基于改进遗传算法优化神经网络的板形控制  

Flatness Control based on Improved Genetic Algorithm to Optimize the Neural Network

作　　者：黄江平[1] 林文钊[1] 吴昊[1] 朱志高[1] 

机构地区：[1]华东交通大学电气与电子工程学院,江西南昌330013

出　　处：《有色冶金设计与研究》2011年第4期108-111,共4页Nonferrous Metals Engineering & Research

基　　金：华东交通大学校立科研基金资助(项目编号:10DQ04)

摘　　要：冷轧板形控制系统是一个强耦合、非线性的多变量复杂系统,难以建立精确的数学模型,一般常规的控制方法难以取得令人满意的控制效果。本文依据现场的轧制数据,提出采用自适应竞争遗传算法优化神经网络对其进行建模,采用模糊控制,可实现实时控制,并利用MATLAB编程,仿真结果显示了算法的有效性和时效性。The system of automatic flatness control is a multi-input and multi-output nonlinear system with strong coupling, therefore, it is difficult to build an accurate mathematical model, the control method of general routine is difficult to obtain satisfactory control effect. Based on the local rolling data, this thesis puts forward a method of using neural network which is optimized by adaptive competition genetic algorithm for modeling the cold rolling, then using fuzzy control realize the real-time control, the simulation results of MATLAB programming show the effectiveness and timeliness of the algorithm.

关 键 词：板形控制 遗传算法 神经网络优化 模糊控制 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置] TG333[自动化与计算机技术—控制科学与工程]