JPS61164480A - Motor control device - Google Patents

Abstract

PURPOSE:To perform the calculations of regulating control constants and controlling an object to be controlled for variation in parameters by a microprocessor without deteriorating the control performance to alternately calculate the regulations and the controls divided into a plurality. CONSTITUTION:In a control calculator 6 of a digital controller 4 by a processor, an operating amount is obtained from a target error of the target values and detected values of the speed and the position of a motor 2. In an identification calculator 7 presumes parameters showing the characteristic of a target to be controlled by a sequential calculation. The one calculating time of the identification calculation and the regulating calculation including control constant calculation is approx. 50ms. In this case, the regulating calculation is divided equally to a plurality to reduce each calculating amount to shorten one control calculation and each regulating calculation time, i.e., sampling period. Thus, the control and the regulation can be performed by a microprocessor without deteriorating the control performance.

Description

[Detailed Description of the Invention] [Field of Application of Invention 1] The present invention relates to a motor control device that adjusts control constants, and particularly to a motor control device that uses a method suitable for adjusting control constants online. Regarding.

[Background of the invention]

In conventional motor control devices, the characteristics of the motor to be controlled and the load connected to the motor change from time to time. For example, in a robot, the inertia of the motor shaft changes as the arm bends and stretches. Therefore, the control characteristics deteriorate due to such parameter fluctuations of the controlled object. Therefore, it is necessary to readjust the control constants in response to parameter fluctuations. However, the amount of calculation required for adjustment is generally greater than the amount of calculation required for control, and if this control calculation and adjustment calculation are performed by one processor, the sum of the time required for control calculation and the time required for adjustment calculation will be Since the sampling period becomes several tens of times longer than the sampling period in the case of only control calculation, the control characteristics deteriorate.

As a conventional means of shortening the sampling period,
For example, as described in Japanese Unexamined Patent Application Publication No. 57-10806, a method is known in which a part of the adjustment calculation is performed by a higher-order processor of the control device. However, this method is intended for large-scale systems such as plants where a host processor can be used, and for small, closed systems such as motor control devices that do not have a host processor. is inconvenient. Therefore, it is conceivable to perform the control calculation and the adjustment calculation using separate processors, but in this case, there is a problem that the circuit becomes larger and the cost becomes higher.

[Purpose of the invention]

SUMMARY OF THE INVENTION An object of the present invention is to provide a motor control device that can perform adjustment calculations for readjusting control constants and control calculations in response to parameter fluctuations of a controlled object using a single microprocessor without deteriorating control performance. be.

[Summary of the invention]

In general, in motor control, the amount of calculation required to adjust control constants in response to parameter fluctuations of the controlled object is greater than the amount of calculation required for control, but parameter fluctuations are sufficiently slower than the response speed of the controlled object. Focusing on the fact that it is not necessary to perform the adjustment and control at the same period, and the sampling period for adjustment can be sufficiently longer than that for control, one adjustment calculation is equally divided into multiple parts, and each adjustment calculation and one adjustment calculation are divided into two. This motor control device is capable of performing control and adjustment with a single microprocessor without making the sampling period fc of the control device too long and without deteriorating control performance by performing control calculations alternately.

[Embodiments of the invention]

An embodiment of the present invention will be described below with reference to the drawings.

FIG. 2 is a block diagram showing an overall configuration of an embodiment of a motor control device according to the present invention. In Figure 2, load 1
The motor 2 to which the motor is connected, the detector 3 that detects the speed and position of the motor, and the drive amplifier 5 are controlled objects. A digital control device 4 using a single microprocessor that calculates the manipulated variable for this controlled object at a fixed sampling period and automatically adjusts control constants, and a drive amplifier 5 that drives the motor 2 in accordance with the manipulated variable. A motor control device is constructed from the following. Furthermore, the digital control device 4 including a single processor includes a control calculation section that calculates a manipulated variable at a constant sampling period based on a control constant from a target error between a target value and a detected value, and a control calculation section that calculates a manipulated variable at a constant sampling period from the manipulated variable and detected value. It consists of an identification calculation section that identifies parameters, and a control constant calculation section 8 that calculates control constants based on the identified parameters.

In this configuration, first, in the control calculation unit 6 of the digital control device 40 using a processor, the calculation formula for the control calculation for determining the manipulated variable from the target values of the speed and position of the motor 2 and the target error of the detected value is used for control calculations generally used for motors. The system is PI or PID control and is expressed by the following equation.

y(t): Manipulated amount e(t): Target error, , Ti, Td: Control constant Here, if the summer processor is an 8-bit CPU, one calculation time required for the control calculation of equation (1). is 1m
s or less.

Next, in the identification calculation section 7, this identification calculation is to estimate parameters indicating the characteristics of the controlled object by sequential calculation, and in this embodiment, Lyabunov's direct method is used. In this method, a model with the same configuration as the controlled object is installed in the processor, and the parameters are identified by the output error using the manipulated variables of this model and the controlled object as input, and the identification rule is expressed by the following equation. .

XM(K): Output of controlled object (detection value) x ITl(K
): Model output E(x): Output error Z(K-1)1z(K-1): Matrix related to convergence and stability of identification (AT is the transposed matrix of A) Ω(K): (Φ(K) ) , 'F(x) )Φ(K)=
A engineering (K) - AM F (K) == Bm (K) ByI AM, BM: parameters of the controlled object Am (K) + Bm (x) e, identification in each cycle In the identification calculation section 7, first When the transfer function of the controlled object is determined using the identification parameters determined by the above equation (2), the following equation is obtained.

G(s)=□...(3)a(1+als
+a2s+... Next, a control constant adapted to the controlled object of the transfer function G(8) of the obtained equation (3) is calculated. In this embodiment, a partial coefficient comparison method is used. In this method, first, a transfer function system of the following equation is set as a control system with an ideal response.

Gm(s)=□...(4) α. +α1δ3+α2δ232+α, δ5,5+...
...α. , α0. α2. α5.・・・・・・：1,1
, 1.5, 0.15 and so on, and then find δ as the smallest real root in the following equation ◎...
(5) The δ height control constants Kp, Ti, and Td are obtained as shown in the following equation.

Here, according to the above-mentioned 8-bit CPU, the calculation time for one adjustment calculation including the above-mentioned identification calculation and control constant calculation is about 50 m5. In this way, the adjustment calculation requires a large amount of calculation and takes a long calculation time compared to the control calculation. The same holds true even if the least squares method is used for the identification calculation in the adjustment calculation, and other methods such as the Jiggly-Nichols method are used for the Jiggi-Nichols calculation.

FIG. 3 is a time chart of the spring/pull cycle when the above adjustment calculation is not divided. In Figure 3, as mentioned above, the amount of calculation required for adjustment is generally larger than that required for control, and if the adjustment calculation is performed without dividing it between 1t1 processors, the time required for control calculation is Since the sum t of tc (1 ms or less) and the longer time tT (about 50 ms) required for adjustment calculation is the sampling period t of the control device, it is several times longer than the sampling period t0 for only control calculations. to number 10
It becomes twice as long and the control characteristics deteriorate. Therefore, according to the present invention, the above adjustment calculations are equally divided into a plurality of parts to reduce the amount of calculation for each part, thereby shortening the time for one control calculation and each adjustment calculation, that is, the sampling period.

FIG. 4 is a time chart of the sampling period when the above adjustment calculation is divided into equal parts. Further, FIG. 1 is a time chart of the sampling period when the above adjustment calculation is time-divided according to the present invention. First, in Figure 4,
The adjustment calculation of the adjustment calculation time tT in each sampling period t in FIG.
Let tTn (=ts/n).

Next, in FIG. 1, according to the present invention, each adjustment calculation of the series of calculation times tT1 # tT21 . . . In other words, the control calculation section 6 (
The control calculation using equation (1) in FIG. Adjustment calculations including identification calculations and control constant calculations using Equation 6) are based on the sampling period t of the control device.
The control constant is adjusted once every n periods of . Therefore, adjustment calculation time 1T adjustment calculation 7&:
Although the sampling period t of the control device is performed every nXt period, deterioration of the control performance can be prevented without increasing the sampling period t. Moreover, in motor control, the fluctuations in the above parameters are generally much slower than the response speed of the controlled object, so there is no need to perform control and adjustment in the same sampling period, and the sampling period for adjustment is considerably larger than the sampling period for control. Therefore, the control constants can be sufficiently readjusted in response to parameter fluctuations. Regarding the parameter fluctuation of the controlled object, for example, in the case of a robot, there is an inertia fluctuation in the motor shaft due to bending and straightening of the arm. However, when this parameter variation is at its maximum, that is, when the arm transitions from an extended state to a bent state or vice versa, the parameter variation period is proportional to the speed of the arm and is several hundred m5 even at the maximum speed, and this can be accommodated. The y4 square/blink period is about 10m8.

The sampling period of one control is proportional to the response speed of the controlled object, and is 5mg for the motor used in the robot.
to 3 mg. Therefore, the sampling period for adjustment may be sufficiently shorter than the sampling period for control. Regarding the equal number n of multiple n equal parts in the above adjustment calculation,
It may be determined in accordance with the period of parameter fluctuation of the controlled object and the adjustment calculation time.

As described above, according to this embodiment, adjustment calculations related to self-adjustment of control constants in response to parameter fluctuations of a controlled object in motor control are divided into n equal parts and are performed between control calculations, so that a single micro Using a processor, control and adjustment can be performed without significantly increasing the sampling period of the control device and without deteriorating the control characteristics, resulting in smaller circuits and lower costs. Note that the present invention is also applicable to control objects other than motors.

〔Effect of the invention〕

As described above, according to the motor control device of the present invention, control and adjustment can be performed using a single microprocessor without deteriorating control performance. Data can be shared between control and adjustment calculations, reducing memory capacity, and the overhead when switching between control and adjustment is reduced, resulting in more efficient processing.

[Brief explanation of drawings]

FIG. 1 is a time chart showing an embodiment of the motor control device according to the present invention when adjustment calculation is time-divided;
This figure is also a block diagram of the overall configuration, FIG. 3 is a time chart when the adjustment calculation is not divided, and FIG. 4 is a time chart when the adjustment calculation is divided into equal parts. 1...Load, 2...Motor, 3...Detector, 4...
...Digital control device using a processor, 5...Drive amplifier, 6...Control calculation section, 7.Identification calculation section,
8... Control calculation unit, t3... Sampling period, t
o... Control calculation time, tTl to tTn... Each adjustment calculation time divided into n equal parts Agent Patent attorney Tadashi Akimoto Figure 1 Figure 2

Claims (1)

[Claims] A single microcontroller that controls a motor to which a load is connected, and is equipped with a control calculation means that uses the target value and detected value of the motor speed (position) as input, and calculates the manipulated variable based on the control constant at each sampling period. In a motor control device consisting of a digital control device using a processor and a drive amplifier that drives the motor based on the manipulated variable from the digital control device, the digital control device using the single microprocessor performs the control calculations described above. In addition to the means, the apparatus includes an identification calculation means for identifying a parameter indicating a characteristic of a controlled object, and a control constant calculation means for determining the control constant based on the identified parameter, and performs an adjustment calculation consisting of the identification calculation and control constant calculation. A motor control device that divides the motor into a plurality of equal parts and alternately performs adjustment calculations for each of the equal parts and one control calculation.