基于改进学习算法的模糊神经网络控制系统

针对一类复杂非线性系统，提出一种模糊神经网络(FNN)控制方案。系统中采用模糊神经网络控制器和神经网络辨识控制器相结合的结构，介绍一种改进的学习算法，对学习公式进行推导，利用改进的遗传算法来优化已经获得的隶属度函数，并结合误差补偿以提高控制精度。同时将混沌机制引入常规BP算法，利用混沌机制固有的全局游动，逃出权值优化过程中存在的局部极小点，解决了网络训练易陷入局部极小点的问题。用该方法对某非线性动态系统进行辨识和控制，仿真结果表明控制精度和实时性优于常规模糊控制器。

A Fuzzy Neural Network Control System Based on Improved Learning Algorithms

A fuzzy neural network(FNN)control scheme for a class of complicated nonlinear systems was presented.In this scheme it has the structure that combines a FNN controller with neural network identification controller,a new improved learning algorithm was derived theoretically.Based on the error-compensation method and using the modified genetic algorithm for optimizing the membership functions,the accuracy of the algorithm was improved.Then chaotic mechanism was introduced to normal BP algorithm,and the problem of local limit value for network was solved by using global moving characteristic of chaotic mechanism.The simulation results show that this design has a better performance than normal fuzzy controller.