论文部分内容阅读
断路器操作时会伴随剧烈的振动和声音信号,该文提出了一种基于自适应权重的证据理论,诊断断路器机械故障方法.首先利用小波包对多传感器的振动和声音信号进行分解,然后提取特征熵输入(library for support vector machines,LIBSVM)获得了基本可信度分配,再利用状态分类准确率对权重自适应赋值,最后通过证据理论(dempster shafer,DS)将多信号加权后的基本可信度进行融合,实现了断路器正常、卡涩、基座松动和拒分状态的识别.通过识别实验表明:自适应权重的证据理论在避免证据冲突情况下,能够有效提高断路器故障类型诊断的准确率.“,”The circuit breaker operates with a sharp vibration and sound signal.In this paper,a circuit breaker mechanical fault method based on adaptive weight was proposed.Firstly,wavelet packet was used to decompose the vibration and sound signal of multi-sensor.Then the characteristic entropy could be extracted and put into LIBSVM(A Library for Support Vector Machines) to obtain the basic probability assignment.The status classification accuracy was applied to the weight adaptive assignment.Finally,the basic reliability of multi-signal weighting was integrated by using D-S (Dempster Shafer) evidence theory.Different status including normal,sticking,pedestal loosing and break rejecting could be recognized.According to the identification experiments,the evidence theory of adaptive weight can effectively improve the accuracy of fault type diagnosis of the circuit breaker in the case of evidence conflict avoidance.