基于BP神经网络的仿人智能控制仿真研究

Human-Simulated Intelligent Control Research Based on the BP Neural Network

在仿人智能控制算法的闭环阶段,须调整比例、积分和微分三种控制作用,形成控制量既相互配合又相互制约的关系,这种关系不一定是简单的“线性组合”,从变化无穷的非线性组合中可以找出最佳的。而BP神经网络所具有的任意非线性表达能力,可以通过对系统性能的学习来实现具有最佳组合的过程控制。在对非线性系统的数值仿真中,取得了较好的阶跃响应。

At the closed loop stage of human-simulated intelligent control algorithm, you must adjust the proportion, the integral and the differential control actions, to form the control quantity both mutually to coordinate the relations and mutually restricts, this kind of relations is not necessarily the simple＂ linear combination＂, may discover the best from the countless changes of those non-linearity combination. But the BP nerve network has the free non-linear expression ability, and it can realize the best combination of the process control through the system performance study. In the nonlinear system value simulation, it obtained a better step respond.