropriate data set has been acquired forβy), β and/_/- may be determined as:

The friction force function/_/- may be further improved by adding more terms in order to include additional friction effects, such as pressure dependent friction, position dependent friction, and viscous damping.

The flow function co(p,u) corresponds to the element denoted "non-linear flow" in fig. 5. The flow function is arranged to provide the nonlinear flow signal, which is non-linear and essentially dependent on actuator pressure p and the actuator control signal u. The characteristics of the flow function a>(j>,u) may routinely be predetermined by experimental tests of the electro-pneumatic actuator.

The output of the observer, denoted 172 in fig. 1, is a composite signal or vector signal which comprises the estimated states, in particular the estimated pressure p and the estimated velocity v . Advantageously, the observer output 172 will also comprise the actual position signal y, i.e. the position signal 192 supplied by the position sensor 190.

Further in the state-space model (1) and with reference to fig. 5, M is the total mass of movable parts in the actuator, A is the actuator piston area, and Po is the pressure on the opposite side of the piston. If the chamber on the opposite side is open, Po will thus equal the pressure external to the actuator, which is typically the atmospheric pressure. Kl, K2, KS are predetermined correction constants that are selected in order to adjust the model by adding an amount of the position difference (y-y) to each state pressure, velocity and position, respectively. Under ideal circumstances, Kl, K2 and K3 could all be set to zero. R is the gas constant for air, R = 288 J/ „ , and To is the reference temperature external to the actuator.

The elements denoted "integration method" in fig. 5 are integrators. Their implementation may be routinely selected by the skilled person according to the actual circumstances, e.g. as implicit or explicit Euler methods, implicit or explicit Runge-Kutta methods or the like.

The practical implementation of an observer 170, based on the above description, is a straightforward task for the skilled person. The model indicated by the equation set (1) may be transformed into executable computer program code, e.g. by means of well known control engineering software tools such as MATLAB and SIMULINK. The resulting code may be implemented in an ECU as explained with reference to fig. 3 above.

Fig. 6 is a schematic block diagram illustrating an example implementation of a controller.

The purpose of the controller 160, illustrated in principle in figure 1 and in closer detail in fig. 6, is to perform the control signal calculating step 440 mentioned with reference to fig. 4 above.

More specifically, the purpose of the controller 160 is to calculate the control signal 162 which is supplied to the clutch actuator 180 and also to the observer 170, based on the reference input 152 and the estimate 172 of at least one unmeasured state, output by the observer 170.

The illustrated controller 160 is non-linear.

The control signal 162 output by the controller 160 is denoted u. The controller input signals y (position), v (velocity) and p (pressure) may be either measured signals, measured and filtered signals, estimated signals, or a combination of measured signals, measured and filtered signals and estimated signals.

In an advantageous embodiment, the controller input signal y is connected to the measured position signal 192 provided by the position sensor 190, while the controller input signals v and p are connected to the estimator output signals v and p , respectively (cf. figure 5 and the corresponding description above).

In fig. 6, the input signals y_d , y_d , y^' _d and possibly higher order derivatives of y_d are calculated by the reference model, which is described in further detail with reference to fig. 7 below.

As appears from fig. 6, internal state variables ∑i, ∑₂ and Zs are defined for the controller.

The variables cc_\ and α₂ are auxiliary (substitute) variables used for simplicity The controller may equivalently be implemented by the following set of equations:

^zi = y -y_d

z₂ = v-a₁

<*2 - (C₁ + c₂>₂]

z₃ = p-a₂ (3)

The control signal output is provided by the following transformation: u = ω^'x(ω{p,u)) (4)

It should be understood that the remaining variables and parameters, as well as the elements denoted "non-linear flow" and "non-linear load" in fig. 6, have the same meaning and significance as previously explained in connection with the observer described with reference to fig. 5 above.

The skilled person will readily realize that numerous other possibilities for the controller design is possible within the scope of the invention. In particular, the controller 160 described in detail with reference to fig. 6 is arranged to cancel all non-linearities, in particular non-linear load and non-linear friction. In practice, however, certain non-linearities that have a stabilizing effect, should advantageously no be cancelled out by the controller. The controller 160 employs input not only from the reference yj itself, but also signals representing its derivatives, i.e. y_d , y^' _d and possibly higher order derivatives of y_d . This leads to advantageous results, and is based on the assumption that a reference model is also employed, or that such derivative signals are available by other means. However, the controller could alternatively be realized with only the reference y_d as input.

The practical implementation of the controller 160, based on the above description, is a straightforward task for the skilled person. The model indicated by the equation sets (3) and (4) may be transformed into executable computer program code, e.g. by means of well known control engineering software tools such as MATLAB and SIMULINK. The resulting code may be implemented in an ECU as explained with reference to fig. 3 above. Preferably, the controller code is implemented in the same ECU as the code of the observer 170. Alternatively, the controller code is implemented in a separate unit.

Fig. 7 is a schematic block diagram illustrating an example implementation of a reference model 150.

The purpose of the reference model 150 is to provide an improved reference trajectory, wherein noise and discontinuities in the original reference signal are reduced or cancelled. Another purpose of the reference model is to make available first and higher order derivatives of the reference trajectory, which may be necessary or at least advantageous for the performance and stability of the controller 160.

In order to simplify the figure, the reference model is illustrated on a vector/matrix form. The output _y_rf of the reference model is thus a vector containing calculated values representing the 0^th, 1^st, 2^nd (and possibly higher) derivatives of the reference signal y_r. For simplicity, the reference model is exemplified as providing only the first and second derivatives, i.e. the output vector has three components: y_d y, = y_d (5)

The matrices A_REF and .SR^ used in the reference model in this case are: W

12

and

0

B 'KBEΓFP — 0 (7)

where mo, mi and m^ are predetermined in order to keep the dynamics of the reference model exponentially stable. This is obtained by selecting the values of mo, mi and 1112 such that the characteristic polynomial s³ + m₂s² + Jn₁S + m₀ is a Hurwitz polynominal. The element denoted "integration method" is an integrator. Its practical implementation may be routinely selected by the skilled person according to the actual circumstances, e.g. as an implicit or explicit Euler method, an implicit or explicit Runge-Kutta method or the like.

The practical implementation of a reference model 150, based on the above description, is a straightforward task for the skilled person. The model indicated by the equation sets (5), (6) and (7) may be transformed into executable computer program code, e.g. by means of well known control engineering software tools such as MATLAB and SIMULINK. The resulting code may be implemented in an ECU as explained with reference to fig. 3 above. Preferably, the reference model code is implemented in the same ECU as the code of the observer 170 and the controller 160. Alternatively, the reference model code is implemented in a separate unit.

Fig. 8 is a graph illustrating an exemplary load characteristic, and has already been mentioned with reference to the description of fig. 5 above.

Although a preferred embodiment has been described in detail, it will be recognized that numerous alternatives and variations can be made without departing from the scope of the present invention, which is defined by the appended claims and their equivalents.

For instance, the detailed description specifies that a position sensor in the electro- pneumatic actuator is used for obtaining a measurement signal representative of the actuator position. However, the skilled person will understand that the signal that is representative of the actuator position may alternatively be obtained by means of an indirect measurement approach. Particularly, in the case of a clutch actuator, the position may be indirectly measured by measuring the torque transferred from the .

13

engine to the driveline. This torque measure is an indication of the position of the friction plates in the clutch, and may be suitable as the measurement signal representative of the actuator position, as required by the present invention.

The invention has been particularly described in relation to a vehicle clutch application. However, the skilled person will readily recognize that the principles of the invention will also be applicable in other applications wherein an electro- pneumatic actuator is to be controlled in a robust, accurate and reliable way. Such alternative applications include industrial automation, robotic control, and others.

The observer, the controller and the reference model have all been described as software realizations, i.e., embodied as computer program portions which are executed by a processor based control unit in the vehicle ECU. The skilled person will realize that the principles of the present invention likewise may be put into practice by other technological embodiments, such as by FPGAs (Field- Programmable Gate Arrays), ASICs (Application Specific Integrated Circuits) or even by analog circuits.

Claims

1. Method for controlling an electro-pneumatic actuator in accordance with a reference input, comprising the steps of

- providing the reference input, - providing a measurement signal representative of the actuator position,

- calculating an estimate of at least one unmeasured state in the electro-pneumatic actuator, based on the position measurement and a control signal,

- calculating the control signal based on the reference input and the estimate of an unmeasured state, and - supplying the control signal as an input signal to the electro-pneumatic actuator, c h ar a c t e r i z e d i n that the electro-pneumatic actuator is a clutch actuator having a non-linear load characteristic, and that said step of calculating an estimate of at least one unmeasured state comprises operating a non-linear model which simulates the dynamics of the electro- pneumatic actuator.

2. Method according to claim 1, wherein the step of providing a measurement signal comprises reading a position sensor in the electro-pneumatic actuator.

3. Method according to claim 1 or 2, wherein the non-linear model comprises a model of the pressure dynamics of the electro-pneumatic actuator.

4. Method according to one of the claims 1-3, wherein the non-linear model comprises a model of actuator load characteristics.

5. Method according to claim 4, wherein the model of actuator load characteristics comprises a spring load characteristic.

6. Method according to claim 5, wherein the model of actuator load characteristics further comprises a friction load characteristic.

7. Method according to one of the claims 1-6, wherein the step of providing a reference input comprises the substeps of

- providing a primary reference input, and

- calculating a first order derivative of the primary reference input by means of a reference model.

8. Method according to claim 7, wherein said step of providing a reference input further comprises calculating a second order derivative of the primary reference input by means of said reference model.

9. Method according to one of the claims 1-8, wherein the electro-pneumatic actuator is a clutch actuator in a vehicle.

10. Method according to claim 9, wherein the reference input is supplied by a CBW signal or an AMT signal.

11. Computer program for controlling an electro-pneumatic actuator in accordance with a reference input, for execution by a processing device in a control device, c h ar a c t e r i z e d i n that the computer program comprises instructions which cause the control device to perform a method in accordance with one of the claims 1-10.

12. Control device for controlling an electro-pneumatic actuator in accordance with a reference input, comprising

- a measurement device for providing a measurement signal representative of the actuator position,

- an observer for calculating an estimate of at least one unmeasured state in the electro-pneumatic actuator, based on the measurement signal and a control signal,

- a controller for calculating the control signal based on the reference input and the estimate of an unmeasured state, and - a control signal output for supplying the control signal as an input signal to the electro-pneumatic actuator, c h a r a c t e r i z e d i n that the electro-pneumatic actuator is a clutch actuator having a non-linear load characteristic, and that said observer comprises a non-linear model arranged for simulating the dynamics of the electro-pneumatic actuator.

13. Device according to claim 12, wherein the measurement device comprises a position sensor in the electro- pneumatic actuator.

14. Device according to one of the claims 12-13, wherein the non-linear model comprises a model of the pressure dynamics of the electro-pneumatic actuator.

15. Device according to one of the claims 12-14, wherein the non-linear model comprises a model of actuator load characteristics.

16. Device according to claim 15, wherein the model of actuator load characteristics comprises a spring load characteristic.

17. Device according to claim 16, wherein the model of actuator load characteristics further comprises a friction load characteristic.

18. Device according to one of the claims 12-17, wherein the reference input is connected to the output of a reference model supplied by a primary reference input, said reference model being arranged to calculate a first order derivative of the primary reference input.

19. Device according to claim 18, wherein said reference model is further arranged for calculating a second order derivative of the primary reference input.

20. Device according to one of the claims 12-19, wherein the electro-pneumatic actuator is a clutch actuator.

21. Device according to claim 20, wherein the reference input is supplied by a CBW signal or an AMT signal.

22. System for controlling an electro-pneumatic actuator in accordance with a reference input, c h a r a c t e r i z e d i n that it comprises a vehicle clutch (120), a clutch actuator (180) arranged to actuate the clutch (120), a position sensor (190) arranged to measure the position of the clutch (120), and a control device (100) as set forth in one of the claims 14-21 for controlling the clutch actuator in accordance with the reference input.