摘要
为实时监控类矩形盾构偏心刀盘工作状态,提出一种基于遗传算法(GA)优化BP神经网络模型的在线故障预测方法。首先,利用现场检测的相关测量数据,建立“特征—故障”数据集;然后,利用最优权值与阈值由遗传算法获取的BP神经网络对数据集进行自我学习,构建工作期故障预测模型;最后,开发偏心刀盘监控系统,对刀盘工作状态进行在线预测。实验结果表明,GA-BP网络模型预测准确率达到93.3%,与传统BP网络模型相比提高6%。基于GA-BP网络的偏心刀盘在线故障预测方法可精准预测刀盘工作状态,满足应用设计要求,为盾构施工安全提供有力保障。
In order to effectively monitor the working status of eccentric cutterhead of quasi-rectangular shields,an online failure prediction method based on genetic algorithm to optimize BP neural network model is proposed.First,the“feature-failure”data sets are established using the relevant measurement data of the field inspection.Then,the BP neural network which obtains the optimal weights and thresholds through the genetic algorithm,performs self-learning on the data sets to construct a failure prediction model during the work period.Finally,the eccentric cutterhead condition monitoring system is developed to predict the working state of the cutterhead online.The experimental results show that the accuracy of GA-BP network model is 93.3%which is 6%higher than that of traditional BP network model.The on-line fault prediction method of eccentric cutterhead based on GA-BP can accurately predict the working state of cutterhead,meet the requirements of application design and provide strong guarantee for shield construction safety.
作者
刘高宏
吴恩启
闵锐
侯天凡
王晓辉
LIU Gao-hong;WU En-qi;MIN Rui;HOU Tian-fan;WANG Xiao-hui(School of Mechanical Engineering,University of Shanghai for Science and Technology,Shanghai 200093,China;Mechanical Engineering Company,Shanghai Tunnel Engineering Co.Ltd.,Shanghai 200137,China)
出处
《软件导刊》
2020年第10期111-115,共5页
Software Guide
基金
国家自然科学基金项目(51976130)。
关键词
偏心刀盘
故障预测
遗传算法
BP神经网络
eccentric cutterhead
failure prediction
genetic algorithm
BP neural network
