DE3833881C2 - Control device - Google Patents

Description

State of the art

From DE-OS 36 03 810 is a control device with the features of The preamble of claim 1 is known.

There, the actuator is usually used in braking systems for commercial vehicles Test used pressure control valve that has no linear control characteristic points. Rather, the current-proportional control curve is strong Hysteresis. This hysteresis was recorded there and in a regression line stored in a memory is compensated and thus the position link linearized. The regression lines necessary for the compensation are there off-line in a one-time initialization phase from the computer added. The regression line should only be recorded from time to time be performed again to control inaccuracy due to aging to eliminate the control device.

DE-OS 35 21 594 describes a method and a system for Improvement of the control quality in controlled systems with non-linear Behavior known, in which a corresponding to the manipulated variable Signal fed to a model of the controlled system becomes. A difference signal, which is the difference between the output signal of the model and one of the controlled variable derived control signal is evaluated and superimposed on the manipulated variable.

The object of the present invention is therefore in a Control device of the type mentioned an improvement the dynamics and robustness of the control loop.

According to the invention, this object is achieved by a rule device with the characterizing features of the claim 1. Further advantageous features are the subclaims remove.

Advantages of the invention

In the solution according to the invention, those required for the correction are used Parameters continuously determined on-line and in the correction device or the controller processed; so the adaptation is happening what an improvement in the dynamics and robustness of the control loop brings. The dynamic and static parameters are processed separately.  

The principle of the invention can be used universally in strongly non-linear systems, such as, for example, a characteristic curve with hysteresis ( FIG. 1a) and / or with a dead zone ( FIG. 1b) and / or with a bias voltage ( FIG. 1c).

Figure description

The invention is explained in more detail below on the basis of exemplary embodiments explained.

Show it:

Fig. 1 is not linear characteristic examples

Fig. 2 shows the model of a group consisting of proportional valve and relay valve valve combination,

Fig. 3 shows a control loop designed according to the invention.

In FIG. 2 a model of the actuator (valve combination) is shown. It is divided into two blocks. The manipulated variable, the control current i (t), is fed to a first block 1 ; it contains the static characteristic curves that can be described by two pairs of static parameters V _P and P _o (rise and fall characteristic). The straight lines can be described by the following expression:

P * = V · _P i (t) + P _o (1)

The resulting output pressure P * is calculated in a second block 2 with the dynamic parameters namely the time constant T _S (it takes a time until the corresponding pressure is present when i (t) changes) and the dead time T _t (after a change from i (t) the actuator needs the time T _t until it starts to react). One can see the influence by the expression

describe. If the two equations (1) and (2) are taken into account equal to the dead time, you get:

the following designations apply:

The parameters (V _P , P _o ) and (T _S , T _t ) of the model of equation (1) are variable, depending on where the operating point is. This fact can be represented by the block diagram of FIG. 2.

It can be seen that the model of equation (3) comprises the valve characteristic (characterized by the static parameters V _P , P _o ) and the dynamic behavior (described by the dynamic parameters T _S , T _t ). If these parameters are known, the characteristic curve compensation and the controller adaptation can be carried out as follows:

Characteristic curve compensation: i (t) = V _iP (t) + i _o (4)

Controller adjustment:

P = P _P + P _I (PI controller) (5)

P _P = K _{Pe W} (5a)

P _I = K _I · We _W dt (5b)

e _W = P _Soll - P _ist (5c)

K _P ∼ f (T _S , T _t ) (5d)

K _I ∼ f (T _S , T _t ) (5e)

Known algorithms such as e.g. B. recursive Least squares method with variable forgetting factor can be selected. The The basic equation for the parameter estimation is Eq. (3):

Fig. 3 shows an embodiment of a control circuit designed according to the invention with the valve combination proportional valve and relay valve as an actuator. The actuator is designated 11 . The inserted by him operated pressure or the corresponding measuring signal is _intended in a comparator 12 with the desired pressure signal P compared. The resulting control deviation e _W is fed to a controller 18 , which is constructed as a PI controller. It has the parameters K _P and K _I given above, which can be varied. Its output signal P is fed to a correction element 17 , which is the characteristic

i = V _iP + i _o

includes in general form; the parameters V _i and i _o are variable. Its output signal i (control signal) controls the actuator 11 .

The control signal i and the measurement signal of the applied pressure P are fed to a block 13 which, on the basis of equation (3), makes an estimate of the parameters and the estimated values _P , _o (static parameters) and _S , _t (dynamic parameters) to adapter elements 15 or 16 delivers. In these, the estimated values become parameter values of the correction element 17

or controller 18

K _P ∼ f ( _S , _t )
K _I ∼ f ( _S , _t )

converted and these parameters supplied to the correction element 17 or controller 18 as valid parameters. This ensures that the pressure P changes linearly with the control deviation (∼ means proportional. There are numerous empirical "setting rules" for a PI controller, eg the setting rule (rule of thumb) according to Chien, Hrones and Reswick) .

For the parameter estimation it is important to get information on where the operating point is. This can be done e.g. B. as in DE 36 03 810.5 by querying | e. . | and gain the sign of the change in the manipulated variable Δ _P. This allows the measured values to be separate parameter estimates, e.g. Assign for pressure build-up and reduction (selection of measured values). This is not shown in FIG. 3 for the sake of simplicity. It is also useful to use surveillance strategies, e.g. B. to specify when the parameter estimator remains active. For example, the parameter estimator must be switched off if Δ _{P is} below a threshold. This is indicated by a block 14 . It is also beneficial to use basic parameters for control and compensation during commissioning or error detection. The basic parameters can e.g. B. can be obtained in an initialization phase. This should be indicated by block 9 . If low-pass filtering of the estimated parameters is used, isolated estimation errors have only a minor effect on the control. The invention is not limited to the pressure control valve described above as an exemplary embodiment. Rather, it can be used for all strongly non-linear processes, the non-linearities of which can be described by static characteristic curves. The general structure corresponds to that shown in FIG. 3.

Claims (4)

1.Control device for regulating a controlled variable to a value corresponding to a setpoint value using an actuator in which the dependency of the controlled variable output at its output on the control variable supplied to the input is nonlinear but can be described by static characteristics, in which the controller is corrected is assigned to the facility, which takes into account the non-linearity of the actuator, the control deviation in such a manipulated variable that there is a linear relationship between the control deviation and the controlled variable, characterized in that with the aid of the current manipulated variable and the current controlled variable continuously in an identification process Serving parameter estimation method, the static and dynamic parameters of the non-linear behavior of the actuator are determined and that the signals corresponding to the dynamic parameters to the controller to influence the Sen parameters and the signals corresponding to the static parameters are fed to the downstream correction element for influencing the parameters of the static characteristic curves. 2. Control device according to claim 1, characterized in that the actuator is a valve or a valve combination that regulates a pressure and that the parameter estimate on the basis of the model equation T _S (t) + P (t) = V _p i (tT _t ) + P _o takes place, where P (t) the measured pressure, i (tT _t ) the manipulated variable, V _p as amplification and P _o as offset the static parameters and T _S as the time constant and T _t as the dead time the dynamic Parameters are. 3. Control device according to claim 1 or 2, characterized records that a separate parameter estimate is dependent speed of the direction of the change in the controlled variable. 4. Control device according to claims 1-3, characterized indicates that the parameter estimation is canceled when the controlled variable change takes a small value.