D-S证据理论和多模型结合的模拟电路故障诊断

Fault Diagnosis of Analog Circuit Based on a Combination of D-S Evidence Theory and Multi-model

为提高模拟电路故障诊断的精确度和正确率,采用信息融合方法进行故障诊断。首先取不同频率下的输出增益作为特征参数,经ANFIS模型、BP模型、RBF模型3种方法的局部诊断,获得彼此独立的证据;然后采用D-S证据理论及方法对证据进行决策融合故障定位,并将局部诊断正确度加入到基本概率赋值的获取中。实例证明,经过融合处理后,诊断的可信度明显增加,有效地提高了故障诊断的正确率和精确度。

In order to increase the precision and accuracy of analog circuit fault diagnosis ,the paper uses information fusion method for fault diagnosis .First ,take the output gain under different frequency as characteristic parameters ,three meth-ods ,including ANFIS model ,BP model and RBF model ,are considered as primary diagnosis ,which can obtain independ-ent evidence .Then use the method of D-S evidence theory evidence to decide the fusion fault location ,and join the local di-agnostic accuracy of the acquisition of basic probability assignment .Through its application in a circuit ,it indicates that by multi-level information fusion treatment ,not only significantly increase the credibility of diagnostic conclusions ,but also improve the diagnostic accuracy .