基于混沌粒子群优化算法的热连轧厚度控制  

Thickness Control of Hot Strip Mill Based on Chaos Particle Swarm Optimization Algorithm

作　　者：周建新 李建坤 

机构地区：[1]华北理工大学电气工程学院,河北唐山063000

出　　处：《工业控制计算机》2016年第12期76-77,共2页Industrial Control Computer

基　　金：河北省教育厅重点项目(ZD2015059)

摘　　要：热连轧厚度控制一向是权衡板带钢质量的重要因素,是一类非线性、时变性、大时滞的典型对象。常规的串级PID厚度控制系统在干扰大、工矿发生变化时,很难到达理想状态。为了解决这种普遍存在的问题,构造了一种调节惯性权重的自适应混沌粒子群算法,即基于已知的早熟收敛程度和个体本身的适应值的新方法,充分利用混沌优化与粒子群优化的搜索特征,训练神经网络的权值阈值,解决了BP神经网络收敛速度慢和易陷入局部极值的问题,从而实现了对热连轧厚度控制的精度,仿真结果表明与传统PID控制相比较,该算法不仅提高了误差精度,同时也加快了训练收敛的速度,该控制方法稳定可行。Conventional thickness control system of cascade PID can＇t reach the ideal state,when it is strongly disturbed or working condition changed.Because of this widely existing problem,this paper construct an adaptive chaotic particle swarm optimization algorithm which can adjust inertia weight,which is a new method based on the known premature convergent degree and individual adaptive value.It fully use the search characteristic of chaos optimization and particle swarm optimization and train the weights and threshold of neural network.It solves the problem of BP neural network that converging too slow and easily getting into the local extremum and realize a precision gauge control of hot strip mill.

关 键 词：厚度控制 混沌 粒子群优化 神经网络 

分 类 号：TG334.9[金属学及工艺—金属压力加工] TP18[自动化与计算机技术—控制理论与控制工程]