电子电路故障诊断的神经网络数据融合算法

针对模拟电路故障元件诊断的不确定性问题，将BP网络引入数据融合之中，结合模糊集合论，构造－模糊神经网络故障分类器，并将其应用于电子电路故障诊断之中。通过测试电子电路中被诊断元件的工作温度和工作电压这2个物理量，求出两传感器对各待诊断元件的故障隶属度，利用模糊BP网络故障分类器进行数据融合，得到融合的各待诊断元件的故障隶属度，从而确定故障元件，并通过单传感器诊断结果与融合诊断结果比较，说明多传感器融合的优越性。

Neural Network Data Fusion Algorithm of Circuit Fault Diagnosis

In order to solve uncertain problem of circuit fault diagnosis, a fuzzy neural network fault classifier is designed based on BP neural network and fuzzy logical theory, and it is used in fault diagnosis. By measuring the temperature and voltage of circuit components, the membership function of two sensors to circuit component is obtained and the data fusion is made by using fuzzy BP neural network classifier. Thus the fusion fault membership function of all circuit components and the fault component can be found. By comparing the diagnosis results based on separate original data and fused date, it is shown that the latter is more accurate than the former in circuit fault recognition.