JPS6144326B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an optimum value calculation method for PID control parameters based on process characteristic test data in a closed-loop control system of a process, and in particular, to This is a simple calculation method that can be easily implemented.

In the past, much research and development has been carried out on optimal value calculation methods for this type of PID control parameters, in which large-scale electronic computers can be used to perform complex calculations on large amounts of data. There has been little research on simple computational processing of the relatively small amount of data obtained.
Its development has long been desired.

On the other hand, regarding the simplification of the test method for process characteristics in a process closed-loop control system, there is a "process test method" described in Japanese Patent Application Laid-open No. 159702/1983, which was separately proposed by the present inventor. The test method makes it possible to easily test the dynamic characteristics of a process in a closed-loop control system under normal operating conditions, thereby obtaining the dynamic characteristics of the process under actual operating conditions.

In general, the calculation formulas used for this type of calculation are relatively simple and are listed in ordinary handbooks, etc., but they only provide values within a rough guideline; It has the disadvantage that it is necessary to separately consider its suitability when using it for PID control.
The calculation formula that can be used for control to perform appropriate control has a complicated structure, and has the disadvantage that it cannot be implemented in a small-scale digital instrumentation system.

The purpose of the present invention is to eliminate the above-mentioned conventional drawbacks and to improve the above-mentioned "process test method" according to a separate proposal.
We have revised the calculation process of the test results to be suitable for calculating the optimum value of PID control parameters, and we have developed an optimum value calculation method for PID control parameters that can obtain calculated values that can be used for actual control with sufficient reliability. It is about providing.

That is, the calculation method of the present invention changes the control input amount of the process in a stepwise manner in the process control system, and changes the characteristic area between the new set value of the control input amount and the transient characteristic curve and the output of the process. Formulas for calculating the characteristic area between the new set value of the force and the transient characteristic curve are constructed based on the approximation formula of the transfer function of the process, and the weighting constants of these formulas are set to the number of process characteristic values. The invention is characterized in that a plurality of arithmetic expressions are created with different values depending on the situation, solutions to these arithmetic expressions are obtained, and PID control parameters of the control system constituting the transfer function are calculated using these solutions. It is something to do.

The present invention will be described in detail below with reference to the drawings.

First, FIG. 1 shows the configuration of a process closed-loop control system testing apparatus using a conventional method. In the illustrated configuration, a changeover switch SW is connected to a process control system 1 consisting of a process section P, an operation section FC that operates the process section P, and an adjustment section C that controls the operation section FC by applying an operation signal CS to the operation section FC. When testing process characteristics, switch SW is switched to break the closed loop of control, and the test signal TS, which is generated by supplying the output amount PS of the process section P to the test device 2, is applied to the operating section FC, and the step shape etc. The characteristics of the process section P are known based on the response of the process section P to the test signal TS having a desired waveform. According to this method, data processing of process characteristic test results is easy, but there is a drawback in that an operator must carefully monitor process fluctuations during actual operation.

On the other hand, in the above-mentioned "process test method" proposed by the present inventor, as shown in FIG. 2, a process control system 1 similar to that shown in FIG.
A test device 2 configured in the same manner as shown in the first diagram is connected, but unlike the normal test method shown in the first diagram, the output amount PS of the process section P is measured without inserting a changeover switch SW into the closed control loop. The test signal TS that is supplied and formed is applied to the regulating section C in the closed control loop, and the operating signal CS that acts on the process section P under normal operating conditions is changed in a stepwise manner. The response of the process section P was tested.

That is, similarly to the normal test method shown in FIG. 1, the output amount PS of the process section P is
Once stored in memory M, it is taken out and supplied to the change detector AD, and the momentary process output
A detection output signal AS that detects a change in is applied to the test signal generator TD for control, and a test signal TS with a waveform that suddenly changes in a step manner is applied to the adjustment section C in the process control system 1 to control the output operation. The signal CS was changed in steps.

Therefore, the operation signal CS from the adjustment section C is
As shown in FIG. 3A, for a new setting value L that changes stepwise by ΔCS, the transient characteristic curve U
(t), and in response to the change in the operation signal CS, the output amount PS of the process section P also changes as shown in the third
As shown in Figure B, the transient characteristic curve
Draw (t) and change. In the "process test method" proposed by the inventor, the process part P
Temporarily store the momentary data of the operation signal CS as a control input in the memory M in the test device 2 together with the momentary data of the process output quantity PS,
New setting value B of operating signal CS and transient characteristic curve U
(t), and the characteristic area shown by diagonal lines in FIG. 3A, and the diagonally shaded area in FIG. The dynamic characteristics of the process section P in the actual operating state can be known by calculating and comparing the characteristic areas shown in FIG. In addition, during such a process characteristic test, the step-shaped test signal can be turned into a pulse-like signal by cutting it when the process output amount reaches a certain ratio of the set value change width, or it can be made into a pulse-like signal as shown in Fig. 4. A large number of process characteristic values were obtained by weighting the characteristic area using a weighting function.

Therefore, the arithmetic processing of test data in the "process test method" proposed by the present inventor uses relatively complex calculation formulas in order to accurately estimate process characteristics by comparing characteristic areas. However, in order to properly control the process section P,
As mentioned above, if the only purpose is to obtain the optimum value of the PID control parameter in accordance with the operation of the process part P in the actual operating state, the transfer function of the process, which is the starting point for calculating the characteristic area described above, can be used. By expressing it with a relatively simple approximate expression, the optimum value of the PID control parameter can be easily calculated through relatively simple arithmetic processing even using a small-scale arithmetic device.

That is, the dynamic characteristics of the process part P can be approximately expressed by the following transfer functions for the localized system and non-localized system processes, respectively.

Localization system Gp (s) = Kp・e ^−Ls / (1+T ₁ S) (1+T
₂ S) (1) Positionless system Gp (s) = e ^{- Ls} / T ₁ S (1 + T ₂ S) (2) Here, Gp is the transfer function, Kp is the gain of the process part P, and L is the process control Dead time, Equation (1)
T ₁ and T ₂ are delay time constants of the process, T ₁ in equation (2) is an integral time constant of the process, and S represents a Laplace transform operator.

Starting from the approximation formula for the transfer function as described above,
Various known methods can be used for the calculation process for determining the optimum value of the PID control parameter, and the calculation method of the present invention does not specify the calculation method.

In order to obtain the constants included in the above equations (1) and (2), the set value of the adjustment section C or the characteristic area for a step change in the output operation signal CS of the adjustment section C is used. In other words, in the calculation method of the present invention, even if the output operation signal CS of the control section C changes in a closed loop state, the area of the shaded area in FIG. In contrast, the following theoretical relationship is utilized. Although there are differences in the calculation formulas for a localization system and a non-localization system, the calculations can be processed using almost the same method.For simplicity, the following describes the case where the localization system process is subjected to differential parallel PID control. This is explained as an example.

Here, Kc is the proportional gain in PID control,
Ti is an integral time constant in PID control, and Td is a differential time constant in PID control.

However, since equations (3) and (4) above only contain the same information regarding the calculation of the constant in equation (1) that represents the transfer function, the weighting that applies weighting to the calculation process of the characteristic area Constant α and other constant β
Instead, the following equation is constructed by performing similar calculations with different weighting.

On the other hand, when weighting is not applied to the calculation process for calculating the characteristic area described above, ΔCS=ΔPS/Kp (7) ∫ ^∞ ₀ {ΔCS+U(o)−U(t)}dt = ΔPS/Kp{Ti/Kp・Kc-(T ₁ +T ₂ +L)
}(8) ∫ ^∞ ₀ {ΔPS+X(o)−X(t)}dt =ΔPS/Kp·Kc (9) The following relational expression is obtained.

Therefore, the values of the four parameters Kp,
There are four relational expressions for L, T ₁ and T ₂ , namely,
(3) or (4), (5) or (6), (8) or (9)
Now that we have obtained equations and equations (7), for example, by mathematical methods such as the Newton-Raphson method,
Unique optimal values can be found for each of the four parameters. Furthermore, equation (3) is a theoretical equation regarding the characteristic area for the control section output CS, and equation (4) is a theoretical equation for the characteristic area for the process section output PS, since they contain only the same information as described above. Conversely, it can be seen that the data obtained as test results are not reasonable unless there is a certain relationship, and the same is true for equations (5) and (6).
This also holds true for equations (8) and (9). In addition, in that case,
The weighting constants α and β are set as α=1/L+T ₁ +T ₂ (10) β=1/2(L+T ₁ +T ₂ ) (11) in consideration of the convergence status of process control and calculation error. Predict and weight.

In addition, in the above calculation process, T ₂ = 0
If we make a simple approximation so that It can be simplified.

There have already been many reports that sufficiently effective optimal adjustment values for PID control can be obtained from the parameters expressing the process characteristics obtained in this way, so there are some that are suitable for the scale of the digital instrumentation system. All you have to do is select and use the method you have chosen.

As is clear from the above description, according to the present invention, based on the results of testing the process characteristics under actual operating conditions for the process control system using the configuration shown in FIG. Since the optimal values of PID control parameters can be determined through relatively simple calculation processing, it is possible to easily test process characteristics without disturbing the operating status of the process, and it is also possible to determine the appropriateness of control parameters for process control. adjustments can be made easily.

In addition, four relational equations were created using relatively easily obtained test data such as the set value of the process output amount and the characteristic area, and since the construction of these relational equations is relatively simple, it is possible to The arithmetic processing can be easily carried out even with the arithmetic device.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of a conventional process characteristic testing device, FIG. 2 is a block diagram showing an example of the configuration of a process characteristic testing device used in the calculation method of the present invention, and FIGS. 3A and B are the same. FIG. 4 is a characteristic curve diagram showing the operating characteristics of each part, and FIG. 4 is a characteristic curve diagram showing the weighting characteristics applied to the operating characteristics of each part. 1... Process control system, 2... Test equipment, C...
...Adjustment section, FC...Operation section, P...Process section,
M...Memory, AD...Change detector, TD...Test signal generator.

Claims (1)

[Scope of Claims] 1. The control input amount of the process in the process control system is changed stepwise, and the characteristic area between the new set value of the control input amount and the transient characteristic curve and the output amount of the process are changed. Formulas for calculating the characteristic area between the new setting value of Create a plurality of arithmetic expressions with different values, find solutions to these arithmetic expressions, and use these solutions to calculate the control system of the control system that configures the transfer function.
An optimal value calculation method for PID control parameters, characterized in that the PID control parameters are calculated. 2. In the PID control parameter optimum value calculation method according to claim 1, the weighting constant of the calculation formula is set based on the predicted process characteristic value. Optimal value calculation method.