基于遗传算法的BP神经网络在塔式起重机故障诊断中的应用  

Application of genetic algorithm-based BP neural network in diagnosis to tower crane

作　　者：尚敬强[1] 原思聪[1] 卫东东[1] 肖畅[1] 

机构地区：[1]西安建筑科技大学,西安710055

出　　处：《起重运输机械》2012年第4期61-64,共4页Hoisting and Conveying Machinery

基　　金：国家十二五科技支撑计划资助(2010BAE00372-2)

摘　　要：为解决BP神经网络存在的收敛速度慢、易陷入局部极小点等不足,利用遗传算法随机搜索性强,易收敛到全局最优解的优点对其进行改进,通过误差平方和的倒数建立遗传算法与BP神经网络的联系,以改进网络权值、阈值为目标,优化神经网络结构,以更加有效地应用改进网络进行预测、识别,将其应用于塔式起重机的状态识别中,可很好地对塔式起重机进行故障诊断。To solve the problems such as slow convergence rate and easily getting the local minimum solutions existing in BP neural network,this paper improves the genetic algorithm taking advantages of its strong random searching capability and easily converging to the global optimum solution,establishes connections between genetic algorithm and BP neural network by reciprocal of error square sum,and with improved network weights and threshold value as an objective optimizes neural network structure in order to more effectively apply the improved network for prediction and identification.When applying to state recognition of tower crane,it can effectively perform diagnosis to tower crane.

关 键 词：塔式起重机 BP神经网络 遗传算法 优化 故障诊断 

分 类 号：TH215[机械工程—机械制造及自动化]