基于电路特征信息矩阵的容差模拟电路故障诊断

Fault diagnosis for analog circuits with tolerance base on ciucuit feature information matrix

针对模拟电路软故障诊断问题,提出了一种基于电路特征信息矩阵的故障诊断方法。依据电路在不同激励下各测点的响应特征向量组建立电路特征信息矩阵,利用加权马氏距离表征各故障特征向量之间的相似程度。再通过将待测电路马氏距离矩阵与各故障模式的马氏距离容差矩阵进行对比计算,获得电路的特征信息相似度矩阵。对其进行相应的离散度和模糊度加权处理,求取待测电路对电路各故障模式的总体相似度矢量。最后根据电路故障判断准则确定电路故障模式。诊断过程利用相关软件进行数据采样、分析和处理,减小了工作量,提高了诊断效率。实验结果表明,方法适用于容差条件下软、硬故障的诊断,相对于利用单一故障特征的诊断方法,具有更高的检测准确率。

In this paper,we propose a soft fault diagnostic method for analog circuits based on a circuit feature information matrix,which we developed with respect to the response characteristics of different excitation levels at different test points. We used the weighted Mahalanobis distance to characterize the degree of similarity of each faulty feature vector. By comparing the weighted Mahalanobis distance matrixes of fault-free and faulty circuits,we were able to construct the circuit feature information similarity matrix. Using this matrix,we can obtain the integrated similarity vector of each circuit＇s fault mode by considering the dispersion and fuzzy weighting factors. Finally,we determine the circuit＇s fault mode based on the judgment criteria of the circuit fault. To reduce the diagnostic workload and improve diagnostic efficiency,we based the data sampling,analysis,and processing of the diagnostic procedures on related software. The simulation experiment results show that our proposed method is applicable not only to catastrophic faults but also to parametric faults in the tolerance circuits,and that it has higher detection accuracy compared to methods with only a single-fault feature.