摘要
在基于 T- S模型的模糊神经网络的基础上 ,提出了一种将混沌优化方法和最小二乘法相结合的优化方法。用变尺度混沌优化方法优化隶属函数参数 ,而用最小二乘法估计规则后件参数。该方法同时利用了变尺度混沌优化的快速全局搜索能力和最小二乘法的快速收敛性 ,因此网络学习速度快 ,精度高。仿真结果表明了该方法的有效性 ,所建立的模型具有良好的泛化能力。
Based on fuzzy neural network(FNN) of Takagi Sugeno type, a new optimization method is proposed, which combines mutative scale chaos optimization algorithm(MSCOA) and least square estimation(LSE). MSCOA is used to optimize the parameters of membership functions, and LSE to estimate the parameters of consequence. Because of the fast search ability of MSCOA and fast convergency of LSE, FNN learns fast with high accuracy. Simulation results demonstrate the effectiveness of this method and generalization ability of the FNN model.
出处
《华东理工大学学报（社会科学版）》
2002年第S1期27-29,33,共4页
Journal of East China University of Science and Technology：Social Science Edition
关键词
模糊控制
模糊神经网络
混沌优化
最小二乘法
fuzzy control
fuzzy neural network
chaos optimization algorithm
least square estimation
