基于神经网络的PID网络化控制系统建模与仿真

Modeling and Simulation of PID Networked Control Systems Based on Neural Network

针对网络化控制系统中的时延补偿问题和PID控制器参数整定问题,提出了一类基于改进BP网络的快速PID网络化控制系统。考虑BP网络时延预测模型求取隐含层节点数耗时问题和训练方法的局部最优问题,分别提出了基于计算公式的隐含层节点数求取方法和基于改进遗传算法的BP网络训练方法;考虑PID网络化控制系统中的积分饱和问题、控制器参数整定问题和应对模型失配的能力,设计了基于在线BP网络的PID参数调节器。仿真结果表明:该系统运行速度较快,具有应对模型失配的能力,且在随机短时延和随机长时延情况下均有良好的控制性能。

According to the problems of the delay compensation and PID parameters tuning of networked control systems, a class of rapid PID networked control systems based on improved BP network was proposed. Considering the problems of obtaining hidden layer nodes number and local optimum of the BP network delay prediction model, a calculation method was proposed to obtain hidden layer nodes number, and an improved genetic algorithm was proposed to train the BP network. Considering the problems of integral saturation, parameters tuning and model mismatch of the PID network control systems, a PID parameter adjuster was designed based on on-line BP network. The simulation results show that this system has the advantage of rapid running speed, coping model mismatch, and good control performance with random short delay or random long delay.