JPH04186401A - Method and device for tuning digital control device - Google Patents

Abstract

PURPOSE:To attain auto-tuning without reducing a control response speed by connecting a specific control gain tuning device to a digital control device and tuning the control characteristics of the device. CONSTITUTION:A data I/O means 103 rewrites control response data stored in a control response storing means 102 and feedback control gain. A control gain tuning device 5 inputs the control response data stored in the means 102 by a control response reading means 51 through the means 103 and finds out an evaluation index by a control response evaluating means 522 based upon set control characteristics set up in a control characteristic setting means 53 and the control response data. A control gain correcting means 54 corrects the control gain based upon the evaluation index and then the means 103 corrects the corrected control gain to tune control characteristics to required ones. Thus auto-tuning can be attained without reducing the control response speed.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a tuning method and device for a digital control device, and more particularly to a method and device for setting desired control characteristics by automatic tuning of control gains.

[Conventional technology]

Conventionally, control characteristics in a digital control device were manually set by an inspector while observing the response of the controlled variable.

For this reason, there were problems such as a long time being spent on setting work and individual differences appearing in the tuning results.

To solve this problem, auto-tuning can be used to automatically set the control characteristics to desired characteristics. Japanese Patent Application Laid-Open No. 64-46101 is a method for such auto-tuning.
There is one that targets the PID controller described in No. This is the amount of overshoot from the control variable response waveform when the command value changes.

The period ratio, which is the ratio between the previous value and the current value of the damping coefficient and period, is obtained as an evaluation index of the control response using the controlled variable response waveform recognition means, and the control gain is corrected by fuzzy inference based on these evaluation indexes and adjustment rules. control performance determining means for controlling the control gain modifying means to operate the control gain modifying means when the overshoot amount or attenuation coefficient exceeds a permissible range, that is, when the desired control characteristics are not achieved; has been established. This method applies fuzzy inference and tunes the control gain in the same way as a skilled inspector, so it does not cause any disturbance to the controlled object during tuning, and it is possible to tune even highly nonlinear controlled objects. It has excellent characteristics.

[Problem to be solved by the invention]

Since the above-mentioned conventional technology is aimed at a PID controller, there are the following problems when applying the above-mentioned conventional technology to a digital speed control device for a motor. The control response of the motor speed is several to several degrees, which is hundreds to tens of thousands of times faster than the control response of the object controlled by the PID controller. Therefore, compared to the microprocessor of a PID controller, the microprocessor built into the digital control device performs control calculations at a shorter sampling period of several hundred to tens of thousands of minutes.

Therefore, since the microprocessor does not have enough processing time and it is difficult to evaluate the control response every sampling period and obtain evaluation indicators such as the amount of overshoot, it is impossible to perform auto-tuning processing. Alternatively, if the speed control calculation cycle is lengthened to provide a margin for processing time and auto-tuning processing is performed, there is a problem that the required high-speed speed control response cannot be obtained. Furthermore, since the constants of the motor, load, etc. do not change significantly during operating conditions, it is sufficient to tune the control characteristics once at the beginning, and conventionally this is done when the digital control device is installed. For this reason, when the auto-tuning means is built into the digital control device, there is a problem in that it has functions that are not used during operation, which makes it complicated and expensive, resulting in poor cost performance. Furthermore, when conventional controlled variable response waveform recognition means are used, there is a problem that the control response is so fast that it cannot keep up with the desired function and auto-tuning becomes impossible.

SUMMARY OF THE INVENTION An object of the present invention is to provide a tuning method and apparatus for a digital speed control device having an auto-tuning function that solves the above-mentioned problems.

[Means to solve the problem]

In order to solve the above-mentioned problems, a control response accumulation means for accumulating a command value and a check value of a control amount for each sampling as control response data in a digital control device, and a control response and feedback control accumulated in the control response accumulation means are provided. A data input/output means is provided to enable rewriting of the control gain.
Further, a control response reading means inputs the control response data accumulated in the control response accumulation means, a control characteristic setting means for setting desired control characteristics, and a set control characteristic set in the control response data and the control characteristic setting means. A control gain tuning device is provided, which includes a control response evaluation means for determining an evaluation index from the above, and a control gain modification means for modifying the control gain based on the evaluation index.

[Effect]

The data input/output means which accumulates the command value and the detected value of the control amount for each sampling of the control response accumulation means rewrites the control response data accumulated in the control response accumulation means and the control gain of the feedback control. The tuning device inputs the control response data accumulated in the control response storage means through the data input/output means using the control response reading means, and performs control based on the set control characteristic set in the control characteristic setting means and the control response data. The response evaluation means obtains an evaluation index, the control gain modification means modifies the control gain based on the evaluation index, the data input/output means modifies the control gain, and the control characteristics are tuned to desired characteristics.

〔Example〕

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a diagram showing an embodiment of the present invention. 1
2 is a digital control device, 2 is a power converter, 3 is a motor, 4 is a speed detector for detecting the speed of the motor 3, and 5 is a tuning device.

The digital control device 1 inputs the speed detection value of the motor 3 detected by the speed detector 4 and the speed command, and calculates the PI for the deviation.
A speed control 101 that performs a D compensation calculation and operates the power converter 2 to control the speed, a speed response storage 102 that stores speed commands and detected speed values for each sampling period, and a reading ratio of the accumulated data of the speed response storage 102. and a data input/output 103 that rewrites the control gain of the speed control 101. Tuning device 5 is digital control device 1
Speed response readout 51. inputs the speed command rn+speed detection value nn stored in the speed response storage 102 of the speed response storage 102. Control characteristic evaluation 5 consisting of interpolation processing 521 and control response evaluation 522
2. Control characteristic setting 53 for setting desired control characteristics; control gain change 5 for changing control gain of speed control 101
41. The control gain modification 54 includes a control gain modification amount determination 542 that determines a control gain modification amount, and a tuning rule 543 that defines a control gain modification method for the control gain modification amount determination 542.

The functions of speed control 101 and speed response storage 102 are digital control devices! It is processed and functions by the program of the microprocessor in 1. The processing program will be explained with reference to the flowchart shown in FIG. Processing is started by a control interrupt that occurs at every predetermined sampling period, and includes processing 1000 of speed control 101 and speed response accumulation 1.
02 processing 1001 is performed. Process 1000 further includes steps 1000a to 1000d.

In step 1000a, a speed command r and a detected speed value n are input. In step 1000b, the deviation e is set in step 10.
It is calculated from the speed command r input at 00a and the detected speed value n. PID compensation calculation is performed in step 1000c, and the value obtained by multiplying the deviation e by the proportional gain, the value obtained by multiplying the integral value of the previous deviation e by the integral gain, and the value obtained by multiplying the differential value of the deviation by the differential gain are added. to determine the current command for the controlled variable.

At step 1000cl, a current command is output to the power converter 2 to operate the motor 3.

Processing 1001 corresponding to speed response accumulation 102 is step 1
Consists of 001a to 1001c. Step 1001
i of the speed command memory REF, which is a memory provided in the microprocessor of the digital control device 1, for the speed command r at a.
Store speed detection value n in speed detection value memory S
The i address of PD is stored by 1, and i is incremented by 1.

In step 1001b, i is IN which is the final address of the speed command memory REF and the speed detected value memory SPD.
It is determined whether data has been accumulated up to address M, and if the final address INH has been reached, step 1001c is performed. Step 1001c sets i to the leading value O of each of the speed command memory REF and the speed detected value memory SPD. Through this process 1001, the command value rn and speed detection value nn for each control sampling are stored in the speed command memory REF and the speed detection value memory SP'D.

The data input/output 103 functions to enable the tuning device 5 to read the accumulated data of the speed response accumulation 102, and to change the control gain of the speed control 101 by the control gain change 541 of the tuning device 5, and its operation is also digital. It is processed and functions by the microprocessor of the control device 1. Data input/output 10 according to Figure 3
The operation of step 3 will be explained.

When reading the stored data, which is inputted by the speed response readout 51, the speed response readout 51 of the tuning device 5 and the data input/output 10 are performed in the data format shown in FIG. 3(a).
Data is exchanged with 3. 'r' as a command indicating reading.

Next, a space I Sp l indicating a break, the address of the stored data memory to be read next, and a space # s indicating a break.
Pj, a data count indicating the number of data to be read, and finally a blank JSpl indicating a nail are applied from the speed response readout 51 to the data input/output 103. When the microcomputer of the digital control device 1 inputs this, it sequentially reads out several data counts of accumulated data from the address position and outputs it to the tuning device 5. The speed response readout 51 of the tuning device 5 reads out the speed response data by sequentially taking the data and storing it in the memory.

When the control gain change 541 rewrites the control gain of the speed control 101, data is exchanged with the data input/output 103 in the data format shown in FIG. 3(b). g kl is sent as a command to indicate writing, a space i sPl indicates a break, the start address of the memory that stores the control gain to be written next, and a space 1s indicates a break.
After pj, a data count indicating the number of data to be written, and a space l s pJ indicating a break, the data to be written is sent as many times as the data count. Digital control device 1
When the microcomputer receives the input, it performs an operation of sequentially writing data of several data counts from the memory at the location specified by the address, thereby fulfilling the function of the data input/output 103.

Next, the operation of the tuning device S will be explained. The overall operation of the tuning device 5 is as follows. Speed response readout 51 reads speed response data stored in the memory of digital control device 1, and control characteristic evaluation 52 extracts feature quantities of the control response. At this time, since the data stored in the memory is the data at the time of sampling, the six special quantities that perform accurate control characteristic feature extraction without compensating for data between samplings are set as control characteristics 53 through interpolation processing 521. A control characteristic evaluation index is determined by comparison with a desired control characteristic. The control characteristic evaluation index is input to the control gain correction amount determination 542, fuzzy inference is performed based on the adjustment rule 543, and the control gain of the speed control 101 is rewritten via the data input/output 103 by the control gain change 541 to change the control characteristics. Perform tuning.

The control characteristic evaluation index determination process performed in the control characteristic evaluation 52 will be explained based on FIGS. 4 and 5.

The control characteristic evaluation index evaluated in the control characteristic evaluation 52 is a time constant, an overshoot amount, and a damping coefficient. FIG. 4 is a flowchart of the process, and FIG. 5 shows the relationship between the accumulated data read out by the speed response readout 51 and the control response. At step 200°, the step change width r of the speed command
, and its start time T0. The process reads out the speed command value r, sequentially increases i to find i that satisfies the following equation, and calculates the step change @rw and step start time T0 using equations (2) and (3).

r turtle≠r1+, ...(
1) r, ≠ri+1−r, ・・
・(2) T o = T s book i
・(3) However, Ts is the value n of 63% of the step change width r for the sampling period step 2001.
Calculate t using the following formula: nt = 0.63+k rw...
- (4) In step 2002, nPpnM sandwiching the 63% value shown in the following equation is searched from the accumulated data of speed detection values in order to obtain a time constant.

nP≦nt≦nq (5) In step 2003, the time constant tr is determined. This step corresponds to the interpolation process 521, in which the time point at which the motor speed exactly matches nt is determined by interpolation using the following equation, and the time constant Tr is calculated.

Tr=(p+(nt-np)/(nq-np))・T
sT.

(6) In step 2004, the extreme value n□pnm2H... is determined. The detected speed values nn are sequentially read out, and points na' and nb where the speed changes from increasing to decreasing are determined.

Since the extreme value n=1 is between the two points na and n6, its value is determined by interpolation processing 521. Interpolation is performed using the following quadratic equation. However, the origin of time t is assumed to be at time na.

n=αt2+βt+γ...(7)
Since equation (7) is a quadratic equation, when α, β, and γ are determined using the three points na, nbt nc, the following equation is obtained.

rs γ"na ... (1o
) The time t at which equation (7) takes an extreme value is determined by the following equation, and the extreme value n□ is found from equation (12).

na-3nb+nc 11 na'+9nb"+nc"-20nanb 6
nbnc+8ncna...(12) Other extreme value n12. ... is found in the same way. Note that extreme values do not occur if the response of the control characteristics is not oscillatory, but
In that case, the extreme value nmzHn+m that cannot be found
The value of 2y... is O.

In step 2005, a time constant evaluation index TR is calculated.

The time constant determined in step 2003 and the target control characteristic trs of the time constant set in the control characteristic setting 53 are determined by the following equation.

TR=tr/rrs
(13) In step 2006, an overshoot amount evaluation index E is calculated. It is determined by the following equation from the first extreme value n+ai determined in step 2004, the step change width rw determined in step 2000, and the step command value.

E −(n wa, −r t+,)/r W
-(14) Note that when the value of the extreme value n1□ is O, the overshoot amount evaluation index E is set to 1.

In step 2007, a damping coefficient evaluation index is calculated.

The extreme value n1□ found in step 2004.

n■2. It is determined from n, , using the following formula.

n=(nm, nmz)/(nmz nm2) ・
...(1s) Furthermore, the extreme value n1□Hn wa2 p n
When any value of 113 is O, the attenuation coefficient evaluation index is set to 0.

Next, control gain modification-54 using fuzzy inference will be explained. Control gain correction amount determination 542 is based on time constant TR
, overshoot amount E, and damping coefficient D. In order to qualitatively evaluate the magnitude of each control characteristic evaluation index, FIG. 6(a) is used for each control characteristic evaluation index.
Membership functions as shown in FIGS. 6(c) to 6(c) are defined. TR1 to TR4, E1 to E, and D in the diagram.
D2 is a constant that defines the membership function, and NB
,ZE.

PM and PB are names given to membership functions to qualitatively evaluate the size, and each has the following meanings.

NB: Negative BigZ E: Z
Er.

PM: Po5itive MediumPB
: Po5itive Big The vertical axis of the figure is the confidence level, which takes a value of 0 to 1.

An example of the tuning rules 543 is shown in FIG. The control characteristic evaluation index is evaluated using the membership function for various control response shapes, and how each control gain should be modified is qualitatively specified based on the qualitatively evaluated results. For example, in the case of rule 1, 1f(TRis Pa and E is
ZE and D is ZE) then(K
p is PB and Tl1s ZE and T
o is ZE). Here, TR is a time constant, E is an overshoot amount, D is a control characteristic evaluation index of a damping coefficient, and Kp, Tr, and To are correction coefficients of a proportional gain, an integral gain, and a differential gain. FIG. 8 shows membership functions for converting the qualitatively determined control gain correction coefficients Kp, TI, and To into quantitative quantities, which are provided in the control gain correction amount determination 542. 01 to C4 in the figure are constants that define the form of the membership function. NB, ZE, and PB are names given to membership functions to qualitatively represent the magnitude of the control gain correction coefficient, and have the same meanings as the names used in FIG. 6. Moreover, the vertical axis of the figure is the confidence level.

The control gain correction amount determination 542 inputs the control characteristic index from the control characteristic evaluation 52, qualitatively determines each control characteristic evaluation index using the membership function shown in FIG. 6, and performs fuzzy inference based on the tuning rule 543. Then, the control gain correction coefficient is determined qualitatively, and the control gain correction coefficient is made into a quantitative value using the membership function shown in FIG. When the qualitative control gain correction coefficient for a given overshoot amount E, damping coefficient and time constant TR is specified by multiple tuning rules as shown in Fig. 7, weighting is applied according to each membership value. Take the average and convert it into a quantitative value.

The control gain change 541 determines the current control gain by multiplying the PID control gain correction coefficient obtained in the control gain correction amount determination 542 by the previous value of the PID control gain, and controls the speed control 101 via the data input/output 103. Change the gain.

According to this embodiment, the process in the digital control device 1 for auto-tuning is as simple as storing the command value and the detected speed value in the memory every sampling period, so that the control response speed is reduced. It can be processed by the microcomputer of the digital control device 1 without having to perform automatic tuning. Furthermore, the digital control device 1
There is no need to incorporate a complicated part that performs fuzzy inference to modify the control gain, and the configuration does not become complicated.

In addition, since the digital control device 1 and the tuning device 5 are separated, one tuning device 5 can be used sequentially for many digital control devices 1, which has the effect of realizing an auto-tuning function at a low cost. . Furthermore, since the control characteristics are correctly evaluated through interpolation processing, there is also the effect that auto-tuning can be performed accurately to the set control characteristics.

In the interpolation processing 521, an example has been explained in which a linear equation is used for interpolation of the time constant tr, and a quadratic equation is used for interpolation of the extreme values nm1y nm2y, etc. n, 2.
. . . Both may be interpolated at the same order, or at a higher or lower order. In addition, control gain correction amount determination 54
In 2, the membership function is triangular, but it is not necessary to stick to this shape, and a quadratic curve, an exponential function, etc. may be used.

Furthermore, not only the shape of the membership function but also the number thereof may be arbitrarily set, and other functions may be used as the control response evaluation index.

〔Effect of the invention〕

According to the present invention, the processing in the digital control device for auto-tuning is as simple as storing the command value and speed detection value in the memory every sampling period, so that the control response speed can be reduced. It can be processed by the microcomputer of the digital control device and auto-tuning can be performed. Furthermore, since there is no need to incorporate a complicated part that performs fuzzy inference and corrects the control gain into the digital control device, the configuration of the digital control device does not become complicated. Furthermore, since the digital control device and the tuning device are separated, one tuning device can be used in sequence for a large number of digital control devices, which has the effect of realizing an auto-tuning function at a low cost. Furthermore, since the control characteristics are correctly evaluated through interpolation processing, there is also the effect that auto-tuning can be performed to the set control characteristics.

[Brief explanation of drawings]

Fig. 1 is a diagram showing the configuration of auto-tuning in a digital control device according to the present invention, Fig. 2 is a flowchart explaining the operation of the digital control device, and Fig. 3 is a data format used for reading accumulated data and rewriting control gains. FIG. 4 is a flowchart of the process for determining the control characteristic evaluation index, FIG. 5 is a control response diagram for explaining the process for determining the control characteristic evaluation index, and FIG.
Figures (a) to 6(c) are diagrams showing membership functions for control response evaluation, Figure 7 is a diagram showing an example of a tuning rule, and Figure 8 is a membership function for control gain correction coefficients. In the diagram showing al. 1... Digital control device, 2... Power converter,
3...Motor, 4...Speed detector, 5...Tuning device. (J) CC/
)Figure 7Figure 8

Claims (1)

[Claims] 1. In a digital control device that performs feedback control by inputting a command value and a detected value of a controlled variable at each predetermined sampling period, the command value and the detected value of the controlled variable are input at each sampling period. A digital control device comprising: a control response accumulation means for accumulating the control response data as control response data; a data input/output means for reading out the control response data accumulated in the control response accumulation means and rewriting the control gain of the feedback control; control response reading means for inputting control response data accumulated in the control response accumulation means of the control device via the data input/output means of the digital control device; control characteristic setting means for setting desired control characteristics;
control response evaluation means for calculating an evaluation index from the control response data inputted by the control response reading means and the control characteristic set in the control characteristic setting means; and control gain modification means for modifying the control gain based on the evaluation index. 1. A tuning device for a digital control device, characterized in that the control characteristic is tuned by connecting a control gain tuning device consisting of the following. 2. The tuning device according to claim 1, wherein the digital control is characterized in that the evaluation index is obtained by estimating the value required for calculating the evaluation index by performing an interpolation calculation from the control response data between samplings. How to tune the device. 3. The tuning device for a digital control device according to claim 1, characterized in that the tuning device is connected to the digital control device only when adjusting the digital control device, and is not connected when the digital control device is operating. How to tune a digital control device.