EP3341604A1

EP3341604A1 - Servo system for controlling the position of an actuator in a motor vehicle

Info

Publication number: EP3341604A1
Application number: EP16744823.2A
Authority: EP
Inventors: Ali ACHIR
Original assignee: PSA Automobiles SA
Current assignee: PSA Automobiles SA
Priority date: 2015-08-24
Filing date: 2016-07-11
Publication date: 2018-07-04
Anticipated expiration: 2036-07-11
Also published as: FR3040504A1; EP3341604B1; FR3040504B1; WO2017032931A1

Abstract

A servo system for controlling the position Y(s) of an actuator to a position setpoint R(s) in a motor vehicle, the actuator having a predefined transfer function G(s) associated with a pure delay d, said system being implemented via a position control U(s) input to the transfer function, the position of the actuator corresponding to the response of the actuator to the position control, said system comprising a pure delay compensator (50) in which the position control is used as a summing input of an adder (41) and as an input of a pure delay block (42), the output of the pure delay block being used as a subtracting input of the adder, the input of a high-pass filter (51) being connected to the output of the adder, the input of an amplifier (52) being connected to the output of the high-pass filter.

Description

ACTUATOR POSITION ASSISTING SYSTEM

IN A MOTOR VEHICLE

The invention relates to actuator position servo systems equipping a motor vehicle. More particularly, the invention relates to the compensation of pure delays present in these position control systems.

Continuous optimization of powertrains, especially to meet the different pollution standards and improve the performance of these groups, leads to an ever greater use of actuators in motor vehicles. Examples of actuators include actuators present in the air loop of a motor vehicle such as a valve / exhaust gas recirculation valve, commonly referred to as the EGR valve, or an air intake butterfly of a turbocharger.

Such actuators are commonly controlled by a computer of a motor vehicle, via the monitoring of a variable setpoint. FIG. 1 illustrates, by way of example, the overall architecture of a system 100 for controlling the position of an actuator 1, implemented in the computer of a motor vehicle. This architecture includes a supervisor 2 in charge of ordering (dashed arrows) the following tasks:

a task dedicated to the acquisition of the position of the actuator 1, performed by an acquisition chain 3;

a task allocated to a control of the position control of the actuator 1, performed according to the position of the actuator 1 acquired by the processing chain 3. This control task is performed by a position controller 4, implementing a control law of the actuator 1;

a task of electrical power control of the actuator 1, developed by a processing chain 5, performed according to the control of the servo in position of the actuator 1 previously established. Following this control task, from a measurement, information relating to the position of the actuator 1 is then returned to the acquisition chain 3 (arrow 101). [0004] FIG. 2a illustrates the servo-positioning performance of an actuator 1 of the air loop of a motor vehicle for a perfect system, that is to say having no pure delay. In this figure are illustrated two curves:

the curve 201 in thick dashed lines shows in ordinate, in percentage, a position setpoint for an actuator 1, as a function of a time in seconds represented on the abscissa;

the curve 202 in fine solid lines illustrates in the ordinate, at the same percentage scale, the position of the actuator 1 as a function of the same time in seconds represented on the abscissa.

Note in this Figure 2a that the position of the actuator 1 follows the position of the set value without exceeding exceeding 5% and has a response time tracking performance setpoint.

In practice, such a curve is difficult to observe, because of the existence of a pure delay existing in the system 100 servo. By way of example, with reference to FIG. 1, this pure delay can be attributed to accumulations of different delays such as:

delays relating to supervisor 2 and the order of execution of tasks;

the delays generated by the acquisition and processing chain 3;

the delays intrinsic to the position control system 100 of the actuator 1.

- Possibly intrinsic delays to the system itself.

Thus, in contrast to Figure 2a, it is common in the absence of compensation for this pure delay, to observe the curves shown in Figure 2b. In this figure are illustrated two curves:

the curve 203 in thick dashed lines illustrates in ordinate, in percentage, a position setpoint for the actuator 1, as a function of a time in seconds represented on the abscissa;

the curve 204 in full-line shows in ordinate, at the same scale of percentage, the position of the actuator 1 according to the same time in seconds represented in abscissa.

FIG. 2b shows a performance degradation compared with FIG. 2a, in particular the presence oscillations around the position set point. In fact, the absence of compensation for the pure delay causes the position controller 4 to significantly degrade the closed loop response time of the actuator 1, in order to maintain acceptable overshoot values of the position setpoint, typically less than 5%.

Another disadvantage of the absence of pure delay compensation, comes from the fact that the position setpoint determined by the computer is derived from an overall regulation of air flow or boost pressure in the control loop. 'air. The degradation of the response time of the actuator 1 then impacts the torque response of the vehicle engine. We can then see oscillations of pressure, supercharging, air flow, or even torque holes. Such a situation therefore has a strong impact on the degradation of driving pleasure. Moreover, from a mechanical point of view, the occurrence of oscillations at the level of the actuators 1 can quickly damage them. Their lifetimes are therefore diminished.

To overcome these problems, a known solution in the general field of automatic, is to add to a system 100 servo, such as that shown in Figure 1, a predictor 6 delay. The latter has the task of correcting the command developed by the position controller 4, in order to mitigate the impact of the pure delay on the quality of the servocontrol in the position of the actuator 1. To do this, the predictor 6 of delay takes as inputs, the inputs 102 and outputs 103 of the regulator 4, and develops a compensation 104 of the pure delay, it returns to the regulator 4 position.

Various solutions are known for producing a delay predictor in a servo system. For example, the document FR2749613 describes a system for regulating the richness of the air-fuel mixture in an internal combustion engine, this system comprising a control device produced in the form of a Smith predictor. The structure of the Smith predictor described in this document proves, however, particularly complex to implement, and remains expensive in terms of computational load induced. In particular, the implementation of such a solution involves a particularly complex adjustment of a set of parameters at the on-board computer in the vehicle, such a setting being expensive in time of realization. More generally, the predictors of existing delays are, in the state, not adapted to the position control for actuators equipping a motor vehicle.

An object of the present invention is to overcome all the aforementioned drawbacks.

A second object of the present invention is to provide a pure delay compensation device for an actuator position servo system fitted to a motor vehicle.

A third object of the present invention is to minimize the complexity of implementation of a pure delay compensation device, for an actuator position servo system in a motor vehicle.

A fourth object of the present invention is to optimally control the position of the actuators in a motor vehicle.

Thus, it is proposed, in a first aspect, a servo system of the position Y (s) of an actuator to a position setpoint R (s) in a motor vehicle, the actuator having a function predetermined transfer G (s) associated with a pure delay e ^{~ ds} , s being the Laplace variable, d representing the pure delay, this system being realized via:

a position control U (s) at the input of the transfer function G (s), the position Y (s) of the actuator corresponding to the response of the actuator to the position control U (s);

the position control U (s) being elaborated using a Proportional Derivative type regulator, formed by a Proportional action loop and a Derivative action loop, made respectively by a first and a second loop of feedback between the Y (s) position and subtractor inputs of a first adder, the controller performing a third feedback loop between the Y (s) position and a subtractor input of a second adder, the second adder having as summing input the position setpoint R (s); the position setpoint R (s) also being used as input of an amplifying loop whose output is connected to an summing input of the first adder;

the output of the second adder being used as an input of an integral action loop whose output is connected to an summing input of the first adder, so that the first adder calculates at its output the position U (s) according to its inputs summators and subtractors;

this system comprising a pure delay compensator in which: the position control U (s) is used as the first summing input of a third adder and as input of a pure delay block of transfer function e ^{~ ds} , the the output of this pure delay block being used as the subtractor input of the third adder, so that the third adder calculates its output from its summing input and its subtractor input;

the input of a high-pass filter is connected to the output of the third adder;

the input of an amplifier is connected to the output of the high-pass filter;

the output of the amplifier corresponds to a delay compensation term P (s), this output being used as a subtracting input by the second adder.

Advantageously, in this servo system, the high-pass filter and the amplifier are sized from a first-order system approximating the transfer function G (s) of the actuator, this system having a static gain k and a time constant τΐ.

Advantageously, in this servo system, the amplifier has for gain, the static gain k.

Advantageously, in this servo system, the high-pass filter has the function of transfer. where τ2 is a parameterizable time constant.

Advantageously, in this servo system, the Proportional action loop is associated with a gain k £, the loop to Integral action is associated with a gain k [, the time constant τ2 being calibrated so that τ2 = - ^R.

Advantageously, in this servo system, the actuator is an actuator of the air loop of a motor vehicle.

It is proposed, in a second aspect, a motor vehicle computer implementing an actuator position control system, performed as described above.

It is proposed, in a third aspect, a motor vehicle comprising the computer described above.

Other objects and advantages of the invention will appear in the light of the description of an embodiment, given below with reference to the accompanying drawings in which:

FIG. 1 illustrates the overall architecture of a position servo-control system of an actuator implemented in a computer equipping a motor vehicle;

FIGS. 2a and 2b illustrate the temporal variations of the position of an actuator, respectively for a perfect and real servo-control system, with respect to a position command;

FIG. 3 illustrates a device for controlling the position of an actuator;

Figure 4 illustrates a position servo system of an actuator including a Smith predictor;

FIG. 5 illustrates a system for controlling the position of an actuator comprising a pure delay compensator according to one embodiment.

In Figure 3 is shown a position control device 300 of an actuator 1, the device 300 having a pure delay d_. The actuator 1 is, for example, an actuator of the air loop of a motor vehicle, such as a valve / a valve of an exhaust gas recirculation system, or a throttle valve. admission of a turbocharger. The servo-control device 300 is here represented in the Laplace domain and can be implemented in a position controller 4, such as that illustrated in FIG. As explained in the introductory part, the position controller 4 is configured to implement a control law of the position of the actuator 1. To do this, the behavior of the actuator 1 in the control device 300 is modeled by a transfer function G (s) in the Laplace domain, where s is the Laplace variable.

The transfer function G (s) can be obtained in different ways, for example communicated directly by a model provided by the manufacturer of the actuator 1, modeled via an appropriate simulation tool (eg Simulink®), or still obtained experimentally by applying to the actuator 1 a step-type input and observing its output response.

The pure delay d. associated with the actuator 1 is meanwhile represented by the term e ^{~ ds} in the Laplace domain.

Thus, in FIG. 3, the actuator 1 and its pure delay are modeled by a transfer function block 30 G (s) e ^{~ ds} , receiving for input a position command U (s) destined for the actuator 1.

In practice, the actual position of the actuator 1, following receipt of the position control U (s), corresponds to an ideal response of the actuator 1, to which is added a disturbance on its position. The ideal response of the actuator 1, following receipt of the position command U (s), corresponds to the application of the transfer function G (s) e ^{~ ds} , whose response is obtained at the output of the block 30. The disturbance in position results, in turn, in particular aeraulic forces applying to the actuator 1, for example pressure variations during its movement.

The disturbance on the position of the actuator 1 is represented in FIG. 3 by the function D (s) at the input of an adder 7, the other input of this adder 7 corresponding to the ideal response of the actuator 1 following the command U (s), that is to say here to the output response of the block 30 without disturbance. Thus, at the output of the adder 7, a real position Y (s) of the actuator 1 is observed in response to the command U (s). It will be noted here that the summator 7 and the variable D (s) are represented here for purposes of theoretical understanding. In practice, and in the remainder of this document, we consider the position Y (s) of the actuator 1 as the "real" output of the transfer function block 30 G (s) e ^{~ ds} , that is to say as the position of the actuator 1 comprising possible external disturbances.

The actuator 1 thus has the output of the transfer function block G (s) e ^{~ ds} a position Y (s) in response to the command U (s), this position Y (s) being the position that it is desired to slave to a position setpoint value R (s).

The position control U (s) is developed as an output of a first adder 31 having two subtracting inputs "-" and two summing inputs "+". The subtractor inputs of the first adder 31 are produced by a Regulator of the Proportional Derivative type, commonly referred to by the acronym "PD regulator". As its name suggests, this regulator is formed of a proportional action loop, performing a gain amplification operation kjD;

a Derivative action loop, performing a derivation operation, symbolized by the block "→ _>> sometimes also symbolized by a block" s ", associated with a gain kd_.

The proportional action loop and the derivative action loop respectively provide a first and a second loop 32, 33 of feedback between the output of the block 30, that is to say the position Y (s), and a subtractor input "-" of the first adder 31.

A third feedback loop 34 is also provided by the controller between the position Y (s) and a subtracting input "-" of a second adder 35.

The second adder 35 further comprises, for summing input "+", the position setpoint value R (s), this value being determined by the motor vehicle computer, as a function, for example, of parameters measured in the loop. air. This position setpoint value R (s) corresponds to the setpoint value at which it is desired to control the position Y (s) of the actuator 1. Thus, the output of the second adder 35 corresponds to the difference between the setpoint of position R (s) (on the summing input "+") and the position Y (s) of the actuator (on the subtracting input "-"), that is to say to a position error e (s) of the actuator 1, where e (s) = R (s) -Y (s). The position reference value R (s) is also used as input of a gain amplifier loop kf_ whose output is connected to a summing input "+" of the first adder 31.

In addition, the output of the second adder 35, that is to say the position error e (s) of the actuator 1, serves as input to a loop 37 with integral action, performing an operation integration, symbolized by the block "I", sometimes also symbolized by a block "1 / s", associated with a gain k [. The output of this integrally acting loop 37 is connected to a summing input "+" of the first adder 31. Thus, as a function of all of its summing inputs "+" and subtracters "-", the first adder 31 determines the control in position U (s) to be applied to the actuator 1.

The gains kf, k [, k £ and kd_ are static values currently determined with respect to the transfer function G (s), previously known, for example via the application of a pole placement method. These parameters are, by way of example, determined by simulation via a development tool of the motor vehicle computer and / or adjusted experimentally at the computer, for example as a function of the course of a series of tests intended to validate the computer. specifications of the vehicle. These gains are here preconfigured.

As just described, it will be noted that the device 300 for controlling the position of an actuator Icomprets in particular: feedback loops 32, 33 on the position Y (s) of the actuator 1, performed by a regulator of Proportional Derivative type;

a loop 37 with integral action performed on the position error e (s) of the actuator 1, before the development of the position control U (s). Because of this configuration, such a loop is commonly referred to as the anticipatory loop.

Thus, the structure of this type of position control device 300 is sometimes referred to as PD-I (acronym for "Proportional Derivative - Integral"). Such a structure differs from common PID (Acronym Integral Derivative) controllers, in that only the integral term lies on the error of position e (s), while the proportional and derivative actions are indexed only on the position Y (s) of the actuator 1. This has the advantage of being able to decouple the performance of tracking setpoint and the performance of rejection of disturbances. Thus, such a servocontrol structure is realized in motor vehicle computers for the position control of the actuators 1.

The position control device 300 of FIG. 3 comprises here for transfer function:

Note in this equation the presence at the denominator of the term e ^{~ ds} , corresponding to the pure delay of this device 300. It is the presence of this term which is at the origin of oscillations (see for example Figure 2b) observed on the position of the actuator 1 when it tries to follow the position setpoint R (s).

To compensate for this delay, we then add to the device 300 described above a delay compensator, such a Smith predictor 40 shown in Figure 4. In this figure, the predictor 40 of Smith is realized in the following manner.

The position control U (s) at the output of the first adder 31 is used as the summing input "+" of a third adder 41 and as input of a block 42 of pure transfer function delay e ^{~ ds} , the output of this delay block being used as the subtracting input "-" of the third adder 41. The block 42 use of pure delay implies that the delay d. is known. This is in practice estimated by applying to the actuator 1 a step-type input and observing its output. It can be calibrated more finely when setting the position controller 4.

The third adder 41 then calculates its output from its summing input "+" and its subtracting input "-". This output is connected as input to a transfer function correcting block C (s) .G (s) whose output is used as a subtractor input "-" of the first adder 31. The term C (s) here denotes a function parameterizable transfer specific to the block 43 corrector. Advantageously, the presence of the Smith predictor 40 makes it possible to extract the pure delay e ^{~ ds} from the feedback loop 34 between Y (s) and R (s), that is to say of the closed loop. the transfer function

Y (s ^' ) J -

Indeed, if one chooses for figure 4, C (s) = k _p + - + k _d s, the transfer function of the system is written then:

l + G (s) / _p + c + / c _d s

It can be seen in this equation that the pure delay is outside the servo loop since the denominator no longer depends on the term e ^{~ ds} .

Advantageously, this eliminates the oscillations observed on the response of the actuator 1. However, we note here that the pure delay e ^{~ ds} has not disappeared. If we denote H (s) the closed-loop transfer function without this pure delay such that

G s) (k _f + ^) _{Y (s)}

H (s) =. it follows that ^ = H (s) e-

₁₊ G (s) (k _{p +} ^ + k _d s) "W

The transfer function of this system is therefore equal to a closed-loop response H (s) of the actuator 1, shifted in time by a pure delay d_. Following receipt of the position command U (s), the position Y (s) of the actuator 1 will therefore be followed by a tracking of the position setpoint R (s) close to FIG. 2a, with a simple offset temporal and no longer oscillations on the position Y (s) of the actuator 1. The pure delay d. has not disappeared, but becomes less detrimental to the position control of the actuator 1. The use of a predictor 40 Smith is therefore effective for the position control of the actuator 1. However, this predictor 40 requires a fine and complex adjustment of all the parameters of the transfer function H (s).

To overcome this drawback, in one embodiment, the transfer function G (s) of the actuator 1 is then approximated by a first-order system:

k.

G (s) ", where k and ri are respectively a static gain and a time constant associated with the first order system, these values also being specific to the transfer function G (s) of the actuator 1. Advantageously, the variables k and he are easily deductible from the actuator 1. These variables are for example determined by simulation via an approximation of the transfer function G (s), or experimentally, by observing the response of the actuator 1 to a step applied at the input of it.

Suppose, moreover, that ^ "0: such an approximation can be performed when calculating the parameters kd. and kl determined during the implementation of the actuator position control device 300 illustrated in FIG. 3. The product C (s) .G (s), that is to say the transfer function of the block 43 corrector, can then be approximated in this way:

C (s) .G (s) = ^k i ^'k _¾ ¾ι¾ (τ ₂ ⁵ + ι) ^ _{τ2 is a cons} t _an t _e _e d configured time.

If a transfer function T (s) such as:

T (s) = ^k ^ ^{2S + 1} ^ _; we note that it is then enough to set the time constant χ2 to the value to approximate the product C (s) .G (s).

Furthermore, it is observed that the transfer function T (s) is k- the product of a gain integrator k1 (term, a high-pass filter of time constant τ2 having the function of transfer ^ and the static gain k of the actuator 1.

Following this observation, in an embodiment shown in FIG. 5, the previously described control device 300 which is associated with a pure delay compensator 50 is implemented in the following manner.

The position control U (s) at the output of the first adder 31 is used as the summing input "+" of a third adder 41 and as input of a block 42 of pure transfer function delay e ^{~ ds} , the output of this delay block being used as the subtracting input "-" of the third adder 41. The third adder 41 then calculates its output from its summing input "+" and its subtracting input "-". This output is connected as input to a high-pass filter 51. Advantageously, the high-pass filter 51 has the function of transfer. where he is the predetermined time constant of the actuator 1 approximated by a first order system and where τ2 = -A

The output of the high-pass filter 51 is connected to a gain amplifier 52 k, where k. is the static gain k. (predetermined) of the actuator 1.

The high-pass filter 51 and the amplifier 52 are therefore sized from the first-order system approximating the transfer function G (s) of the actuator 1.

The output of the amplifier 52 then corresponds to a delay compensation term P (s). This output is used as the subtracting input "-" for the second adder 35.

Advantageously, as explained above, the variables kjD and k [are preconfigured during the realization of the servo system. Similarly, the static gain k and the time constant II are predetermined variables of a first-order system approximating the transfer function G (s) of the actuator 1. Thus, for the compensator 50 of pure delay, the the only remaining magnitude to be calibrated is the variable r2 of the high-pass filter 51.

This calibration is easily performed by calibrating τ2 to the value -. The realization of this compensator 50 of pure delay is therefore very fast, easy to implement, and does not imply complexity in terms of computing load.

Advantageously, thanks to the embodiments described above, the calibration of actuator position servo systems are simplified, which allows to gain time during their development.

Advantageously, the embodiments described above are applicable to any actuator 1 of the air loop of a motor vehicle, for example to a valve / valve for the recirculation of the exhaust gas, or to an air intake butterfly of a turbocharger.

These embodiments make it possible to obtain better control of the actuators 1 in the air loop, and thus to limit oscillation phenomena in this loop, such as supercharging pressure oscillations, air flow rates. , or engine torque. The performance of the air system controls are seen therefore improved, as well as driving pleasure. Moreover, the improvement of the position control of the actuators 1 makes it possible to limit the phenomena of micro-actuations due to their position oscillations. Advantageously, the service life of the actuators 1 is extended. In general, the robustness and accuracy of actuator position control systems 1 are improved.

Claims

1. System for controlling the position Y (s) of an actuator

(1) at a position setpoint R (s) in a motor vehicle, the actuator (1) having a predetermined transfer function G (s) associated with a pure delay e ^{~ ds} , s being the Laplace variable, d . representing the pure delay, this system being realized via

a position command U (s) at the input of the transfer function G (s), the position Y (s) of the actuator (1) corresponding to the response of the actuator (1) to the position control U (s); the position control U (s) being elaborated using a Proportional Derivative type regulator, formed of a Proportional action loop and a Derivative action loop, made respectively by a first and a second loop ( 32,33) between the Y (s) position and subtracting inputs of a first adder (31), the controller realizing a third feedback loop (34) between the Y (s) position and a subtracting input of a second adder (35), the second adder (35) having as summing input the position setpoint R (s);

the position setpoint R (s) also being used as input of an amplifying loop (36) whose output is connected to an summing input of the first adder (31);

the output of the second adder (35) being used as an input of an integral-acting loop (37) whose output is connected to a summing input of the first adder (31), so that the first adder calculates at its output the position U (s) according to its summing and subtracting inputs;

this system being characterized in that it comprises a compensator

(50) pure delay in which

the position control U (s) is used as the first summing input of a third adder (41) and as the input of a block (42) of pure transfer function delay e ^{~ ds} , the output of this block (42 ) of pure delay being used as a subtractor input of the third adder (41), so that the third adder (41) calculates its output from its summing input and its subtracting input;

the input of a high-pass filter (51) is connected to the output of the third adder (41);

the input of an amplifier (52) is connected to the output of the high-pass filter (51);

the output of the amplifier (52) corresponds to a delay compensation term P (s), this output being used as a subtracting input by the second adder (35).

2. A servo control system according to claim 1, wherein the high-pass filter (51) and the amplifier (52) are sized from a first-order system approximating the transfer function G (s) of the actuator (1), this system having a static gain k and a time constant τΐ.

3. Control system according to claim 2, wherein the amplifier (52) has for gain, the static gain k.

4. Servo control system according to claims 2 or 3, wherein the high-pass filter (51) has the function of transfer, where τ2 is a parameterizable time constant.

5. Servo-control system according to claim 4, wherein the proportional action loop is associated with a gain J <p_, the loop (37) integral action is associated with a gain k [, the time constant χ2 being calibrated so that τ2 = - ^.

6. Control system according to any one of the preceding claims, wherein the actuator (1) is an actuator of the air loop of a motor vehicle.

7. Motor vehicle calculator implementing an actuator position servo system, made according to any one of the preceding claims.

8. Motor vehicle comprising the calculator of claim 7.