基于遗传算法的最优模糊控制器设计  

Design of Optimized Fuzzy Controller Based on Genetic Algorithm

作　　者：李旭明[1] 康赐荣[1] 

机构地区：[1]华侨大学电子工程系

出　　处：《华侨大学学报（自然科学版）》1999年第1期95-99,共5页Journal of Huaqiao University(Natural Science)

摘　　要：模糊控制规则集是模糊控制系统的核心部分，对控制的快速性和精度有很大影响．采用改进的ＧＡ算法优化ＢＰ网络，加快了ＢＰ网络的收敛速度，也避免了ＢＰ网络易于陷于局部极小的麻烦．用优化的ＢＰ网络生成规则集，经过优化的ＢＰ网络隐层神经元数减少，即优化了规则库，从而提高系统的精度，降低了对ＢＰ网络系统误差的精度要求．以倒立摆模糊控制的仿真为例表明，倒立摆稳定快、精度高，系统稳定时，摆角摆动范围不超过１．７４×１０－７ｒａｄ，系统稳定时间仅需１．５ｓ左右，从而证明方法的有效性．As the core of fuzzy control system, the rule set of fuzzy control exerts great influence on speed and accuracy of control. To optimize BP network by adopting improved GA algorithm will accelerate the convergence of BP network and will avoid to get BP network into local trivial trouble. To generate rule set by using optimized BP network will lead to the optimization of rule library after cutting down the number of neurons at the hidden layer of neurons at the hidden layer of optimized BP network. Thus the accuracy of system will be improved and the accuracy demand on system error of BP network will be lowered. As exemplified by the simulation of fuzzy control of inverted pendulum, which is quick in stabilization and high in accuracy, for a stable system, the angle of swing has a variance within the range of 1.74×10 -7 rad, and the system needs only about 1.5 second to be stabilized. The effectiveness of our method is thus proved.

关 键 词：遗传算法 BP算法 模糊控制器 设计 最优控制 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置] O242.23[自动化与计算机技术—控制科学与工程]