基于异类信息融合的模拟电路故障诊断方法

Fault diagnosis method of analog circuits using dissimilar information fusion

针对模拟电路故障诊断中存在的可测节点电压信息不足、单一信息不能表征所有故障状态等问题,提出了一种基于神经网络和D-S证据理论的模拟电路异类信息融合故障诊断方法。该方法首先提取两类故障信息并分别将其输入不同的神经网络对其进行初步诊断,并得到各自的诊断结果,然后根据初步诊断结果,运用D-S证据理论对其进行融合,做出最终决策。

In order to solve the problems of insufficient accessible nodes voltage and single information can not represent all the fault state in the analog circuit fault diagnosis, a dissimilar information fusion fault diagnosis method based on neural network, D-S evidence theory is proposed. The different neural network to carry on the preliminary diagnosis to the multi-sort fault information distilled from test circuit. According to the preliminary diagnosis results, ultimate decision-making by means of the fusion using D-S evidence theory.