考虑异质信息不相容的模拟电路融合诊断

研究提出了一种基于异质信息融合的模拟电路故障诊断方法,以解决模拟电路诊断中故障特征信息缺乏和决策融合中异质信息不相容而导致误判等问题.该法首先采用基于可转移置信模型的核近邻法提取电路温度特征信息,并进行故障预识别,再结合基于可测点电压的神经网络故障预识别结果,在考虑信息不相容的情况下计算关联权重系数、先验权重系数,从而实现电路的综合诊断.电路仿真试验结果表明,所提诊断方法对容差模拟电路的多故障和单故障均适用,故障定位准确率高.

Analog circuit fault diagnosis based on heterogeneous inconsistent information fusion

A new analog circuit diagnosis method based on heterogeneous information fusion is proposed in order to solve the problems of insufficient diagnosis information and unreliable fusion results from inconsistent information. The characteristic temperature information is extracted by transferable belief model case-based classifier and then primary fault identification is performed. Meanwhile, another separate primary fault diagnosis is performed with a neural network based on accessible node voltages. By calculating prior weight coefficient and correlation weight coefficient under the consideration of information inconsistency, the synthetic diagnosis is completed and more reliable fault diagnosis results are obtained. Simulation experiment results show that the proposed method has the capability to diagnose multiple faults and single faults in tolerance analog circuits and gains satisfactory accuracy.