A brain-inspired learning control method for a multi-degree-of-freedom robot

Abstract

The invention discloses a brain based learning control method of a multi-freedom-degree robot. The method comprises the steps that firstly, a system dynamical model of the multi-freedom-degree robot is built; secondly, an intelligent controller u based on MSAE-NN is built; thirdly, the controller u acts on the robot system built in the first step, the output y(t) traces the desired trajectory xd(t) according to the given accuracy beta0, and meanwhile, it is ensured that the system tracking error e(t) is bounded within t larger than or equal to 0. Different from most robot neural network control methods, the brain based learning control method completely inherits the excellent characteristics of the MSAE-NN, the network has structural diversity primary functions and time varying ideal weights, and according to the system current output deviation, the number of nerve cells can be adjusted in real time. On the one hand, the cumbersome process that NN related parameters are configured by manual work in a repeated test manner is avoided; on the other hand, self learning and self-adaptation capacity of the system are consolidated and reinforced, and the whole intelligent degree is improved.

Description

technical field [0001] The invention relates to the field of robot control and the field of uncertain nonlinear dynamic system control, in particular to a bionic intelligence control method of a multi-degree-of-freedom robot. Background technique [0002] A multi-joint manipulator is a typical object for MIMO system control. Combined with the adaptive NN stability analysis technology, many achievements have been made in the control of manipulators. For example, in order to improve the accuracy and comprehensive performance of trajectory tracking control, some scholars have designed the position tracking task of the end effector without payload mass first. online NN adaptive controller based on experimental information; some scholars realized the interaction between robots and uncertain viscous environments through NN control; A robust position control method with limited output tracking error is proposed. Since the NN model with a fixed structure is used in these control s...

AI Technical Summary

Problems solved by technology

Since the NN model with a fixed structure is used in these control strategies, there are the following problems in actual use scenarios: NN control is based on the strict premise of the Universal Approximation Theorem (UAT), and it is necessary to ensure that the NN can be safe and effective during the controller design and integration stages. Once the NN parameters are set improperly, it will not only lose the overall approximation ability of NN, but also affect the smooth and safe operation of the system

Generally speaking, the design and analysis process of traditional NN control is very complicated, and the established controller usually has a complex structure and requires a large amount of system online computing resources.

Images