基于前向模糊神经网络 ANFIS提出了一种新型的动态模糊神经网络 ( DFNN) ,将模糊逻辑、神经网络和 PID控制器三者的优点有机地融合在一起 .通过在 ANFIS的归一化层和输出层之间加入递归层 ,构成了动态模糊神经网络 ( DFNN) ,并推导了基于 BP的反传学习算法 .与 ANFIS和 PID控制器相比 ,DFNN具有更好的控制效果 .DFNN的参数具有明确的物理意义 ,可根据专家的经验选择初值 ,加快了网络的收敛速度 ;由于 DFNN为动态神经网络 ,从而具有更强的处理动态系统的能力 A new type of dynamic fuzzy neural network (DFNN) is proposed based on the forward fuzzy neural network ANFIS, which combines the advantages of fuzzy logic, neural network and PID controller organically.Through the normalization layer of ANFIS and The recursive layer is added between the output layers to form a dynamic fuzzy neural network (DFNN), and a back propagation learning algorithm based on BP is derived. Compared with ANFIS and PID controllers, DFNN has better control effect. The parameters of DFNN have Clear physical meaning, according to the expert’s experience choose initial value, speed up the convergence of the network; DFNN dynamic neural network, which has a stronger ability to handle dynamic systems