基于动态结构神经网络的自组织模糊控制

Self-organized fuzzy control based on neural network of dynamic structure

给出了一种利用动态结构的BP神经网络进行模糊推理的方法 ,即在学习过程中BP网络的隐层层数和维数不断变化 ,其目的在于使网络性能在训练过程中始终保持最优 ,从而实现规则自组织 ,达到在控制过程中不断优化控制性能的目的。

This paper presents a kind of BP neural network of dynamic structure. It can inference ,that is, the numbers of layers and dimensions vary during it studies. Its aim is to make the capability of network is always optimization. So the rules are self-organized. Thus the control capability is optimization in the course of control.