JPS6364102A - Learning control system - Google Patents

Abstract

PURPOSE:To quickly execute the converging by performing a next reproduction operation in accordance with a command value to add a value to multiple respective loop gain values to respective deviations to teaching value or a present command value concerning the acceleration, speed and position of a controlled system for respective freedom degrees. CONSTITUTION:A command value arithmetic unit 1 stores the position of a work locus, which is a target, to an object and sets the gain value of respective control loops. Based on the initial setting, a target locus thetad is given as a first command value. At this time, anacceleration signal, a speed signal and a position signal detected for each sample time by a detecting device 7 from a controlled system 6 are stored through an A/D converter 8 to a memory 9. When a first reproducing operation is completed, based on the stored data, an evaluation function J, for example, like an error square integrated value is counted by the device 1 and it is decided whether the function J is larger or smaller than a value Jmin of a prescribed value. When the function J is smaller than the value Jmin, the control is completed and when the function is larger, the device 1 corrects and calculates the command value by using the deviation of the acceleration, etc., with a target value and an output signal.

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention relates to a learning control method for objects that are repeatedly controlled, such as playback robots, and particularly to a learning control method with fast convergence.

[Prior Art] Generally, when positioning an object using repetitive control such as in a playback type robot, for example, Japanese Patent Laid-Open No. 60
As shown in Publication No. 57409, first, a teaching operation is performed to make the target object memorize the position data of the target work trajectory (hereinafter referred to as the teaching value), and regeneration operation is performed according to this teaching value, and the above teaching value and A learning control method is adopted in which a difference from the driving trajectory (hereinafter referred to as deviation) is detected, and this deviation is added to the teaching value as a command value for the next regenerative operation.

In addition to the general learning control described above, as shown in the above publication, there is a control method that takes into account the dynamic delay time for each degree of freedom, a method that takes the error in speed between the target trajectory and the driving trajectory as the deviation,
Methods that calculate errors in acceleration have been proposed, with the aim of improving the accuracy and stability of learning control.

[Problems to be solved by the invention] However, in the conventional learning control method, the gain value of the control system is only selected within a range that satisfies stability, and is not determined in consideration of the characteristics of the system. Therefore, there was a problem that the trajectory would not match the target trajectory unless a large number of trials were repeated.

This invention has been made to solve the above-mentioned problems, and aims to provide a learning control system with good positioning accuracy and fast convergence.

[Means for Solving the Problems] The learning control method according to the present invention calculates, for each degree of freedom, the deviations of the acceleration, velocity, and position of the controlled object multiplied by the respective loop gain values. , the next regeneration operation is performed in accordance with the taught value or the command value added to the current command value. [Function] In this invention, acceleration, velocity, and position are Since the gain value of each control loop is taken into consideration, learning suitable for the control system is possible, and learning control with good positioning accuracy and fast convergence can be realized.

[Example] Hereinafter, an example of the present invention will be described with reference to the drawings. 1st
The figure is a block diagram showing one embodiment of the present invention. In the figure, (1) is a command value calculation device, such as a digital computer, that calculates and outputs a control command value, and (2) is a command value calculation device ( A D/A converter that converts the digital signal from 1) into an analog signal, (3) a comparator that is, for example, an operational amplifier, (4) a control circuit, (5) a servo amplifier,
(6) is an object to be controlled, (7) is a detector that detects output signals representing the position, velocity, and acceleration from the object to be controlled (6), and (8) is a signal that is fed back from the detector (7). A/D converter that converts analog signals into digital signals, (9)
is a memory that stores digital signals from the A/D converter (8).

Next, the operation will be explained based on FIG.

FIG. 2 is a flowchart showing a command value calculation program executed by the command value calculation device (1).

First, in the initial setting (step (11)), the target object is memorized with the position data of the target work trajectory through a teaching operation, and the gain values of each control loop are set.

That is, motor inertia Im is set for acceleration, speed servo gain Kv is set for speed, and position servo gain KP is set for position. Next, a regeneration operation is performed based on this initial setting (step (12)). 1
As the second command value, if the target trajectory is θd, then U z (t) = I m tj d(t) 10 Kv
Q d (t) + Kp B d (t) ... (1)
give. At this time, the acceleration signal υaX (t) and the speed signal OdY (1
) and position signal θd (1) are A/D converter (8)
It is stored in the memory (9) through. When one regeneration operation is completed, an evaluation function J such as an error squared integral value is calculated in the command value calculation device (1) based on the stored data (step (13)), and the evaluation function It is determined whether J is larger or smaller than a predetermined value J win (step (
14)) If the evaluation function J is smaller than the predetermined value J1lin, the control ends, but if not, the command value calculation device (1) calculates the acceleration between the target value and the output signal, Deviation in velocity and position δi (t) = σd(t) − σd(1), (t)
= 6d(t) -bd(1)e i (t) =θ
The command value U(1) is corrected using d(t)-θdl(t), and a new command value U2(t) is calculated using equation (2) (step (15)).

U2(t)=U1(t)+In2. (t)+Kve
1(t)+Kpe, (t) (2) Similar operations are repeated until the evaluation function J becomes smaller than J win.

In general, the equation of motion of a controlled object includes, in addition to each item shown in equation (2) above, a nonlinear term that depends on acceleration such as the inertia of the controlled object, a nonlinear term that depends on velocity such as centrifugal force,
It includes nonlinear terms such as gravity terms and gravity correction terms.

This time, the linear terms Imjd(t), Kv6d(t), and KpOd that are dominant in the system are used as the first input U(1).
Since (t) is included, the error at this time is only due to the nonlinear term described above, and a significant correction is made. Furthermore, for second and subsequent corrections, learning is corrected based on acceleration, velocity, and position based on the coefficients of the dominant linear terms in the system, which improves positioning accuracy and provides learning control with fast convergence. realizable.

In the above embodiment, the control system and the controlled object are analog servo systems, but they may also be digital servo systems.

Further, in the above explanation, only one degree of freedom has been explained, but it goes without saying that the same operation can be performed for other degrees of freedom.

[Effects of the Invention] As described above, according to the present invention, in the learning control method, acceleration, velocity, and position are calculated for each degree of freedom by considering the coefficients of the linear terms that are dominant in the control system. Since learning correction is performed, learning suitable for the control system is possible, and learning control with good positioning accuracy and fast convergence can be realized.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing one embodiment of the present invention, and FIG.
The figure is a flowchart showing the operation. In the figure, (1) is a command value calculation device, (4) is a control circuit, (6) is a controlled object, (7) is a detector, and (9) is a memory.

Claims (3)

[Claims] (1) A controlled object having multiple degrees of freedom is operated in a regenerative manner according to the taught value, and the deviation between the taught value and the regenerated trajectory is measured. During the next regenerative operation, the taught value or the current command value is adjusted based on the deviation. In a learning control method in which learning control is performed for each degree of freedom, regenerative operation is performed according to a command value to which a drop correction value has been added. A learning control method characterized in that a value multiplied by a loop gain value is used as the correction value. (2) In the above learning control, the initial taught value is set as a value obtained by multiplying the acceleration, velocity, and position of the target trajectory of the object by a predetermined acceleration gain value, velocity gain value, and position gain value, and the next command value to the above taught value or current command value,
Claim 1, which is a value obtained by multiplying each acceleration deviation, speed deviation, and position deviation by the acceleration gain value, speed gain value, and position gain value, respectively.
Learning control method described in section. (3) Claim 2 in which the acceleration gain value is an inertia value of the drive motor to be controlled, the speed gain value is a speed servo gain value, and the position gain value is a servo gain value for the position of the control target. Learning control method described in section.