模糊建模与控制的神经网络方法及其仿真研究

Simulation Study on Neural Network Approach to Fuzzy Modeiling and Control

本文提出了一种复杂系统模糊建模及控制的神经网络方法。利用改进的ｐｉ－ｓｉｇｍａ神经网络，对模糊规则的结论参数和隶属函数进行在线修正，实现模糊规则的自组织。这种方法被用于降水量预报和机器人解耦控制，取得了满意的仿真结果。为增强神经网络仿真算法的快速性，本文采用了一种基于向量的数据结构，并用标识阵指示神经元的连接状态，以实现有效的内存运算。

A neural network approach to fuzzy modelling and control has been proposed.The newscheme is based on an updated pi-sigma neural network and can realize on-line fuzzy rule self-organiza-tion.The new method is applied to rainfall prediction and robot control.In both cases,satisfyingsimulation results are obtained.A vector-based neural network data structure and an index array whichindicates the connection state of the neurons are adopted to increase the simulation speed.