结合S变换和mRMR特征选择的电缆早期故障识别方法  

AN INCIPIENT CABLE FAILURES IDENTIFICATION METHOD BASED ON S TRANSFORM COMBINED WITH MRMR FEATURE SELECTION

作　　者：王森 龚俊 杨晓梅[1] Wang Sen;Gong Jun;Yang Xiaomei(College of Electrical Engineering,Sichuan University,Chengdu 610065,Sichuan,China;Chengdu Product Quality Supervision and Inspection Institute,Chengdu 610100,Sichuan,China)

机构地区：[1]四川大学电气工程学院,四川成都610065 [2]成都市产品质量监督检察院,四川成都610100

出　　处：《计算机应用与软件》2022年第1期206-211,265,共7页Computer Applications and Software

摘　　要：电力电缆早期故障严重威胁用电安全且难以准确识别,在基于特征提取与特征选择的识别方法中,一旦不能准确获得关键特征信息会直接导致识别精度下降。鉴于此,提出一种基于S变换特征提取和最大相关最小冗余(mRMR)特征选择的电缆早期故障识别方法。对故障相电流进行S变换,提取一些具有相关性、冗余性的统计量、熵和能量等构成初始特征集;采用mRMR选择出具有最佳分类效果的特征子集;利用带核函数的SVM分类器对多种电缆故障进行识别。仿真结果表明,在不同噪声环境下该方法在识别精度和鲁棒性方面都优于同类算法。Incipient failures of power cable is a serious threat to power safety and difficult to be accurately identified.In the recognition based on feature extraction and feature selection,once the key feature information cannot be obtained accurately,the recognition accuracy will be decreased directly.In view of this,this paper proposes an incipient cable failures identification method which combines S transform based on feature extraction with the maximum-relevance-minimum-redundancy(mRMR)feature selection.S transform of the fault phase current was used to extract characteristics to form the primary feature set with correlation and redundancy,e.g,some statistics,entropy and energy.The mRMR algorithm was used to obtain the feature subsets that were helpful for incipient failures classification.The SVM classifier with kernel function was used to recognize various cable fault signals.The simulative results show that the proposed method is superior to the similar algorithm in recognition accuracy and robustness in different noise environments.

关 键 词：早期故障 S变换 特征选择 最大相关最小冗余 识别 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]