基于改进SVM的车辆传动系统故障诊断方法  

A Fault Diagnosis Method of Vehicle Transmission System Based on Improved SVM

作　　者：马立玲[1] 郭凯杰 王军政[1] MA Li-ling;GUO Kai-jie;WANG Jun-zheng(School of Automation, Beijing Institute of Technology, Beijing 100081, China)

机构地区：[1]北京理工大学自动化学院,北京100081

出　　处：《北京理工大学学报》2020年第8期856-860,共5页Transactions of Beijing Institute of Technology

摘　　要：利用车辆传动系统试验数据对车辆进行故障诊断和性能评价可以实现车辆故障预警,提高可靠性,从而提高车辆性能,但测试数据有数据量大、不平衡、维度高、噪声多的特征,使得传统数据分析算法会产生次优的分类模型.针对上述问题,提出了一种改进的不平衡数据分类支持向量机算法.该算法赋予各样本不同的权值,用马氏距离改进模糊隶属度的设计以排除变量相关性干扰,同时可以输出正常状态下的故障概率.实验结果表明,该算法能够有效提高故障诊断的准确性,概率输出模型可用于故障预警和性能分析.Fault diagnosis and performance evaluation with vehicle transmission system test data can play a role in fault warning,improving reliability,and further improving vehicle performance.However,the test data are very large and unbalanced,possess high dimensionality and noise,which make the traditional data analysis algorithm produce sub-optimal classification model.In order to solve the above problems,a new improved support vector machine(SVM)algorithm was proposed for imbalanced data classification.The algorithm was arranged to present different weights for each sample,improve the design of fuzzy membership degree with Mahalanobis distance to eliminate the interference of variable correlation,and to output the failure probability under normal state at the same time.The experimental results show that the algorithm can effectively improve the accuracy of fault diagnosis,and at the same time can use the probability output model to carry out fault warning and performance analysis.

关 键 词：支持向量机 不平衡数据 概率输出 模糊隶属度 性能分析 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]