基于故障区分度DAG-SVM的模拟电路故障诊断  

A Novel Fault Diagnosis Approach in Analog Circuits

作　　者：孙贤明[1] 樊晓光[1] 禚真福[1] 黄雷[1] 陈少华[1] 

机构地区：[1]空军工程大学航空航天工程学院,西安710038

出　　处：《空军工程大学学报（自然科学版）》2016年第4期64-69,共6页Journal of Air Force Engineering University(Natural Science Edition)

基　　金：航空科学基金(201428960220)

摘　　要：为了提高模拟电路故障诊断的精度,针对现有DAG-SVM用于解决多类分类问题固有的不稳定性结构以及"误差累积"的特点,提出了一种基于故障区分度构建DAG-SVM的新方法。根据从不同测试点获取的故障数据信息,定义故障区分度,并以此为依据优化DAG-SVM的拓扑结构,从而消除DAG-SVM结构固有的不稳定性,获得稳定而较高的诊断精度。实验结果表明,与现有的"1vs 1"SVM、DAG-SVM及其改进方法相比,该方法在诊断精度上有明显提高,对于模拟电路的故障诊断具有很好的借鉴意义。In order to improve the accuracy of fault diagnosis in analog circuits,aimed at the instability structure and error transferring of the existing directed acyclic graph support vector machine （DAG-SVM）,a novel approach based on fault distinguish degree to construct DAG-SVM is proposed.According to the fault information acquired from all of the testable points,this paper defines the concept of fault dis-tinguish degree,and takes this as a basis to optimize the topology of the DAG-SVM to eliminate the inher-ent instability of DAG-SVM structure.For this reason,there is a stable and quite good accuracy of diagno-sis.The experimental results show that this method improves obviously diagnosis accuracy compared with＆quot;1 vs 1＆quot;,SVM,and traditional DAG-SVM,and simultaneously the method can be used for reference in an-alog circuit fault diagnosis.

关 键 词：模拟电路 故障诊断 多类分类 有向无环图支持向量机 故障区分度 

分 类 号：TP277.3[自动化与计算机技术—检测技术与自动化装置]