Dispersed neural robust controlled trajectory tracking algorithm for mechanical arm

Abstract

The invention discloses a dispersed neural robust controlled trajectory tracking algorithm for a mechanical arm. According to an algorithm flow, the algorithm sequentially comprises constructing a high-order recurrent neural network model, an estimating a nonlinear system by an RHONN model, estimating unknown weight coefficients in the models, designing a dispersed robust neural controller and proving stability. The invention aims to design a neural controller for eliminating disturbances to improve trajectory tracking accuracy of the mechanical arm.

Description

technical field [0001] The invention belongs to the technical field of robots, and specifically relates to a track tracking algorithm for decentralized neural robust control of a manipulator. Background technique [0002] With the rapid development of the robot industry, the requirements for the robot's work indicators are getting higher and higher, so the trajectory tracking accuracy of the manipulator is becoming more and more important. At present, the commonly used robot manipulator control methods mainly include PID control. Although this control method is simple and easy to implement, it often requires a lot of control energy, and it cannot guarantee that the robot has good static and dynamic performance; adaptive robust control, its The disadvantage is that the huge calculation required for online identification parameters has strict real-time requirements, especially when there are non-parameter uncertainties, it is difficult for adaptive control to ensure system sta...

AI Technical Summary

Problems solved by technology

At present, the commonly used robot manipulator control methods mainly include PID control. Although this control method is simple and easy to implement, it often requires a lot of control energy, and it cannot guarantee that the robot has good static and dynamic performance; adaptive robust control, its The disadvantage is that the huge calculation required for online identification parameters has strict real-time requirements, especially when there are non-parameter uncertainties, it is difficult for adaptive control to ensure system stability; neural network control and fuzzy control, due to the non-uniformity of the dynamic model of the robot Linearity, the model parameters are often difficult to obtain accurately, which affects the trajectory tracking accuracy of the robot arm to a certain extent, the neural network fuzzy control has a high degree of nonlinear approximation ability, and can accurately approximate the unknown part of the dynamic equation of the robot arm online , to achieve high-precision tracking of the robot, but neural network control often requires a large amount of training, which greatly increases the amount of calculation

Images