基于改进PSO神经网络冷轧带钢厚度控制仿真研究  

Simulation of Cold Rolled Strip Thickness Control Based on Improved PSO Neural Network

作　　者：朱亮宇 赵志挺 高珣洋 ZHU Liangyu;ZHAO Zhiting;GAO Xunyang(School of Mechanical and Power Engineering,Shenyang University of Chemical Technology,Shenyang 110142,China)

机构地区：[1]沈阳化工大学机械与动力工程学院,沈阳110142

出　　处：《机械工程师》2024年第8期49-53,共5页Mechanical Engineer

基　　金：国家自然科学基金项目(52074085,U21A20117,U21A20475);中央高校基础研究基金项目(N2004010)。

摘　　要：在面对变时滞、强耦合的非线性系统时,传统的比例-积分-微分控制器存在参数整定困难和控制精度不高的实际问题。为了解决这些问题,提出了一种使用粒子群算法优化反面传递的方法,该方法可以有效地提高控制精度,并且可以有效地减少参数整定的困难,从而提高控制的效率。神经网络PID控制是BP的一种重要方法,这种方法能够实时调整PID系数值,大大改善了BP-PID系统的控制特性,从而有效地改善了系统的控制速度和控制精度。将改进的粒子群算法训练的神经网络应用于冷轧带钢厚度控制。仿真结果表明,与普通粒子群算法相比较,该算法响应速度快、控制误差精度高,具有较好的动态特性和系统稳定性。In the face of nonlinear systems with time-varying delay and strong coupling,the traditional proportional-integral-differential controller has practical problems of difficult parameter tuning and low control accuracy.In order to solve these problems,this paper proposes a method of using particle swarm optimization algorithm to optimize the reverse transfer.This method can effectively improve the control accuracy,and can effectively reduce the difficulty of parameter to improve the efficiency of control.BP neural network PID control is an important method of BP.This method can adjust the PID coefficient in real time,greatly improve the control characteristics of BP-PID system,and effectively improve the control speed and control precision of the system.The neural network trained by the improved particle swarm optimization algorithm is applied to the thickness control of cold rolled strip steel.The simulation results show that compared with the ordinary particle swarm optimization algorithm,the algorithm has fast response speed,high control error precision,good dynamic characteristics and system stability.

关 键 词：冷轧机 带钢厚度 IPSO 神经网络 时滞 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置]