摘要
[目的]为了提高液压机械无级变速器(HMCVT)换段液压系统的平稳性和安全性,设计了一种应对换段液压故障的诊断方法。[方法]利用自主研发的液压机械无级变速器试验平台,获得5种油路故障状态模式数据集;通过数据预处理及随机抽取的方法得到120组单一故障样本集和21组组合故障样本集,基于天牛须搜索算法(beetle antennae search,BAS)和BP(back propagation)神经网络,对处理后的120组单一故障样本建立了故障诊断模型;对标准BP神经网络模型和优化型BP神经网络模型进行试验测试和比对研究。[结果]所使用的BAS-BP方法对试验样本实现了5种油路状态模式分类;与标准BP神经网络相比,BAS-BP方法可以更好防止网络受限于局部极小值,且故障诊断正确率提升10%。[结论]与常规优化算法相比较,BAS-BP方法所需训练时间短、收敛速度快,算法运行速率提升85.76%,拥有更好的稳定性和判别精度。特别需要指出的是,该方法对于组合故障的判别仍然有效。
[Objectives]In order to improve the stability and safety of the hydraulic system of the hydro-mechanical continuously variable transmission(HMCVT),a diagnosis method to deal with the hydraulic fault of the hydraulic system was designed.[Methods]Using the independently developed hydro mechanical continuously variable transmission test platform,five oil circuit fault state mode data sets were obtained.120 sets of single fault samples and 21 sets of combined fault samples were obtained by data preprocessing and random sampling.Based on the beetle antenna search(BAS)algorithm and BP(back propagation)neural network,a fault diagnosis model was established for 120 sets of single fault samples after processing.The standard BP neural network model and the optimized BP neural network model were tested and compared.[Results]The BAS-BP method used in this paper realized the classification of five oil circuit state patterns for the test samples.Compared with standard BP neural network,BAS-BP method could better prevent the network from being limited by local minimum,and the accuracy of fault diagnosis was improved by 10%.[Conclusions]Compared with the conventional optimization algorithm,BAS-BP method required shorter training time and faster convergence speed.The running speed of the algorithm increased by 85.76%,and it had better stability and discrimination accuracy.In particular,it should be pointed out that this method was still effective for the identification of combined faults.
作者
王家博
张海军
赵余祥
刘永华
肖茂华
鲁植雄
王光明
WANG Jiabo;ZHANG Haijun;ZHAO Yuxiang;LIU Yonghua;XIAO Maohua;LU Zhixiong;WANG Guangming(College of Engineering,Nanjing Agricultural University,Nanjing 210031,China;Jiangsu Vocational and Technical College of Agriculture and Forestry,Jurong 212400,China;College of Mechanical and Electronic Engineering,Shandong Agricultural University,Tai’an 271018,China)
出处
《南京农业大学学报》
CAS
CSCD
北大核心
2023年第3期626-634,共9页
Journal of Nanjing Agricultural University
基金
国家重点研发计划项目(2016YFD0701103)
江苏省现代农机装备与技术示范推广项目(NJ2020-35)
江苏省高校优秀科技创新团队项目(2020kj069)。
关键词
故障诊断
BP神经网络
BAS算法
液压机械无级变速器
fault diagnosis
BP neural network
BAS algorithm
hydro-mechanical continuously variable transmission
