基于粒子群算法优化PIDNN的流浆箱解耦控制  

Decoupling control of headbox with PIDNN based on particle swarm optimization

作　　者：陈帅帅[1] 熊智新[1] 胡慕伊[1] 

机构地区：[1]南京林业大学江苏省制浆造纸科学与技术重点实验室,南京210037

出　　处：《中华纸业》2017年第6期20-24,共5页China Pulp & Paper Industry

基　　金：国家林业局"948"项目"农林剩余物制机械浆节能和减量技术引进"

摘　　要：针对气垫式流浆箱的总压和浆位之间存在严重耦合的情况,探讨使用粒子群算法优化的PID神经元网络对其进行解耦控制。建立一个双输入双输出PID神经元网络控制系统,用粒子群算法优化网络的初始权重后,再采用误差反向传播算法进行修正,最终实现完全解耦。仿真结果表明,经粒子群算法优化后的神经元网络能够有效地消除总压与浆位之间的耦合,大大缩短调试时间,具有更高的控制精度,能够用于气垫式流浆箱的控制。Aiming at the coupling problem between total pressure and liquid level in air-cushioned headbox control system,this paper investigated the feasibility of using PID neural network decouple controller which optimized with particle swarm optimization.A double input two output PID neural network control system was established,and the initial weights of the network had been optimized by PSO and amended by error back propagation algorithm.Finally,the decoupling control of aircushioned headbox was realized.The simulation results show that the PSO-PIDNN controller can effectively reduce the coupling influence between total pressure and liquid level,and reduce the debug time while improve control precision,thus it’s qualified for the air-cushioned headbox control.

关 键 词：气垫式流浆箱 解耦控制 PID神经元网络 粒子群算法 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置] TS734.2[自动化与计算机技术—控制科学与工程]