论文部分内容阅读
1 引言神经网络自学习控制代表性结果之一 [1 ] 是采用一个多层前馈网络 (对象仿真器 )辨识被控对象 ,采用另一个神经网络 (控制器 )学习控制仿真器后 ,再控制真实对象 .这个思想对于解决复杂工业过程对象优化控制问题具有一定意义 .但是 ,直接应用仍存在问题 :1)其中对象仿真
I. INTRODUCTION One of the representative results of neural network self-learning control [1] is the identification of the controlled object using a multi-layer feedforward network (object simulator). After another neural network (controller) learns the control simulator, Real object.This idea has certain significance to solve the problem of the optimization control of complex industrial process objects.However, there are still some problems in the direct application: 1) The object simulation