【摘要】 提出了基于椭圆基函数(EBF)的广义动态模糊神经网络(GD-FNN)算法。算法提取的模糊规则具有很好可理解性,可以作为建模工具,也可以作为知识提取的工具。广义动态模糊神经网络由于基于模糊ε-完备性,同时提出了一种新颖的在线参数分配机制,从而缓解了初始化的随机选择,且与输入变量不同值域没有关系,因而更容易构造一个较好性能的模糊系统。开发了仿真程序,对具体案例进行仿真,取得了较为理想的结果。更多还原

【Abstract】 General dynamic fuzzy neural network( GD-FNN) algorithm is proposed based on the elliptic basis function( EBF). Fuzzy rules generated from the algorithm are intelligibility. It can be used as a modeling tool. and a tool of knowledge extraction. Because of a novel on-line parameter allocation mechanism for allevialing the random selection in initialization without relation to different input variable range,the proposed GD-FNN based on fuzzy ε-completeness is more easy to construct a good fuzzy system in performance. The simulation program is also developed based on the GD-FNN algorithm and ideal results are achieved by simulation in specific design case.更多还原