JPH01315801A - plant control equipment - Google Patents

Abstract

PURPOSE:To optimally keep the gain of a feedback control by assuming a status quantity including a time lag up to detection with a status quantity feedback and a control output by a control object model and executing a feedback control. CONSTITUTION:In a model estimating device 100, a model is corrected so as to satisfy Y-Y 0, namely, Y Y and a feedback control is executed so as to satisfy Y 0. For this reason, when a stationary deviation is included in the model and the deviation is present at Y, it takes time until it is converged to deviation '0', and then, the compensation is executed so that the deviation of a detecting value Y can be '0' with a stationary deviation compensating device 200. Thus, the time necessary to execute the applied correction of a model and make the control deviation into '0' can be made shortest. Thus, the time lag exists from the generation of the status quantity up to the detection and the optimumness of the control gain of the control system for the system to change the dynamic characteristic of a control object 2 can be always held.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to a plant control device, and particularly to a plant control device suitable for controlling a system in which there is a non-negligible time delay from generation to detection of a state quantity. It is related to the device.

[Conventional technology]

As a control method for large-scale systems, a blocked non-interference optimal control method has been proposed, which divides the system into several sub-blocks and controls the system while avoiding interference between each block.

For example, when applying this block non-interference optimal control method to each stand of a tandem rolling mill in order to achieve non-interference and optimal control between the stands, ■ there is a distance from the rolling position to the plate thickness detection position. , There is a time delay between the generation of the state quantity (plate thickness) and detection, and the plate thickness directly below the rolling roll cannot be determined. - The dynamic characteristics of the controlled object change due to acceleration and deceleration, and at a certain speed, the optimum gain may not be achieved. However, there was a problem that the speed was no longer optimal when the speed changed.

This problem is not limited to tandem rolling mills, but is common to systems in which there is a time delay from the generation of the state quantity to its detection, and the dynamic characteristics of the controlled object fluctuate.

Conventionally, this problem has been addressed using observer control methods or model-based control methods, but these have not provided a fundamental solution.

Incidentally, as an example of this type of conventional technology, there is
-131103 etc.

[Problem to be solved by the invention]

An example of the basic configuration of a conventional plant control device is shown in FIG. FIG. 4(A) shows the observer control method, and FIG. 4(B) shows the model reference type control method.

In these figures, X is the state quantity vector actually obtained from control measure 2, X is the state estimation quantity vector obtained from the controlled object model 10, and Y is the detected quantity of the detection target parameter obtained by converting the state quantity. vector, Y is the detection estimator vector obtained from the controlled object model 10 side, C
is the detection matrix 4.12 used for the conversion.

Conventionally, the deviation Y between the estimated value Y and the detected value Y by model 1o
By adding -Y to the input of only the controlled object model 10 in the case of observer control, and to the input of the controlled object 2 and the controlled object model 10 in the case of model reference control,
An attempt was made to match the dynamic characteristics of the controlled object 2 and the dynamic characteristics of the controlled object model 10.

In the case of observer control, the state estimator vector X becomes equal to , there was a possibility that the control system would collapse. On the other hand, in the case of model-based control, a control output is output to the controlled object so that the dynamic characteristics of the controlled object 2 match the dynamic characteristics of the controlled object model 10. Therefore, the controlled object model 1o
When there is a large deviation from the actual value, a large control output may result, and in most cases, a limiter or the like is applied to the control output. Therefore, the dynamic characteristics of the controlled object 2 and the dynamic characteristics of the controlled object model 10 are It is difficult to match.

This was not a solution to the above problem.

An object of the present invention is to keep the estimation error of the state quantity to a minimum even when the dynamic characteristics of the controlled object change, and to perform feedback control or feedforward control in addition to the feedback control in accordance with the change in the dynamic characteristics of the controlled object. It is an object of the present invention to provide a plant control device that can maintain optimum gain.

[Means to solve the problem]

The above object is achieved by the following means.

(2) Estimate the state quantity including a time delay until detection using the controlled object model and the state quantity feedback and control output, and execute feedback control.

■ Feedback control gain and feedforward control gain for observable disturbances that are calculated using the influence coefficient by adaptively modifying the influence coefficient of the controlled object model using the deviation between the estimated value and the detected value that takes time delay into account By adapting and modifying.

In addition to matching the controlled object model with the dynamic characteristics of the controlled object, the control system is also modified to optimize the control gain.

■ When there is a steady-state deviation in the controlled model.

A control error occurs, but in order to quickly remove this control error, a steady error compensator using a detected value is provided.

That is, in order to achieve the above object, the present invention has a model of a controlled object of a plant, detects the actual state quantity of the controlled object with a time delay from its occurrence, and detects the time-delayed state quantity using the model. In a plant control device that performs feedback control of the plant based on the deviation between the estimated value with a time delay and the detected value of the state quantity, The present invention proposes a plant control system that includes a model estimator that adaptively modifies the controlled object model itself, and means that modifies the gain of the feedback control based on the output from the adaptively modified model.

Further, a feedforward control means for an observable disturbance to the controlled object and a means for correcting the gain of the feedforward control using an output from an adaptively corrected model may be provided together.

In either case, it is possible to provide means for compensating for steady-state deviations included in the model based on detected values with a time delay.

By using the above means, the model to be controlled and the control system can be adaptively modified and controlled using estimated values.
Even when the dynamic characteristics of the controlled object change, it is possible to construct an optimal control system that is stable against disturbances and does not leave any steady-state deviation.

[Effect]

FIG. 1 is a diagram showing an example of the basic system configuration of a plant control device of the present invention.

The present invention uses a controlled object model to estimate a state estimation amount X* for a detected amount Y including a time delay until detection.

The state quantity X does not include the time delay until detection X*
However, let us assume that X=(X)k, X)k). At this time, Y=X*=
C(X*,X*)...(1) The detection matrix C is determined in advance.

X is delayed in consideration of time delay, and is set as a detection estimate Y, which is compared with the actual detection value Y, and the influence coefficient of the model itself is adaptively corrected by a method such as model identification.

At this time, if a constant term is added to the model, the stationary deviation of the model can also be removed.

The control gain of feedback control using state quantity feedback and feedforward control for observable disturbances is determined using the influence coefficient of the model, but by adaptively modifying the control gain according to the modification of the influence coefficient, control The system can be optimized.

The model estimator performs the above functions. In the model estimator, the model is corrected so that Y-Y→0, that is, Y-+Y, and feedback control is executed so that Y→0. Therefore, if the model includes a steady-state error, if there is a deviation in Y, the deviation O
It takes time to converge. Therefore, a steady-state deviation compensator is used to compensate for the deviation of the detected value Y to be O. This makes it possible to minimize the time required to adaptively correct the model and to set the control deviation to O.

〔Example〕

As an embodiment of the present invention, a case where the present invention is applied to automatic plate thickness control of a tandem rolling mill will be described.

In each stand of a tandem rolling mill, there is a distance from directly below the roll where rolling is actually performed to the inlet or outlet thickness gauge, which is the plate thickness detector.

For example, there is a time delay before the plate that was directly under the roll reaches the outlet thickness gauge. Rolling is done at low speed (10~2Q
After passing the board between the rolls at a high speed (LOOO
mpm), maintains a constant speed, and finally decelerates to finish by pulling out the plate from between the rolls at low speed. When the rolling speed changes, the friction coefficient changes and the dynamic characteristics of the system fluctuate. For this reason, the control gains of feedback control and feedforward control, which are determined by the dynamic characteristics of the system, deviate from the Hatake optimum values, resulting in poor control performance.

The present invention is used to correct this deviation and maintain an optimal control gain.

Figure 3 shows an overview of rolling at each stand. Rolling is performed by passing the material to be rolled between rolls. In the figure, ■( represents the entrance side plate thickness, h represents the exit side plate thickness, P represents the rolling load, S represents the roll gap, T□ represents the exit side tension, and Tb represents the input side tension. In the following explanation, The deviation from the set value is represented by the symbol Δ.

FIG. 2 is a block diagram of each stand control device to which the present invention is applied. Modeling the exit side plate thickness deviation △h, (1)
The estimated value Δh* is calculated by the controlled object model (state estimator) 120 using the formula.

Δh*=a□ΔS+a2ΔP+a, ΔH+a4ΔTb+
a, ΔT1+e ... (2) where, ΔS: Roll gap deviation ΔH: Rolling load deviation ΔTb: Inlet tension deviation ΔTi: Output tension deviation a~a,: Influence coefficient e: Estimated steady deviation term plate thickness Using 66 deviations, the feedback control output Utb is: Utb= ((M+Q)/M)66...(3)
Calculate and execute feedback control.

Here, M is Mill's constant and Q is a plasticity constant.

Estimated board thickness Δh - The estimated board thickness (
Δh*). and detected value (Δh). Which difference Δh=(Δh*)
. −(△h).

=Δh*e−TS−Δhe−TS ・= (4) and (Δb books). ΔS used to calculate.

The model identifier 110 identifies the controlled object model 120 using ΔP, ΔH1ΔT b , and ΔTt, and a□
Adaptively modify ~a5 and e. Model identification methods include 1, for example, RL S (recursive 1ea
There are various methods such as (stsquares), but since there are many documents regarding these methods, their explanation will be omitted here.

In the model identifier 110, a1 to a, and e are adaptively modified, but in the rolling mill.

△h= ((M/ (M×Q)'j△S...(
5)△h= ((Q/ (M+Q))△H...(6)
There is a relationship, a1. = M / (M, + Q)
...(7) a, = Q / (M,
+Q)...(8) There is a relationship between the influence coefficient and the influence coefficient.

Therefore, the control gain in feedback control is (
From equation 2), it becomes 1/a, so Utb= (1/
a1) 66 lines - By calculating according to (9), it becomes possible to adaptively correct the control gain. Also when the dynamic characteristics of the controlled object 2 change.

Optimum gain can be maintained.

Feedforward control is possible by detecting the entrance plate thickness deviation ΔH with an entrance plate thickness gauge and tracking it to just below the rolling mill, but the control output Uii in this case is Uzz= (Q/M)AH-(io ) is calculated by. Therefore, Uxt=(ax/at)ΔH−(11) allows an adaptive technique for the feedforward control gain.

The estimated values Δh may include steady-state deviations. For example, if there are 80 offset errors in ΔS, Δh books = Δh *o + a□s, ・=
(12) However, the Δh tree is the true value of the plate thickness deviation. Here, in order to cancel alS, model 12
A 0 term is included in 0, and this e is.

Correct so that a, So+e=O-(13). Feedback control is performed using a Δh tree, and the modification of the model 120 is performed using Δh*→
Since Δh becomes Δh→0 in the end, however, in the case of Δhf−O, the time until Δh=o is determined by the convergence time of model identification. Therefore, in order to remove the steady-state error as quickly as possible, a steady-state error compensator 200 using an integrator is used to remove (steady-state error) → 0.
Control so that

An example in which the present invention is applied to automatic plate thickness control in a rolling mill has been described above. Here, we have considered a control system with one human power and one output, but the same method can be applied even when performing multivariable optimal control with multiple inputs and multiple outputs. In the case of conventional multivariable optimal control, the control gain is calculated by a model, so if the dynamic characteristics of the model do not match the dynamic characteristics of the controlled object, the desired control performance cannot be obtained. By using this method to adaptively control the dynamic characteristics of the model, the control gain is optimized, and even when the dynamic characteristics of the controlled object change, optimal control is always performed.

The present invention can be applied not only to tandem rolling mills, but also to all systems in which there is a time delay between the generation and detection of a state quantity, and the dynamic characteristics of the controlled object change.

〔Effect of the invention〕

According to the present invention, it is possible to always maintain the optimality of the control gain of the control system for a system in which there is a time delay between the generation and detection of the state quantity and the dynamic characteristics of the controlled object change.
Control deviation can be reduced and a stable control system can be constructed.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing an example of the basic configuration of a plant control device according to the present invention, FIG. 2 is a diagram showing the configuration of a specific embodiment in which the device shown in FIG. This figure is a diagram showing an outline of a rolling mill, and FIG. 4 is a diagram showing an example of a system configuration of a conventional observer type control device and a model reference type control device. 2... Controlled object, 4, 12... Detection matrix, 8, 14
... Delay time calculator, 10... Conventional controlled object model, 16.20... Adder, 18... Feedback control circuit, 22... Feedforward control circuit,
1. OO... Model estimator, 110... Model identifier, 120... Controlled object model (state estimator) of the present invention, 180... Variable gain feedback control circuit, 200... Steady-state error compensator , 220... variable gain feedforward control circuit.

Claims (1)

[Claims] 1. A model of a controlled object of a plant is provided, and the state quantity of the actual controlled object is detected with a time delay from its occurrence, and the time-delayed state quantity is estimated by the model, and the state quantity is In a plant control device that performs feedback control of the plant based on a deviation between an estimated value with a time delay and a detected value, the control target model itself is controlled based on a deviation between an estimated value with a time delay and a detected value. A plant control device comprising: a model estimator for adaptively correcting the model; and means for correcting the gain of the feedback control based on an output from the adaptively corrected model. 2. The plant control device according to claim 1, further comprising: feedforward control means for an observable disturbance to the controlled object; and means for modifying the gain of the feedforward control based on the output from the adaptively modified model. A plant control device characterized by: 3. The plant control device according to claim 1 or 2, further comprising means for compensating for a steady-state deviation included in the model based on the detected value with the time delay.