Design of hierarchical fuzzy systems based on subtractive clustering and adaptive neuro-fuzzy inference systems

提出了一种设计递阶模糊系统的简易而有效的方法.在得到一个单级模糊系统的基础上,用灵敏度分析法对每一个输入变量的重要性进行排序,从而确定每一级子系统的输入变量.利用减法聚类和自适应神经 模糊推理系统逐级对子系统进行训练.所得到的递阶模糊系统可进一步得到简化.仿真实例证实了设计方法的有效性.An easy and effective method to design hierarchical fuzzy systems is presented.The degree of importance of each input variable was obtained using sensitivity analysis method based on a single stage fuzzy model.After ranking of importance of each input variable,input variables of every subsystem of the hierarchical fuzzy system can be determined.Every subsystem was trained from the first stage to the last stage using subtractive clustering and ANFIS (adaptive neuro-fuzzy inference systems).A method to reduce the hierarchical fuzzy system was proposed.The design method was proved to be feasible.福建省自然科学基金项目(A0110002)