摘要： 研究了基于多类电量测试信息及其多诊断方法融合的模拟电路故障诊断。获取可及节点电压,运用K故障诊断法进行故障在线检测与初步定位,再离线测量电路在不同的测试频率下输出对输入的增益,运用最小标准差法进行诊断。由所提故障隶属函数得到基于各类测试信息的元件故障隶属度,以此计算D-S证据理论中各传感器的信度函数分配,再运用D-S合成法则实现融合诊断。模拟实验结果表明:所提融合诊断方法大大提高了故障定位的准确率。

关键词: 故障定位, 隶属函数, 证据推论, 决策融合, 模拟电路

Abstract: Based on D-S evidential reasoning and fuzzy theory, this paper studies analog fault diagnosis by means of multiform circuit responses and varied diagnosis methods. Accessible node voltages are employed to on-line fault detection and location by K-fault nodes diagnosis. Then the circuit gains of output to input under different test frequencies are used to off-line diagnosis by least dispersion method. The proposed fault attaching functions connect these diagnosis results to belief function assignment of a corresponding sensor. Fusion diagnosis is performed based on D-S combination theorem. Experimental results show that the proposed fusion diagnosis method can avoid limitation of single information and achieve satisfactory accuracy.

中图分类号: 

• TN710