基于GA-PID神经网络的板形模式识别方法  

Method of Flatness Pattern Recognition based on GA-PID Neural Network

作　　者：张秀玲[1,2] 徐腾[1] 赵亮[1] 樊红敏[1] 臧佳音[1] 

机构地区：[1]燕山大学河北省工业计算机控制工程重点实验室,河北秦皇岛066004 [2]燕山大学国家冷轧板带装备及工艺工程技术研究中心,河北秦皇岛066004

出　　处：《沈阳大学学报（自然科学版）》2013年第3期209-215,共7页Journal of Shenyang University：Natural Science

基　　金：国家自然科学基金资助项目(50675186)

摘　　要：针对传统板形模式识别方法存在抗干扰能力差和识别精度有限等缺点,提出了基于遗传算法(GA)优化的PID神经网络板形模式识别方法.PID神经网络不仅具备传统多层前向网络的特点,而且其隐含层具有动态特性,可以直接用于动态系统辨识.GA具备良好的并行设计结构,具有全局优化的特点,利用GA对网络权值进行优化,克服了传统BP算法易陷于局部极小的不足.仿真结果表明:GA-PID神经网络的板形模式识别方法能够识别出常见的板形缺陷,提高了板形模式识别精度,可以满足板带轧机高精度的板形控制要求.For the traditional methods has the shortages of poor anti-interference ability and limited recognition accuracy, flatness pattern recognition method via PID neural network optimized by genetic algorithm （GA） is proposed. PID neural network not only has the same characteristics as the traditional multi-layer forward network and the hidden layer also has dynamic characteristics, it can be directly used for the identification of dynamic systems. GA has good parallel design structure and characteristics of global optimization. GA, instead of BP algorithm, was used to optimize the weights of PID neural network in order to overcome the shortage of easy trapped in local minimum of BP algorithm. The simulation results show that flatness pattern recognition method via PID neural network optimized by GA can identify common shape defects, improve the accuracy of flatness pattern recognition and meet the high-precision requirements of flatness control for strip mill.

关 键 词：板形 模式识别 PID神经网络 遗传算法 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]