基于概率相关慢特征分析的高速列车牵引系统故障检测  

Fault detection of high-speed train traction system basedon probability related slow feature analysis

作　　者：张瑞婷 翟双[1] 程超[1] 曹文松 ZHANG Ruiting;ZHAI Shuang;CHENG Chao;CAO Wensong(School of Computer Science&Engineering,Changchun University of Technology,Changchun 130102,China)

机构地区：[1]长春工业大学计算机科学与工程学院,吉林长春130102

出　　处：《长春工业大学学报》2023年第3期239-245,共7页Journal of Changchun University of Technology

基　　金：国家自然科学基金(61903047,U20A20186);吉林省科技厅项目(20210201113GX);长春市科技局项目(21GD05)。

摘　　要：利用牵引系统特性提出一种基于概率相关慢特征分析的故障检测方法,以提高故障检测性能。首先,将Kullback-Leibler散度引入慢特征分析方法中,在此方法提取牵引系统数据中慢速特征的基础上,进一步计算离线数据和在线数据慢速特征的概率分布距离,构建统计量。最后使用中南大学的高速列车牵引系统模拟仿真平台,验证了方法的可行性,并对比传统SFA方法,结果表明,提出方法能有效提高检测性能。Using the characteristics of traction system,a fault detection method based on probability related slow feature analysis(PRSFA)is proposed to improve the fault detection performance.Firstly,the Kullback-Leibler divergence is introduced into the slow feature analysis method.On the basis of the slow feature extracted from the traction system data by the slow feature analysis method,the probability distribution distance of the slow feature of offline data and online data is further calculated to construct statistics.Finally,the feasibility of the method is verified by using the simulation platform of the high-speed train traction system of Central South University,and compared with the traditional SFA method,the results show that the proposed method can effectively improve the detection performance.

关 键 词：慢特征分析 Kullback-Leibler散度 故障检测 牵引系统 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]