基于FastICA算法的电路故障诊断方法  

A method of circuit fault diagnosis based on FastICA algorithm

作　　者：雍静[1] 雷冬梅[1] 曾礼强[1] 徐欣[1] 

机构地区：[1]重庆大学输配电装备及系统安全与新技术国家重点实验室,重庆400030

出　　处：《电路与系统学报》2013年第2期291-297,共7页Journal of Circuits and Systems

基　　金：中央高校基本科研业务费资助(CDJZR10150011)

摘　　要：在电路故障检测中针对元件种类较少且同一种元件重复出现的电路,提出基于FastICA(fast independent component analysis)算法的故障诊断方法。该方法首先对电路中的元件进行分类再按信号叠加算法分组建立测试电路,电路正常工作时测试电路不给电。利用FastICA算法对测试信号进行盲分离,初步判断分离结果后与预先建立的元件故障信号库Mf的信号作相关分析判定故障类型。这种故障诊断方法简单易实现且对元件类型没有限制,在检测过程中元件之间没有相互干扰可确保诊断的准确性。仿真表明该方法能有效判断出电路的故障情况。For the electric circuit with few types of components and the same kind of component repeated many times,a method of circuit fault diagnosis based on FastICA algorithm is presented.By designing a test circuit to the original circuit,and dividing the original circuit into several sub-circuits,this method samples the mixed current signals of the sub-circuit and separates the mixed observation signals by using FastICA algorithm.Then determine the fault type by analyzing the correlation between the separation signals and the Mf.There is no interference among circuit components in the process of the electric circuit fault detection and no limit to the component types,so this kind of diagnosis is fast,uncomplicated and effective.The simulation results can effectively show the location and types of the circuit fault.

关 键 词：故障诊断 FASTICA算法 相关性分析 测试电路设计 

分 类 号：TN79[电子电信—电路与系统] TP391[自动化与计算机技术—计算机应用技术]