Fuzzy Neural Network In The Ac Servo System

This paper is aimed to improve the performance of an artillery servo system which requires high control qualities such as the speeding, accuracy and stability under hard circumstances like the huge inertia, the great variation of the load and the great impulsion.The main contents of this paper include three parts: the hardware select and design of the servo system, the software programming of the system and the whole system's debugging. The part of the hardware emphasizes on the design of the servo amplifier and the testing of the circuit, then deducing the model of the servo system. Because of the hard circumstance that the system works on, the part of the software emphasizes on attempting and exploring the combining of the fuzzy control with neural net, then bring forward a newly controller which is named as fuzzy-neural net controller , including neural-net for system identification. The arithmetic is programmed with Borland C++. In the end, a whole debugging of the servo system is done.According to the results of the simulation and real experiments, it is known that with the controller which is based on fuzzy and neural net the system can satisfy all targets and performance even under the hard circumstances.