摘要
针对液压支架前连杆疲劳寿命值计算繁琐的问题,提出采用遗传算法优化后的BP神经网络预测液压支架前连杆疲劳寿命。首先采用最优拉丁超立方方法选取40组参数水平下的设计参数,然后采用ANSYS有限元分析软件计算每个参数水平下的疲劳寿命值,最后采用经遗传算法优化后的BP神经网络学习前30组数据,建立设计参数与疲劳寿命之间的神经网络模型,用剩余的10组数据检测模型预测精度。预测结果表明,遗传算法优化的BP神经网络对液压支架前连杆疲劳寿命的预测误差较小。
For theseover-elaborate calculation of fatigue life of hydraulic support front link,a method of fatigue life prediction method of hydraulic support front link was proposed which adopted BP neural network optimized by genetic algorithm.Design parameters under 40 level are selected by use of Latin Hypercube.Fatigue life value under each parameter level is calculated through ANYSY finite element analysis software.A neural network model between design parameters and fatigue life is established by BP neural network optimized by genetic algorithm learning the first 30 groups of data,and the rest of 10 groups of data is used to verify prediction accuracy of the model.The prediction result shows that the method has small prediction error.
出处
《工矿自动化》
北大核心
2015年第10期46-48,共3页
Journal Of Mine Automation
基金
国家自然科学基金资助项目(61305123)
江苏省学研前瞻性联合研究项目(BY201302505)
关键词
液压支架
前连杆
疲劳寿命
寿命预测
BP神经网络
遗传算法
hydraulic support
front link
fatigue life
life prediction
BP neural network
genetic algorithms
