模糊核聚类在优化模糊控制隶属函数中的应用

摘要

Considering the lack of self-learning ability of fuzzy controller’s membership function designed and optimized,the fuzzy kernel clustering algorithm and Trust-Region optimization method were combined.In which,having fuzzy kernel clustering algorithm used to cluster both input and output samples of fuzzy control and then having Trust-Region optimization method adopted to finish curve-fitting of the clustered results so as to realize fuzzy control’s space division of input and output and the determination of membership function’s type and parameters optimization.The simulation in Matlab shows that the control quality of the fuzzy controller op-timized with the proposed optimizing algorithm has been improved and enhanced greatly.%针对模糊控制器隶属函数的设计与优化缺乏自学习能力的缺点，将模糊核聚类算法与 Trust-Region（信赖域）方法结合起来。首先利用模糊核聚类算法对模糊控制输入、输出样本集进行聚类，然后用 Trust-Region最优化方法对聚类结果进行曲线拟合，实现了模糊控制输入、输出空间的划分、隶属函数类型的确定和参数的优化。在 Matlab中的仿真结果表明：模糊控制器经过笔者提出的算法优化后控制品质有较大的改善和提高。