基于有限元和RBF神经网络的液压支架前连杆疲劳寿命预测
Fatigue-life prediction of hydraulic support's front connecting rod based on finite-element analysis and RBF neural network
许志鹏 1刘婵 2冯红翠2
作者信息
- 1. 盐城工业职业技术学院汽车与交通学院,江苏盐城 224005
- 2. 南京工业职业技术大学电气工程学院,江苏南京 210023
- 折叠
摘要
针对工程应用中液压支架前连杆疲劳寿命预测的需求,提出了一种基于RBF神经网络的疲劳寿命预测方法,选取5个关键设计参数为输入量,以前连杆疲劳寿命为目标函数建立了疲劳寿命预估模型.首先,运用有限元分析获得前连杆的疲劳寿命,再通过优化拉丁采样的方法获得训练样本点,并以此建立前连杆疲劳寿命的RBF神经网络预估模型,通过优化RBF神经网络的目标值和扩散值提高模型的预估精度.结果表明:优化后的前连杆疲劳寿命预估模型计算结果与测试样本点拟合精度较高,平均相对误差为6.72％,满足工程目标,适当增加训练样本点的数量有利于进一步提高疲劳寿命的预估精度.
Abstract
Since it is necessary to predict the fatigue life of the hydraulic support's front connecting rod in engineering appli-cation,in this article,a method of fatigue-life prediction is proposed based on the RBF neural network.Five key design parame-ters are selected as input variables;with the front connecting rod's fatigue life as the objective function,the model of fatigue-life prediction is set up.Firstly,the finite-element analysis is used to work out the front connecting rod's fatigue life.Then,the training sample points are obtained by means of optimized Latin sampling,and the model of predicting the front connecting rod's fatigue life is set up based on the RBF neural network.The model's prediction accuracy is improved by optimizing the RBF neu-ral network's goal and spread values.The results show that the results obtained from the optimized model of predicting the front connecting rod's fatigue life has high fitting accuracy with the training sample points,with the relative error of 6.72％on aver-age,which meets the related requirements of engineering application.Properly increasing the number of training sample points is beneficial to further improve the accuracy in fatigue-life prediction.
关键词
液压支架/疲劳寿命预测/神经网络/疲劳性能Key words
hydraulic support/fatigue-life prediction/neural network/fatigue performance引用本文复制引用
基金项目
江苏省高等学校自然科学研究面上项目(21KJB480007)
江苏风力发电工程技术中心开放基金项目(ZK22-03-06)
江苏省产学研合作项目揭榜挂帅(BY20221395)
&&(2023HX-13)
出版年
2024
NETL
NSTL
万方数据