基于人工鱼群的PID神经网络非线性系统控制  

Nonlinear Control System of PID Neural Network Based on Artificia Fish Swarm Algorithm

作　　者：冯冬青[1] 马超阳[1] 刘艳红[1] 

机构地区：[1]郑州大学电气工程学院,河南郑州450001

出　　处：《计算机仿真》2012年第9期247-250,296,共5页Computer Simulation

基　　金：国家自然科学基金项目(60974005);河南省教育厅自然科学研究资助项目(2010B510019)

摘　　要：研究造纸工业中的流浆箱非线性系统优化控制问题。流浆箱系统是工业过程中常见的一类非线性系统,存在着非线性、强耦合等特性。针对流浆箱要求动态响应好、精度高的特点,提出并设计了人工鱼群算法训练的PID神网络控制器。人工鱼群算法克服了PID神经网络采用BP算法训练权值时,初始权值难以确定,易陷入局部最优的缺点,实现对流浆箱的有效控制。在MATLAB环境下,对流浆箱系统进行了控制仿真。仿真结果表明,人工鱼群算法训练的PID神经网络在动态性、稳定性和精确性等方面均优于BP算法,明显改善了流浆箱这类非线性系统的控制性能,具有很好的应用效果。The control problem of the air-cushioned headbox in papermaking process was studied. The system of air-cushioned headbox is an usual nonlinear system in industrial processes, which is of the the nonlinearity and seri-ous coupling. The air-cushioned headbox demands good dynamic quality and high control precision. In this paper, the PID neural network controler based on the artificia fish swarm algorithm（AFSA） was used in the control system. The AFSA can overcome the shortcomings of the PID neural network training with BP algorithm, such as, the initial weight value is hard to be decided and the training is easily fall in the local minimum. The simulation of the nonlinear control system was presented with MATLAB, and the result shows that the PID neural network based on AFSA is bet-ter than the BP algorithm in rapidity, stability and accuracy. It makes the air-cushioned headbox better in control performances and effect.

关 键 词：流浆箱 神经网络 人工鱼群算法 反向传播学习算法 非线性控制系统 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TP202.7[自动化与计算机技术—控制科学与工程]