摘要
为了解决道岔设备智能故障诊断中特征指标难以提取以及模型训练时间较长的问题,以ZDJ9型转辙机带动的道岔设备为研究对象,以转辙机功率曲线为数据基础,提出一种基于参数优化变分模态分解(Variational Mode Decomposition,VMD)和改进最小二乘支持向量机(Least Squares Support Vector Machines,LSSVM)的道岔故障诊断方法。首先,采用鲸鱼优化算法(Whale Optimization Algorithm,WOA)优化VMD参数,得到模态(Intrinsic Mode Functions,IMF)分量个数和惩罚因子的最优参数组合。其次,计算IMF分量与功率曲线的相关系数,优选相关性较大的前3阶IMF分量,并计算功率谱熵、模糊熵及包络熵值,建立多特征融合样本数据库。最后,针对麻雀搜索算法(Sparrow Search Algorithm,SSA)易陷入局部最优的问题,通过改进Tent混沌映射初始化策略随机生成种群,正余弦算法(Sine Cosine Algorithm,SCA)更新追随者的位置,并采用改进SSA优化LSSVM算法的惩罚因子和核函数方差,构建基于TSSSA-LSSVM的道岔故障诊断模型。实验结果表明:所提道岔故障诊断方法是可行的,采用多特征融合能够更加全面地提取道岔典型故障特征,反映道岔的真实运行状态,提高了故障诊断准确率,而且较TSSSA-SVM,PSO-LSSVM,GWO-LSSVM以及SSA-LSSVM等方法具有较高的故障诊断准确率、召回率以及较低的漏报率,减少了模型训练时间,完全满足现场道岔故障导向安全的原则,具有更好的故障诊断性能,对现场道岔设备的故障维修具有一定的指导意义。
In order to solve the problem that the feature index was difficult to extract and the model training time was long,the intelligent fault diagnosis of turnout equipment driven by ZDJ9 switch machine was taken as the research object.The power curve of switch machine was taken as the data basis,a turnout fault diagnosis method based on parameter optimization Variational Mode Decomposition(VMD)and improved Least Squares Support Vector Machines(LSSVM)was proposed.Firstly,the Whale Optimization Algorithm(WOA)was used to optimize the VMD parameters,and the optimal parameter combination of the number of Intrinsic Mode Functions(IMF)components and the penalty factor was obtained.Secondly,the correlation coefficient between the IMF component and the power curve was calculated.The first three IMF components with large correlation were selected.The power spectral entropy,fuzzy entropy and envelope entropy were calculated.The sample database of multi-feature fusion was constructed.Finally,aiming at the problem that Sparrow Search Algorithm(SSA)was easy to fall into local optimum,the population was randomly generated by the improved tent chaotic map initialization strategy,and the follower position was updated by the positive chord algorithm.The improved SSA algorithm was used to optimize the penalty factor and kernel function variance of the LSSVM algorithm,and the TSSSA-LSSVM fault diagnosis model of high speed railway turnout was constructed.The experimental results show that the turnout fault diagnosis method proposed in this paper is feasible.The multi-feature fusion can extract the typical fault features of the turnout more comprehensively,reflect the real operating state of the turnout,and improve the accuracy of fault diagnosis.Compared with TSSSA-SVM,PSO-LSSVM,GWOLSSVM and SSA-LSSVM,it has higher fault diagnosis accuracy,recall rate and lower false negative rate,reduces the model training time,fully meets the principle of on-site turnout fault-oriented safety.It has better fault diagnosis performance,which has certain guiding significance for the fault maintenance of on-site turnout equipment.
作者
王彦快
孟佳东
张玉
杨建刚
WANG Yankuai;MENG Jiadong;ZHANG Yu;YANG Jiangang(School of Railway Technical,Lanzhou Jiaotong University,Lanzhou 730070,China;School of Mechatronic Engineering,Lanzhou Jiaotong University,Lanzhou 730070,China;State Grid Gansu Electric Power Research Institute,Lanzhou 730070,China;China Railway Signal&Communication Corp,Beijing 100070,China)
出处
《铁道科学与工程学报》
EI
CAS
CSCD
北大核心
2024年第5期2072-2085,共14页
Journal of Railway Science and Engineering
基金
中央引导地方科技发展资金项目(22ZY1QA005)
甘肃省科学计划项目(21JR7RA305)
兰州交通大学青年科学研究基金(2021027)。
作者简介
通信作者:张玉(1983-),男,安徽蚌埠人,高级工程师,从事高铁信号设备的故障监测与健康管理研究,E−mail:1782667905@qq.com。
