遗传算法在水轮发电机模糊神经网络励磁控制器设计中的应用研究

Study on application of genetic algorithm in design of fuzzy neural network excitation controller of hydro generator

本文在分析了模糊神经网络(FNN)控制器的工作原理及设计方法的基础上,提出了一种采用遗传算法优化设计水轮发电机模糊神经网络励磁控制器的方法。其基本过程是利用遗传算法得到初始模糊控制规则,并对初始规则进行过滤,在此基础上利用遗传算法结合模拟退火对得到的模糊神经网络进行训练。仿真结果表明与根据专家经验获得模糊规则和BP算法进行学习的常规FNN比较,采用遗传算法优化设计的模糊神经网络励磁控制器所构成的励磁系统具有更好的动态性能。

Based on the analysis of the work theory and design method of the fuzzy neural network(FNN) controller,a new method based on genetic algorithm(GA) for design of fuzzy neural network(FNN) excitation controller of hydro generator is presented in this paper.The basic process is that at first the initial fuzzy rules are acquired and initial fuzzy rules are filtrated by genetic algorithm(GA),then the FNN is trained by genetic algorithm(GA) combining with simulated annealing(SA).The system emulation proves that comparing with the excitation control system acquired by the initial fuzzy rules from expert experience and FNN trained by the BP algorithm,the control system based on genetic algorithm has better dynamic performance.