基于信息融合的模拟电路故障诊断方法分析

Analysis of Analog Circuit Fault Diagnosis Technique Based on Information Fusion

采用2种基于信息融合故障诊断方法,说明用于模拟电路故障诊断的特点。首先利用指定频率下可测点电压、不同测试频率下输出端电压和测试元件的温度3组测试数据,分别用一个改进的BP网络对电路状态进行预处理,得到每个传感器对各待诊断元件的隶属度函数分配,再分别用模糊融合和D-S融和算法进行决策层信息融合并进行故障定位。仿真结果表明:信息融合方法能够克服基于单一信息诊断的不足,提高电路故障诊断的正确率,对单、多软、硬故障均可识别,D-S融合算法在解决电路故障诊断中的不确定性问题方面优于模糊融合。

Two fault diagnosis methods based on information fusion were adopted in order to illuminate their characteristics in analog circuit fauh diagnosis. According to three measurement results of accessible node voltages at an appointed frequency, output voltages at different frequencies and temperatures of being monitored components, the circuit state was processed preliminarily using three improved BP neural network respectively to obtain the membership function assignment of the three sensors to the circuit components under test, then decision-making information fusion results were obtained using fuzzy method and D-S method respectively and the faults were oriented. Experiment results show that information fusion method can overcome the diagnosis shortcomings based on simple information, improve fault diagnosis accuracy, identify simple and multiple, soft and hard faults, D-S method is more accurate than fuzzy method in solving the uncertainty of circuit fault diagnosis.