基于改进JSOA-SVM的地铁站台门故障诊断  

Fault Diagnosis of Metro Platform Doors Based on Improved JSOA-SVM

作　　者：王若凡 朱松青[1] 杨柳 郝飞[1] 徐涛 WANG Ruofan;ZHU Songqing;YANG Liu;HAO Fei;XU Tao(School of Mechanical Engineering,Nanjing Institute of Technology,Nanjing 211167,China)

机构地区：[1]南京工程学院机械工程学院,南京211167

出　　处：《噪声与振动控制》2025年第2期112-117,125,共7页Noise and Vibration Control

基　　金：国家自然科学基金青年科学基金资助项目(52005248);江苏省高等学校基础科学资助项目(自然科学)研究重大项目资助(23KJA460009);大学生科技创新基金资助项目(TB202317007)。

摘　　要：为准确地对地铁站台门进行故障诊断,并针对支持向量机(Support Vector Machine,SVM)在故障诊断中的参数选择问题,将跳蛛算法(Jumping Spider Optimization Algorithm,JSOA)用于SVM参数优化提升诊断性能,同时针对JSOA易陷入局部最优、收敛速度慢等不足,提出一种多策略改进跳蛛算法(Improved Jumping Spider Optimization Algorithm,IJSOA)优化SVM的站台门故障诊断方法。首先使用Teager能量算子、变分模态分解(Variational Mode Decomposition,VMD)以及精细复合多尺度模糊熵(Refined Composite Multiscale Fuzzy Entropy,RCMFE)提取信号特征;其次,通过IJSOA寻找SVM最优参数组合构建诊断模型;最后,使用提取的特征向量输入诊断模型实现站台门故障诊断。结果表明提出方法平均识别率为97.774%,诊断精度较其余几种方法更具优势,能够有效提升故障诊断分类效果。In order to accurately diagnose the fault of Metro platform doors and to solve the problem of parameter selection in fault diagnosis of support vector machine(SVM),the jumping spider optimization algorithm(JSOA)was used to optimize the parameters of SVM to promote its diagnostic performance.In view of the shortcomings of JSOA,such as easy to fall into local optimization and slow rate of convergence,A multi-strategy Improved Jumping Spider Optimization Algorithm(IJSOA)was proposed to optimise the SVM for platform door fault diagnosis.Firstly,Teager energy operator,variational mode decomposition(VMD)and refine composite multi-scale fuzzy entropy(RCMFE)were utilized to extract signal features.Secondly,the optimized JSOA algorithm was utilized to find the optimal parameter combination for SVM and construct the diagnostic model.Finally,the extracted feature vectors were input into the diagnostic model to achieve fault diagnosis for the platform doors.The results show that the average recognition rate of the proposed method is 97.77%,which is more accurate than other methods and it can effectively improve the classification effect of fault diagnosis.

关 键 词：故障诊断 地铁站台门系统 变分模态分解(VMD) 跳蛛优化算法(JSOA) 支持向量机(SVM) 

分 类 号：TH17[机械工程—机械制造及自动化] TP18[自动化与计算机技术—控制理论与控制工程]