Research On Neural Adaptive Inverse Control Theory Of Linear Servo System

A new kind of inverse control tactic based on neural network control for dynamical and rapid high-precision-feed in a single axis is proposed in this thesis, The background of the thesis is "Theory of Dynamic Precision synchronization Traverse and Research of Realization Methods for Linear Servo Dual Position Loops System (NO.50075057)", which is supported by National Natural Science Foundation of China.The linear permanent magnet synchronous motor(LPMSM) has avoided the effects of the mechanical transmission chains from rotary motions to linear ones, and has strong electromagneticm thrust, lower cost, small electrical tune constant and rapid response etc., which becomes one of the best executive machines in high-precision and micro-feed servo system. In this project, using LPMSM as driving parts in modern mechanical systems involved high-precision synchronized feed technology such as gantry-moving type milling machine is first proposed so as to bring their high-speed dynamic response ability into playfor realizing dynamically synchronized feedConventional vector control algorithm is dependent on the mathematical model of motor.Thus, the control performance of system is influenced to a great extent by the model precision and the linear parameters of motor. Based on artificial neutral network control theory, a MEMO neutral network PID neural network vector control system realizing the maximum torque with minimum current is conformed for a LPMSM drive system. A number of simulation and experiments show that the system not only has good performance both in dynamic and steady state but also avoids the adverse effect resulted from the inaccurate mathematical model and the parameter changes of the LPMSM. Therefore, the system possesses good self-adapting and robustThe innovative ideas in this thesis are that neural network adaptive inverse control theory based on LPMSM control system is applied to the high-precision feed of a single axis, and the controller is composed of inverse model of linear motor. A neural network inverse control system restrains disturbances and uncertainties to keep the robust and stable performances. Theadaptive inverse control system, which has the ability of rapid response, is applied to satisfy the rapid performance. Both the inner system and external system are used to satisfy the required tracking performances.The proposed control scheme has an adaptive inverse control theoretic and neural network base. The controllers designed not only guarantee the stability robustness and performance robustness of the system but also the tracking performance of the system. The simulation results show that the design is reasonable and effective.