神经网络的强耦合控制系统仿真研究

摘要

研究工业控制领域的优化控制问题,工业控制对象具有强耦合特性,传统方法无法对其进行精确解耦,导致系统控制精度比较低.为提高工业控制系统的控制精度,提出一种PID控制和神经网络相融合的控制方法.利用PID优良动态控制特性和BP神经网络非线性控制特性对控制系统进行解耦,在权值调整算法式中加入增大动量项,提高网络学习效率,并采用粒子群算法优化权值初始值,提高控制精度,减少振荡产生.在MATLAB环境下,对非线性控制系统进行仿真研究,仿真结果表明,PID神经网络提高系统的抗干扰能力,提高系统控制精度,能够对系统进行精确解耦,使工业控制系统的性能得到改善.%The traditional control method cannot accurately decouple industry control objects with strong coupling, so the control precision is low. In order to improve the precision, a control method was advanced. The method combines the PID control with neural network. PID control has the excellent dynamic characteristic and BP neural network has the nonlinear control characteristic. The augmenting momentum item was added to improve the network learning efficiency. Meanwhile, particle swarm optimization was adopted to optimize the weight initial value. The method can improve the control precision and reduce the oscillation. The nonlinear control system was simulated by MATLAB. The results show that the PID neural network can improve the anti - interference ability of the system and the control precision. The method achieves precision decoupling and improves the performance of industrial control system.