DE29722692U1 - Device for designing a control system for an oscillating line and control device - Google Patents

Description

Device for designing a control system for an oscillating system and control device 5

The invention relates to a device for designing a control system for an oscillating system according to the preamble of claim 1 and to a control device for an oscillating system with a controller that calculates a manipulated variable in a closed control loop from a control difference that is formed from a reference variable and a controlled variable measured on the system.

Numerous analytical and empirical design methods are known for the design of controllers for aperiodic systems that do not exhibit oscillation behavior when excited with a step function. For example, the method based on the absolute value optimum is described in the book "Control Engineering" by Otto Föllinger, Hüthig-Verlag, Heidelberg, 6th edition, 1990, pages 258 to 261. The absolute value optimum, however, requires a system whose behavior can be approximated by a model with only real poles or with sufficient damping. This condition is often not met by PT2 systems capable of oscillation. Oscillation-capable systems contain at least two energy storage devices. Oscillations can arise through the exchange of energy between the storage devices. Examples of this are spring-mass systems, coil-capacitor systems or electromechanical converters such as moving coil instruments. Feedback systems, such as controls, can also represent oscillation-capable systems. As stated in the above-mentioned book by Otto Föllinger on pages 41 to 45, a 2nd order oscillatory system (PT2 system) is described by the following differential equation:

GR 97 G 4464 DE „

r-*x + 2*d*T*x + x = K*y withK,T,d>0 Eq. (1)

Here, T is the time constant, the dimensionless number d is the damping factor and K is the gain of the path. y(t) describes the course of the input variable and x(t) the course of the output variable. For this, in the formula, y is simplified
or &khgr;. One or two dots indicate the
temporal derivative of the respective quantity. From the differential equation, the transfer function of a PT2 system is obtained using Laplace transformation:

For d = 0 the step response of the system executes an undamped continuous oscillation, for 0 < d < 1 the step response is a decaying oscillation and for d >_ 1 an aperiodic transient response.

A different damping of the transient response is obtained,
by connecting a 1st order delay element in series with a PT2 model. The transfer function is then:

with b > 0 Eq. (3]

4t = 7 T^-,

y{s) (\ + 2dTs + T ² s ² )*(\

The factor b is a dimensionless number whose value is greater than zero. The additional delay element can, for example, determine the behavior of an actuator or a
Model the measuring device required for control.

Choosing a suitable model and determining the
Model parameters can be determined using theoretical or experimental

GR 97 G 44 64 EN

teller method. In an experimental model, the system is stimulated with a change in the control signal y. The parameters of a model according to equation (2) or (3), for example, are then determined from the transient response of the system using parameters or numerical optimization. A design tool suitable for this identification method is, for example, the software tool SIEPID S5 from Siemens AG, which is used to commission and optimize control systems.

If the identification results in a model of the system that has a damping value d < 1 and thus a conjugated complex pole pair, then known design methods for aperiodic systems cannot be easily transferred to the design of a controller for the oscillating system. If, for example, the setting rules for controller parameters based on the optimum amount are applied to these systems, the control loop will behave unstably for small values of d. A controller for oscillating systems can be designed with great effort in an iterative process by manually varying the controller parameters based on empirical setting rules until a control quality that satisfies the requirements is achieved, or by numerically optimizing the controller parameters. Controllers designed in this way are not always optimal, e.g. because the respective controller type does not allow negative reset times or lead times, because the transient response is too slow for an iterative design, or because the controller reacts sensitively to parameter changes in the system.

The invention is based on the object of creating a device for designing a control system for an oscillating system, with which a control device with improved control quality can be obtained in a simple manner.

GR 97 G 44 64 DE ,.

To solve this problem, the new device of the type mentioned at the beginning has the features specified in the characterizing part of claim 1. Advantageous further developments of the device are mentioned in claims 2 and 3. A control device obtainable with the new design device is described in claims 4 and 5. Claims 6 to 9 specify advantageous embodiments of the new control device.

The invention has the advantage that an iterative approach to designing the controller for oscillating systems is no longer necessary. The required controller parameters and the transfer function of a compensation element to be introduced into the control loop can be determined directly using the system model. The method can be integrated into existing software tools, e.g. SIEPID S5 from Siemens AG, with little effort. By loading a software tool created in this way onto a suitable computer, the new device for designing a controller for oscillating systems can be created without any particular difficulties. The control device designed in this way can be implemented as an analog computing circuit using an electronic circuit or as a digital sampling controller by loading appropriate software into a digital controller.

The compensation element can alternatively be used in a loop in the form of a negative feedback or in a feedforward control. A negative feedback structure has the advantage that it is very clear and easy for the user to handle. In contrast, a feedforward control offers a faster response of the controlled variable to changes in the reference variable in sampling controls.

If the controller’s feedforward control is used to superimpose the compensation signal on the manipulated variable,

GR 97 G 4464 DE &ldquor; ,,

Standard controllers can be used. A bumpless switchover from manual to automatic operation is guaranteed if the controller also takes into account the compensation signal fed to the disturbance variable feedforward when adjusting the I component required for this. The system supplemented by the compensation element has an aperiodic transient response. The controller can therefore be designed analytically with little effort according to the optimum amount. The optimum amount provides a controller with good control quality that is robust against parameter fluctuations in the system model. Better robustness against measured value noise is achieved by using a real lead element as a compensation element. Very good results for systems with operating point-dependent parameters are provided by a control device in which the gain factor Kv can be changed depending on the current operating point.

In particular, compared to a simple PID controller, the new control system delivers better control quality through faster, aperiodic oscillation of the controlled variable and a smoother manipulated variable curve. An actuator in the control loop is therefore subjected to less load. By wiring the disturbance variable feedforward of a standard controller, a control structure similar to a cascade control is created, but without the need for an additional controller or measuring sensor.

In cascade control systems, subordinate control loops can be set more "sharply" using the new design facility.

This means that disturbances are better regulated and the control quality is improved. In cascade control, a slave controller no longer has to be weakened to such an extent that the master controller can be designed using the known methods for aperiodic systems, but the subordinate control loop can certainly exhibit oscillatory behavior.

GR 97 G 4464 EN

As already explained, the oscillation behavior of a system is influenced by the damping d: For d > 1, the transient response is aperiodic, but for a very small damping it oscillates strongly. Since the damping d of the system can be changed by the compensation element, systems with variable damping can also be easily controlled with the new control device. An example of this is a cascade control in which the underlying control loop contains a variable, operating point-dependent dead time. With the new design device, the compensation element can easily be adapted to an operating point-dependent damping. For the controller, the behavior of the system supplemented by the compensation element is always aperiodic and can be easily controlled.

The invention, as well as its embodiments and advantages, are explained in more detail below with reference to the drawings in which embodiments of the invention are shown.

Show it:

Figure 1 shows a control system as a block diagram, Figure 2 shows a control system supplemented by a compensation element,

Figure 3 shows a closed control loop with a control system supplemented by a compensation element, Figure 4 shows a table with controller setting rules according to the

Amount optimum,
Figure 5 shows a closed control loop with compensation by a feedforward control and Figure 6 shows a cascade control.

Figure 1 shows a system with a transfer function G(s), an input variable y(s) and an output variable x(s) in the Laplace domain. The new design method is based on the knowledge that an oscillatory

GR 97 G 4464 EN

PT2 system with a transfer function G(s) according to equation (2) can be converted into a feedback structure as shown in Figure 2. This structure consists of a system 1 with aperiodic transient response, i.e. a system 1 with a transfer function according to equation (2) and a damping d = 1, which is fed back via a lead element 2. An output variable x(s) is fed via the lead element 2 and is additively superimposed on an input variable y(s) by means of a summer 3. Since the output signal of the lead element 2 is fed to the summer 3 with a positive sign, this feedback can be referred to as positive feedback. The lead element 2 only evaluates changes in the controlled variable, so that the feedback is only effective in the dynamic case. It has the transfer function V(s) = Kv * s with Kv - gain factor and s - Laplace operator. The transfer function of the line 1 supplemented by the lead element in the feedback structure is:

G(s) ₌ K

y(s) lG(s)*Kv*s 1 + (2* T-KvK)* s+T ² * s ²

The transfer functions according to equation (2) and equation (4) are identical if the gain factor Kv of the lead element 2 satisfies the condition:

2 * (I - d) * T

Kv = Eq. (5)

An aperiodic transient response with d > 1 is achieved with negative Kv, i.e. with a dynamic negative feedback. For a damping value d < 1, the section 1 supplemented by the lead element 2 shows an oscillating transient response. In this case, Kv is positive and the feedback structure represents a true positive feedback.

GR 97 G 4464 EN

The idea for designing a control system for oscillating systems is to compensate for this positive feedback, which can convert an aperiodically damped system 1 into an oscillating system, by means of an upstream negative feedback, so that a controller 4 in a closed control loop according to Figure 3 can be designed for a system with aperiodic behavior due to an additional feedback structure, even though the system 6 actually to be controlled is oscillating. The controller 4, for example a PI or PID controller, calculates a control variable y'(s) from a control difference xd(s), which is formed by a subtractor 5 from a reference variable w(s) and a controlled variable x(s) measured on the oscillating system 6. A compensation signal, which is generated by a compensation element 7 depending on the controlled variable &khgr;(s), is superimposed on this control variable y'(s) by a subtractor 8 with a negative sign. The result of this superposition is fed to the path 6 as a control signal y(s). The compensation element 7 is implemented as a lead element and in turn has the transfer function Kv * s. The gain factor Kv is calculated so that the path 6 supplemented by the compensation element 7 has an aperiodic transient response. The feedback structure formed from the oscillating path 6, the compensation element 7 and the subtractor 8 therefore represents a fictitious path with a predeterminable aperiodic response, which has the same order as the oscillating path 6 to be controlled. Conversely, in a positive feedback structure, the output signal of another compensation element with the same transfer function as the compensation element 7 could be fed back to the input of the fictitious, aperiodic path in order to obtain the damping d of the oscillating path 6 in the loop thus formed from the fictitious, aperiodic path and the positive feedback. According to Figure 3, the output value of the controller 4, preferably a PI-

GR 97 G 4464 EN

or PID controller, a compensation signal from the compensation element 7 is superimposed on the control signal of the PID controller 8, which can advantageously be implemented by a disturbance variable feedforward integrated in the controller 4. This compensation signal is formed by multiplying the controlled variable x(s) by the transfer function of the lead element 7, which has the gain Kv. The controller 4 now controls a system with a predeterminable aperiodic PT2 behavior, i.e. a system with a damping d > 1. In the aperiodic limit case, the damping d = 1. In principle, however, any other value of the damping d can be specified. For the system 6 supplemented by the compensation element 7, the controller parameters can now be calculated using conventional methods for controlled systems with real poles, e.g. according to the known optimum amount.

Alternatively, an aperiodically damped system could be converted into an oscillating system by adding a negative feedback structure with a simple gain in the feedback. In this case, the damping can also be set using the gain factor in the feedback. The disadvantage of this, however, is that the gain factor and the time constant of the system also change depending on the damping. If a PT2 system according to equation (2) with the damping d = 1 is negatively fed via a proportional element with the gain Kv, the following transfer function is obtained for the loop formed in this way:

x(s) _ G(s) _ K

y(s) l + G(s)*Kv (l+Kv*K)+2*T*s +

\±Kv*K

2*7

Eq. (6)

GR 97 G 4464 EN

In principle, this approach is also possible for designing a control system for oscillating systems. Overall, however, feedback via a derivative element is more advantageous.
5

If a PT3 model was determined according to equation (3) during the route identification, the compensation element can be calculated using a modified procedure described below.
10

The denominator of the transfer function of the PT3 system is:

(l+2dTs+T ² s ² )*(l+bTs)=l+(2d+b)Ts+(l+2bd)T ² s ² +bT ³ s ³ Eq. (7)

The denominator of a PT3 model supplemented by a derivative element 2 as shown in Figure 2 is:

(l + 2Tls + Tl ² s ² ) * (l+bl*Tls) -K*Kv*s=

1+ (2Tl+bl*Tl-K*Kv) s+ (l+2*bl) *Tl ² s ² +blTl ³ s ³ Eq.(8)

The damping factor d is contained in the denominator of the transfer function of a PT3 model in two coefficients. Therefore, additional new values bl and Tl must be determined in the transfer function of the feedback structure so that the PT3 model and the feedback structure have the same transfer behavior. The parameters Kv, bl and Tl are calculated according to the following equations, which can be derived from equations (7) and (8) by comparing the coefficients:

s ¹ : (2d + b)*T = (2 + bl)*Tl - K*Kv Eq. (9)

s ² : (1 + 2bd)*T ² = (1 + 2bl)*Tl ² Eq. (10)

s ³ : b*T ³ = bl*Tl ³ Eq. (11)

GR 97 G 4 4 64 EN

From equation (11) it follows for Tl:

Eq. (12)

In equation (10), Tl is replaced by this value and abbreviated by T ² . The equation for determining bl is then:

2/3

Eq. (13)

An acceptable value for bl can be determined using a simple algorithmic loop in which bl is incremented starting from zero and the value for which equation (13) has the smallest absolute value is sorted out. In principle, equation (13) could also be solved numerically. However, a numerical solution could yield solutions that are mathematically correct but technically unusable.

Using equation (9), Kv is calculated as:

bl)*Tl-(2d

Eq. (14)

In the case of an oscillatory system with PT3 behavior, it is advantageous to design the controller 4 (Figure 3) for the system with predetermined aperiodic behavior, which is formed from the feedback structure, with the three real poles 1/Tl, 1/Tl and l/(bl*Tl).

Tests of the new control system have shown that it is so robust that a value of the gain Kv calculated according to equation (5) and a controller design for an aperiodic PT3 system with the parameters K, b, T of the

GR 97 G 4464 EN

already provides solutions that can be used to advantage for the identified, oscillating PT3 section.

Using the setting rules given in the book by Otto Föllinger mentioned at the beginning, the parameters of a PI or PID controller listed in the table in Figure 4 for the aperiodic limit case of a PT2 or PT3 model can be calculated based on the optimum amount. An empirically obtained value is given as the controller gain of a PID controller for the PT2 model. The controller type is entered in the first column of the table from the left, the respective controller parameter in the second, and the values of the parameters for the PT2 and PT3 systems are entered in the third and fourth columns.

The setting rules refer to the following ideal PID control structure:

4^ ⁼ ^&iacgr; ^{1 +} &Tgr;

xd(s) y Tns

with Kp - controller gain,
Tn - reset time and
Tv - lead time.

The design process is very robust and tolerates large parameter fluctuations. If a real lead element with a delay in the transfer function is used in the feedback instead of the ideal lead element as a compensation element, the robustness of the new control system against measured value noise can be improved.

As an alternative to the feedback via the compensation element 7 shown in Figure 3, a feedforward control can also be used in which the compensation signal is not

GR 97 G 4464 EN

the controlled variable x(s), but rather from the setpoint w(s), as shown in Figure 5. Via a compensation element 9 with the transfer function Kv * s, the reference variable w(s) is superimposed by a subtractor 10 to calculate a manipulated variable y(s) on the output value y'(s) of a controller. The manipulated variable y(s) is applied to an oscillating system 12 to be controlled. The controlled variable x(s) is recorded on the system and fed to a subtractor to form a control difference xd(s) from the reference variable w(s) and the controlled variable x(s). The control difference xd(s) is in turn the input variable of the controller 11. Rl denotes the transfer function of the controller 11 and R2 the transfer function of the compensation element 9. G is the transfer function of the system 12. The control behavior of this control system can be calculated as:

x=G* [-R2*w + Rl*(W-X)] Eq. (15)

x \Rl-R2]*G

&mdash; = 1 * Eq. (16)

w l + G*Rl

For better clarity, the bracket expression with the Laplace operator s has been omitted from the names of the individual quantities.

The transfer function of the control behavior of the control device according to Figure 3 is:

&khgr; = G * [ -Kv*s*x + R*(W-X)] Eq. (17)

Λ = ϕ° , Eq. (18)

w l + G*[R + Kv*s]

GR 97 G 4464 DE .. ., ,

The transfer functions of the system and the controller are denoted here by G and R respectively.

The transfer functions in equations (16) and (18) are identical if the D component in Rl in equation (16), i.e. the D component of the controller 11, is increased by the value Kv/Kp with Kp as the controller gain and the transfer function of the control, i.e. the transfer function R2 of the compensation element 9, is set equal to Kv * s. This is immediately obtained by comparing coefficients and solving the equations thus obtained. With the subtraction element 10, the pre-control has a "braking" effect when the setpoint changes and reduces the risk of overshoot.

On the other hand, according to equation (5), Kv changes sign in the case of strongly damped, aperiodic PT2 systems with a damping d > 1. The D component of the controller transfer function Rl is thus weakened and the feedforward control with the compensation element 9 "accelerates" the control.

Although the feedback structure according to Figure 3 and the feedforward control according to Figure 5 mathematically result in the same transfer functions, the feedforward control in sampling controls has the advantage that changes in the reference variable w(s) become effective without delay as a change in the manipulated variable y(s) via the compensation element 9.

The new device for designing a control system for oscillating systems can also be used to design cascade control systems. Figure 6 shows a cascade control system. A manual control variable yh for manual operation, a reference variable w and a controlled variable x2 are fed to a master controller 14 with a transfer function R4. The master controller 14 supplies a control variable y2 for a subordinate control loop. The control variable y2 is in the

GR 97 G 44 64 EN

subordinate control loop is connected to a controller 15 with a transfer function R3, which forms a control variable yl for a system 16 with the transfer function Gl from the manipulated variable y2 and a controlled variable xl. The system 16, on which the controlled variable xl is measured, is followed by a system 17 with a transfer function G2, on which the controlled variable x2 is finally recorded.

A comparatively complex way to determine the controller parameters in such a cascade control is the following iterative procedure:

1st step:

The transient response for the underlying control loop is recorded. A transfer function Gl is determined from the measured and stored time profiles of yl(t) and xl(t), which models the behavior of the system 16. The parameters of the transfer function R3 of the slave controller 15 are determined using this model. 20

Step 2:

The transient response for the master control loop is recorded. From the time profiles of the manipulated variable y2(t) and the controlled variable x2(t), a model is determined for the path of the master controller 14 formed from the subordinate control loop and the path 17. Using this model, the parameters in the transfer function R4 of the master controller 14 are calculated.

30 Step 3:

Depending on the dynamics of the system 17, the parameters of the controller 15 may need to be corrected empirically if no system model for the master control loop or no acceptable parameters of the master controller 14 can be determined with the selected setting of the controller 15. In this case,

GR 97 G 4464 EN

&bull; &bull; ti * · · *

the second step must be repeated. The process is only completed when the design of the master controller 14 also produces acceptable setting parameters.

With the new control design facility, the controller design process can be simplified and carried out with less effort in the following way:

1st step:

The control is initiated and the transient processes of the relevant variables in the control loop are recorded. To do this, the master controller 14 is put into manual operation and, as shown in Figure 6, a change in the manipulated variable is applied via the manual value yh of the controller. The "manual operation" operating mode is characterized by the fact that the manipulated variable output by a controller follows a specified manual value regardless of the control difference currently present at the controller. In automatic operation, however, the manipulated variable is calculated from the control difference present in accordance with the controller transfer function. Another way of inducing the control to undergo transient processes would therefore be to leave both controllers in automatic operation and to change the setpoint w of the master controller 14. The excitation leads to a transient process in the course of the manipulated variables yl(t) and y2(t) as well as the controlled variables xl(t) and x2(t). Therefore, no separate transient response process needs to be recorded for modeling the system 16 and for modeling the system formed from the subordinate control loop and the system 17 for the master controller 14.

Step 2:

To identify the sections 16 and 17, the transfer functions Gl and G2 of the models of sections 16 and 17 are calculated from the stored transient processes according to the general formulas

GR 97 G 44 64 EN

Gl(S) = Fl[Xl(S), yl(s)] G2(S) = F2[x2(s), xl(s)]

and

Eq. (19)

calculated.

Step 3:

Depending on the transfer function Gl of the identified path 16, the controller 15 is designed, for example, according to the optimum magnitude.

Step 4:

Using the previously defined transfer functions R3, Gl and G2, a model of the system is calculated for the master controller. The model can be calculated in two stages. First, an ideal model is determined using the transfer function of the subordinate control loop and the transfer function G2 of the system 17 as:

y ₂ (s) I+ Rl* Eq

G2

Eq. (20)

This possibly complex model with the transfer function G _F (s) is then approximated by a PT3 equivalent model of an oscillatory system with a transfer function G _E (s) according to the following equation:

y ₂ (s) (l + IdTs + T ² s ²

Eq.

Step 5:

With the new facility for designing a control system, a compensation element is determined for this PT3 model to generate a compensation signal that is superimposed on the manipulated variable y2, such that the control signal supplemented by the compensation element from the underlying control loop and the system 17

GR 97 G 4464 EN _§

formed system has an aperiodic transient response. The controller 14 is designed to control the system supplemented by the compensation element according to a design method for aperiodic systems.
5

Compared to the iterative method described above, the method explained with the new facility for designing a control system has the advantage that a transient response is only recorded once and there is no need to readjust the underlying control loop.

The new facility for designing a control system and the new control device obtained with it can also be used advantageously in a cascade control system in which the underlying control loop has a variable dead time.

If the system 16 of the subordinate control loop contains a variable dead time that depends on the operating point, the controller 15 can initially be designed for the dead time that occurs in the nominal range of the control. If the control leaves this range, the dead time of the system 16 changes and the subordinate control loop either becomes too sluggish or begins to oscillate. For the ranges outside the nominal range, aperiodic or oscillatory replacement models, for example with PT3 behavior, must be determined for the master controller 14, which differ primarily in the damping d. With a compensation element that has a gain Kv that can be set depending on the respective damping d, the control can be stabilized over a larger operating range and the control behavior can be improved. Based on the period duration and the amplitude ratio of the transient response after a setpoint step on the cascade control, approximate values for the damping and time constant of a PT2 model for the design of the master controller can be determined. With the different values of the damping d for the different

GR 97 G 4464 EN

The parameters of the master controller 14 and of any compensation element required are determined for the control operating ranges.

Claims (9)

GR 97 G 4464 DE .·\.·\ .: .·*. .·\ Protection claims 1. Device for designing a control system for an oscillating system (6) with a controller (4) which calculates a control variable (y') in a closed control loop from a control difference (xd) which is formed from a reference variable (w) and a controlled variable (x) measured on the system (6), characterized in that a compensation element (7) for generating a compensation signal which is superimposed on the control variable (y') of the controller (4) is determined using a parametric system model, such that the system (6) supplemented by the compensation element (7) has an aperiodic transient response, and that the controller (4) for controlling the system (6) supplemented by the compensation element (7) is designed according to a design procedure for aperiodic systems. 2. Device according to claim 1, characterized in that the controlled variable (x) is superimposed on the manipulated variable (y') of the controller (4) via a compensation element (7) in a negative feedback structure, the compensation element (7) being a differentiating element with at least approximately the transfer function Kv * s with s - Laplace operator and Kv - gain factor, which is determined in such a way that in the loop thus formed from the path (6) and the negative feedback, at least approximately the damping of a path with a predeterminable aperiodic behavior, which has the same order as the oscillating path to be controlled (6) is obtained. GR 97 G 4464 EN 3. Device according to claim 1, characterized in that that to determine the compensation element (7) a gain factor Kv of a differentiating element with the transfer function Kv * s in the Laplace range with Kv - gain factor, s - Laplace operator is determined, over which the controlled variable (x) would have to be superimposed on the manipulated variable (y) in a negative feedback structure in order to obtain at least approximately the damping of a system with a predeterminable aperiodic behavior, which has the same order as the oscillating system (6) to be controlled, in the loop thus formed from the system (6) and the negative feedback,
that the compensation element (9) has at least approximately the transfer function Kv * s in the Laplace range, that the reference variable (w) is superimposed in the manner of a feedforward control via the compensation element (9) with the opposite sign of the manipulated variable (y') of the controller (11) and that the value Kv/Kp is added to the derivative time Tv of a PD or PID controller optimized for the system with predefinable aperiodic behavior, where Kp is the gain of the controller (11). 4. Control device for an oscillating system (6) with a controller (4) which calculates a control variable (y') in a closed control loop from a control difference (xd) which is formed from a reference variable (w) and a controlled variable (x) measured on the system (6), characterized in that the controlled variable (x) is superimposed on the control variable (y') of the controller (4) via a compensation element (7) in a negative feedback structure, the compensation element being a differentiating element with at least approximately GR 97 G 4464 EN approximation of the transfer function Kv * s is with s - Laplace operator and Kv - gain factor, which is determined in such a way that in the loop thus formed from the system and the negative feedback, at least approximately the damping of a system with a predeterminable aperiodic behavior, which has the same order as the oscillating system to be controlled, is obtained. 5. Control device for an oscillating system (12) with a controller (11) which calculates a control variable (y') in a closed control loop from a control difference (xd) which is formed from a reference variable (w) and a controlled variable (x) measured on the system (12), characterized in that that the reference variable (w) is superimposed on the manipulated variable (y') of the controller in the manner of a pre-control via a compensation element (9) with a negative sign, whereby the compensation element (9) in the Laplace range has at least approximately the transfer function Kv * s with Kv - gain factor and s - Laplace operator, that a value Kv/Kp with Kp - controller gain is added to the lead time Tv of a PD or PID controller optimized for a system with a predeterminable aperiodic behavior and that the amplification factor Kv is determined in such a way that in a negative feedback structure the controlled variable (x) of the oscillating path (6) would have to be fed back to the manipulated variable (y) via the compensation element (7) in order to at least approximately obtain the damping of the path with the predeterminable aperiodic behavior in the loop thus formed from the path (6) and the negative feedback. GR 97 G 44 64 EN 6. Control device according to claim 4 or 5, characterized in that the controller (4) has a disturbance variable feedforward at its output, by means of which the compensation signal formed by the compensation element is superimposed on the manipulated variable (y'). 7. Control device according to one of claims 4 to 6, characterized in that the controller (4) is designed according to the optimum amount for the system with a predeterminable aperiodic behavior. 8. Control device according to one of claims 4 to 7, characterized in that the compensation element (7) as a real lead element with the transfer function ^w l + Tz*s is carried out, where the delay time Tz is: 1/2 * Kv > Tz > 1/10 * Kv. 9. Control device according to one of claims 4 to 8, characterized in that the gain factor Kv is variable in dependence on the current operating point of the control.