一种新型神经网络结构的PID控制器及其仿真研究

DESIGN AND SIMULATION OF A NEW TYPE NEURAL STRUCTURAL PID CONTROLLER

出了一种新型的神经网络结构的ＰＩＤ控制器（Ｎ－ＰＩＤ），该控制器将神经网络和ＰＩＤ控制规律融为一体，既具有常规ＰＩＤ控制器结构简单、参数物理意义明确之优点，又具有神经网络自学习、自适应之能力．还给出了Ｎ－ＰＩＤ控制器的结构、计算公式及二阶对象下系统的仿真结果．

This paper describes a new type neural structural proportional-integral-derivative (N-PID) controller. The N-PID is a combination of the neural network and the PID rule. It has an input layer, a hidder layer and an output layer, and its hidder layer units are proportional, integral and derivative units. The N-PID has advantages of both neural network and PID algorithm. It is proved by computer simulation program that N-PID controller has very good dynastic and static properties.