JPH0822306A - Automatic adjusting device for arithmetic control parameter - Google Patents

Abstract

PURPOSE:To make an automatic adjusting device applicable to an open loop system and a closed loop system, and further a cascade control system and to adjust an arithmetic control parameter in a short time during the automatic operation of a plant. CONSTITUTION:This device is equipped with a plant identification part 16 which identifies a plant 14, inputting the output of a PID controller 12 inputting the deviation 11' between a target value 11 and the output 15 of the plant 14, with integral characteristics and an idle time model by using the input 13 and output 15 of the plant 14, and a PID parameter determination part 17 which calculates a proportional gain, an integral time, and a differential time as parameters of the PID controller 12 with the idle time model obtained by the plant identification part 16 and supplies them to the PID controller 12.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a single loop proportional-plus-integral-derivative controller (hereinafter referred to as "PID controller") and cascade PID controller used in various plants such as power plants and scientific plants. The present invention relates to a calculation control parameter automatic adjustment device for determining the control parameter of the above and providing it to the PID control device.

[0002]

2. Description of the Related Art In a conventional PID control parameter automatic adjusting device, there are roughly two types. One is Figure 3
As shown in (a), it is a method of identifying the plant 4 by an open loop and determining the PID parameter.
As shown in (b), this is a method of correcting parameters from the response waveform of the output of the plant 4 in a closed loop.

FIG. 3 (a) shows the PID control parameter automatic adjustment device of the open loop system as described above. In the figure, 2 is a PID control device (indicated as "PID" in the drawing), and 3 is a control target. Input 4 is a plant as a control target to which this input 3 is input, 5 is an output of the plant 4, 6 is a plant identification unit for identifying the plant 4 from the input 3 and the output 5 to the plant 4, and 7 is this A PID parameter determination unit that determines the control parameters required for the PID control device 2 according to the identification of the plant identification unit 6;
Reference numeral 8 is an irregular signal or step signal given to the plant 4.

In this case, the M-sequence signal (maxi
A random signal 8 such as a mum-lenght linear shift register sequence) is provided as an input 3, and the plant identification unit 6 identifies the plant 4 using an ARMA model obtained from the input signal and the output of the plant 4, or to the plant 4. There is a method of inputting the step signal 8 and identifying the plant 4 from the response waveform into a model such as a first-order lag + dead time, and the PID parameter determining unit 7 uses the identified model to adjust the PID parameter according to the Zigler-Nichols adjustment rule or the like. Is determined, and the determined PID parameter is provided to the PID control device 2. However, when the plant 4 is an atypical system, it is identified by the model of integral characteristic + dead time.

Further, FIG. 3 (b) shows a closed-loop type PID control parameter automatic adjusting device, in which 1 is a target value and 9 is a plant response waveform evaluated from an output 5 of a plant 4. The evaluation result is replaced with the identification result of the plant identification unit 6 instead of the PID parameter determination unit 7
It is a plant response waveform evaluation unit for sending to.

In the case of this apparatus, the plant response waveform evaluation unit 9 is used so that the response is improved from the output response waveform of the plant 4 due to the change in the target value or the disturbance using the response time, the damping rate, the settling time, etc. as an index. To modify the PID parameter.

The M-system signal is a pseudo-white signal as shown in FIG. 4 (a), which is formed from a binarized signal, and has ARMA (autoregressive moving averag).
e) The model is a model that represents the response of the object in the form of the following equation. Ie

[0008]

[Equation 1]

(However, y: output, u: input, k: sampling time.) The "first-order delay + dead time" model is a model in which the dynamic characteristics of the controlled object are expressed by the following equation. That is, G (s) = Ke- ^Ls / (1 + Ts) (2) (where K: gain, T: time constant, L: dead time, s: Laplace operator), where: When the step input is 1, the output response is as shown in FIG.

Further, the "integral characteristic + dead time" model is a model in which the dynamic characteristic of the controlled object is expressed by the following equation. That is, G (s) = Ke- ^Ls / s (3) (where, K: gain, T: time constant, L: dead time, s: Laplace operator.) Where 1 is a step input, The response is Figure 4.
As shown in (c).

The Ziegler-Nichols adjustment rule is the control parameters KP, TI, and
This is one of the adjustment rules for determining TD, and K, T, and L are used for the controlled object of the dynamic characteristics of the first-order delay and the dead time.
K for the controlled object of the integral characteristic + dead time dynamic characteristic,
KP, TI, and TD are uniquely obtained from L.
That is, K (s) = KP (1 + 1 / TIS + TDS) (4)

[0012]

In the conventional apparatus of the system shown in FIG. 3 (a), since the plant 4 is identified by the open loop, automatic operation cannot be performed during operation of the plant 4 and manual operation is not possible. However, there is a problem in that a necessary signal must be input to the plant 4 as needed.

Further, in the conventional apparatus of the system shown in FIG. 3 (b), the number of trials for correcting the PID parameter increases, and the adjustment time becomes long. Further, both of the above-mentioned systems have a problem that, for a cascade control system having two PID control devices, adjustment of these two PID control devices cannot be performed at the same time.

The present invention has been made in view of the above circumstances, and an object thereof is an open loop,
An object of the present invention is to provide an arithmetic control parameter automatic adjustment device which can be applied to a cascade control system regardless of the closed loop and can be adjusted in a short time during automatic operation of a plant.

[0015]

[Means and Actions for Solving the Problems] That is, the present invention is as follows: (1) PI for inputting a deviation between a target value and an output of a plant
A plant identification means for identifying the plant by an integral characteristic and a dead time model by the input and output of the plant to which the output from the D control device is input, and the PID control device of the PID control device by the dead time model obtained by the plant identification means. It is provided with a parameter determining means for calculating the proportional gain, the integral time, and the differential time which are parameters and providing them to the PID control device.

With such a configuration, it becomes possible to make adjustments in a short time during automatic operation of the plant regardless of whether it is an open loop or a closed loop. (2) PI that inputs the deviation between the target value and the plant output
In the cascade-type control system having two PID control devices that use the output from the D control device as a target value and the output from the PID control device that inputs the deviation between the target value and the output of the plant to the plant, The plant input and output of the plant of the loop are input to identify the plant by the first-order lag + dead-time model or integral characteristic + dead-time model, and the model is derived from the first plant identification means. The first parameter determining means for inputting the proportional gain, the integral time, and the derivative time, which are the parameters of the proportional-integral-derivative control, and providing them to the PID controller, the output from the PID controller in the outer loop, and the plant Second plant identification means for identifying a system including an internal loop with an output as an input by a first-order lag + dead-time model; The model is input from the runt identifying means, and a second parameter determining means for calculating a proportional gain, an integral time, and a differential time, which are parameters of the proportional-integral-derivative control, and outputting them to the PID controller is provided. is there.

With such a configuration, even in a cascade type control system having two PID control devices, adjustment can be performed in a short time during automatic operation of the plant.

[0018]

An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 shows the basic configuration concept thereof. 11 is a target value, 12 is a PID control device for inputting a deviation 11 'between the target value 11 and an output 15 which will be described later, 14 is a plant 14 or 13 as a control target. Is an input to the plant 14 obtained by the PID controller 12, and 15 is the plant 1
4 output.

Here, a plant identification for inputting the input 13 and the output 15 of the plant 14 to automatically identify the plant 14 by the integral characteristic + dead-time model is provided to form a single-loop feedback control system. Part 16,
The gain K of the model and the dead time L are input from the plant identification unit 16 to calculate the proportional gain KP, the integration time TI, and the derivative time TD which are the parameters of the PID control, and the PID parameter is output to the PID control device 12. The determining unit 17 is provided.

In the above configuration, the plant identifying section 16 identifies the plant 14 by the integral characteristic + dead time model because the response of the process system can be approximated by this model when the response is an atypical system. Since this model can be approximated by the following equation (5), the above model can be expressed by the gain K and the dead time L that constitute this equation (5). That is, G (s) = Ke- ^Ls / s (5), and the plant identification unit 16 obtains the gain K and the dead time L in the equation (5). The procedure for calculating the gain K and the dead time L by the plant identification unit 16 will be described below.

For simplification, the initial value is set to zero in the response calculation from now on. However, it is assumed that the output is not a disturbance to the plant, but the output fluctuates due to fluctuations in the input such as changes in set values. Therefore, the change of the target value 11 need not be a step.

Since the input / output 13 and 15 of the plant 14 input by the plant identification unit 16 are sampling data, they are discretized by the Z-transform with zero-order hold, where the sub-time is Δt. That is,

[0023]

[Equation 2] It is assumed that. Here, the output y is calculated from the equation (6).
Is represented by the following difference equation (8). That is, y (k) = y (k−1) + K · Δt · u (k−1−1) (8) As the estimated value Y (k), the deviation e from the actual output y (k)
Taking (k) gives e (k) = y (k) -Y (k) (9). Regarding this deviation e (k), it is necessary to find K and L such that this becomes the minimum, but in order to calculate this by the least squares method, the sum of squares of the above equation (9) is taken as E. ,

[0024]

(Equation 3) Becomes Differentiating this by K and setting it as D,

[0025]

[Equation 4] By solving this equation (7) for K, K
Can be requested. That is,

[0026]

(Equation 5) It will be.

Regarding the dead time, L = l in the above equation (7)
As Δt (l: positive number), the gain K is obtained for the certain l by the above method, and the gain K is used to calculate an evaluation function J (l) represented by the following equation (13).
The gain K and the dead time L that minimize (1) are adopted. That is,

[0028]

(Equation 6)

The gain K and the dead time L obtained by the plant identification unit 16 are input to the PID parameter determination unit 17. The PID parameter determining unit 17 determines the proportional gain KP, the integration time TI, and the differential time TD which are PID parameters according to the above-mentioned Ziggler-Nichols adjustment rule and inputs them to the PID control device 12, and inputs them to the PID controller 12.
Have the parameters modified.

Although the plant identification section 16 has been described as an open loop identification, identification can be performed even in a closed loop as long as the inputs and outputs 13, 15 of the plant 14 do not change as long as identification is performed in the above manner. It is possible. However, in the case of this closed loop, plant 1
There should be no disturbance to 4 and the sampling time should be constant.

Next, another configuration example according to one embodiment of the present invention is shown in FIG. In addition to the configuration of FIG. 1 described above, FIG. 2 further includes a second PID control device 18 after the PID control device 12, and a second plant 20 between the PID control device 18 and the plant 14. The above-mentioned cascade type feedback control system is formed.

That is, the deviation 11 'between the target value 11 and the output 15 of the plant 14 is input to the PID controller 12, and the deviation 13' between the input 13 from the PID controller 12 and the output 21 of the plant 20 is calculated. It is input to the second PID controller 18.

Further, the output of the PID controller 18 is supplied to the second plant 20 as an input 19, and the output 21 of the plant 20 is supplied to the PID controller 12 as described above.
The deviation 13 'is obtained by subtracting from the input 13 days obtained in (1) and is also supplied to the plant 14 as an input.

However, the plant identifying section 16 uses the inputs 13 and 19 and the outputs 21 and 15 to determine the plant 2
0 and the dynamic characteristics from the input 13 to the output 15 are automatically identified by a first-order delay + dead time model or an integral characteristic + dead-time model. If the model used for this identification is a first-order delay + dead time model. Gain K, time constant T,
The dead time L, the gain K and the dead time L in the case of the integral characteristic + dead time model are sent to the PID parameter determination unit 17.

The PID parameter determination section 17 calculates the proportional gain KP, the integration time TI, and the differential time TD which are control parameters by the input from the plant identification section 16, and outputs them to the PID control devices 12 and 18.

In the above-mentioned configuration, the operation when the plant identification unit 16 identifies the plant by the integral characteristic + dead time model is as shown in FIG. 1 above, and therefore its explanation is omitted here. Here, the operation when the plant identification unit 16 identifies the plant by the first-order delay + dead time model will be described.

The first-order lag + dead time model is as follows.
It can be expressed by a gain K, a time constant T, and a dead time L as shown in equation (4). That is, G (s) = Ke- ^Ls / (1 + Ts) (14) The procedure for calculating K, T, and L in the equation (14) will be described below. However, there is no disturbance to the plant, and the output fluctuates due to input fluctuations such as changes in set values. Further, for simplicity, the initial value is set to zero in the subsequent response calculation.

Inputs / outputs 19 and 21 of the plant 20 and the PID control device 1 input to the plant identification section 16
Since the input 13 that is the output of 2 and the output 15 of the plant 14 are sampling data, the sampling time is Δt.
As a result, it is discretized by Z-transform with zero-order hold. That is,

[0039]

(Equation 7)

From the equation (15), the output y is represented by the following difference equation (17). That is, y (k) = Py (k−1) + K (1-P) u (k−1−1) (17) As the estimated value Y (k), the deviation from the actual output y (k) Taking e (k), e (k) = y (k) −Y (k) (18). With respect to this deviation e (k), it is necessary to find K, T, and L that minimize, but in order to calculate this by the first-square method, the sum of squares of the equation (18) is taken as E. . That is,

[0041]

(Equation 8) Differentiating this by P and K and setting it as D,

[0042]

[Equation 9] Here, the elements of the respective terms of the above equations (20) and (21) are represented by a1, a2, a3, a4, a5 as shown in the following equation (22).
And That is,

[0043]

[Equation 10]

By substituting the elements of the above equations (20) and (21) with the above equation (22), the following equations (23) and (2)
4) is obtained. That, a1 + a2 K + a3 P + a4 K 2 (1-P) = 0 ... (23) -a2 + a4 P + a5 K (1-P) = 0 ... (24) above simultaneous equations (23), by solving (24),
K and P shown by the following equations (25) and (26) are obtained. That is, K = (a1 a4 -a2 a3) / (a1 a5 -a3 a5 + a4 ² -a2 a4) (25) P = (a2 -a5 K) / (a4 -a5 K) (26) This (26) By using the equation), the time constant T can be calculated from the following equation (27) obtained by converting the equation (16). That is, T = −Δt / log _e P (27) For the dead time, L = lΔt (l = a positive number)
For a certain l, T and K are obtained by the above method, the evaluation function J (l) represented by the following equation (28) is calculated using the K and T, and K and T that minimize this J (l) are obtained. , L are adopted. That is,

[0045]

[Equation 11]

If the first-order lag + dead time model obtained by the plant identification unit 16 is K, T, L, and if the integral characteristic + dead time model, K, L are input to the PID parameter determination unit 17, and the PID is determined. The parameter determining unit 17 uses the Zigler-Nichols adjustment rule or the like to determine the proportional gain KP, the integration time TI, and the derivative time T, which are PID parameters.
D is determined and sent to the PID controller 12 and the PID controller 18, and the PID parameters are corrected.
Thereby, by changing the target value 11 for example, P
The PID parameters of the ID controller 12 and the PID controller 18 can be simultaneously modified.

[0047]

As described above, according to the present invention, it is possible to apply not only open loop and closed loop but also to a cascade control system, and it is possible to adjust in a short time during automatic operation of a plant. A parameter automatic adjustment device can be provided.

[Brief description of drawings]

FIG. 1 is a block diagram showing a basic configuration concept according to an embodiment of the present invention.

FIG. 2 is a block diagram illustrating a configuration concept when applied to the cascade control system according to the embodiment.

FIG. 3 is a block diagram showing a basic configuration concept of a conventional arithmetic control parameter automatic adjustment device.

4 is a diagram illustrating each signal by the apparatus of FIG.

[Explanation of symbols]

1, 11 ... Target value, 2, 12, 18 ... PID control device,
3,13,19 ... input, 4,14,20 ... plant,
5, 15, 21 ... Output, 6, 16 ... Plant identification unit,
7, 17 ... PID parameter determination unit, 11 ', 13' ...
deviation.

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI technical display location G05B 24/02

Claims (2)

[Claims] 1. A plant identification means for identifying the plant with an integral characteristic and a dead time model by the input and output of the plant, which receives an output from a proportional-plus-integral-derivative controller which inputs a deviation between a target value and an output of the plant. , A parameter determining means for calculating the parameters of the proportional-plus-integral-plus-derivative controller by the dead time model obtained by the plant identifying means, the proportional gain, the integral time, and the derivative time, and providing the proportional-integral-derivative controller with the parameter determining means. An automatic control parameter adjusting device characterized by the above. 2. The output from the proportional-plus-integral-derivative controller that inputs the deviation between the target value and the output of the plant is set as the target value, and the deviation from the proportional-integral-derivative controller that inputs the deviation between the target value and the output of the plant is input. In a cascade-type control system having two proportional-integral-derivative controllers that provide the output to the plant, the input and output of the plant in the inner loop are input and the plant is subjected to a first-order delay + dead time model or integral characteristic + dead time model. The first plant identifying means identified by the above, and the model is input from the first plant identifying means to calculate the proportional gain, the integral time, and the derivative time which are the parameters of the proportional integral derivative control, and the proportional integral derivative control is performed. The first parameter determining means provided to the device, the output from the proportional-plus-integral-derivative controller of the outer loop and the output of the plant are used as inputs. The second plant identification means for identifying the system including the loop with the first-order lag + dead-time model, and the above-mentioned model from the second plant identification means are input as the proportional gain and the integral which are the parameters of the proportional-integral-derivative control. An arithmetic control parameter automatic adjusting device comprising: a second parameter determining means for calculating time and differential time and outputting to the proportional-integral-derivative control device.