基于信息融合的模拟电路故障的特征提取与融合方法

Feature Extraction and Fusion Method for Analog Circuit Fault Based on Information Fusion

摘要:

在模拟电路故障诊断中,故障特征的提取是一个非常重要的环节,其提取结果的好坏将直接影响最终的诊断正确率;对现有文献研究发现,每种特征提取方法单独使用时都有一定的局限性,为了能够更加充分地提取模拟电路故障特征,提出了小波包分析与主元分析并行应用的方法,并将两种方法提取的特征向量依据不同规则进行了三种类型的融合,方便对比实验;为获取最优小波特征,提出了特征偏离度,并以此为标准选择最优小波基;最后,通过设计一种改进的神经网络分类器模型,将融合后的三种特征向量送入其中进行仿真验证,得出最终诊断结果;结果表明,该方法能够有效克服单一特征提取方法提取不充分的缺点,提高故障诊断的正确率,并且融合因子μ适中时诊断正确率最高。

Abstract:

In the fault diagnosis of analog circuits, the extraction of fault features is a very important link, and the result of extraction has a direct impact on the final correctness of fault diagnosis.Because of the limitation of single fault feature extraction and to extract the fault characteristics more fully ,a method of fault feature extraction based on wavelet packet analysis and principal component analysis (PCA) is proposed, and three different feature vector fusion models are proposed. In order to obtain the optimal wavelet feature, the characteristic deviation degree is proposed, and the optimal wavelet basis is selected as the standard. Finally, an improved neural network classifier model is constructed, and the results of the fusion are sent into it to verify the results. The specific algorithm and simulation examples are given in this paper, The results show that the proposed method can effectively improve the correctness of fault diagnosis compared with single fault feature extraction method, and when the fusion factor is moderate, the correctness of diagnosis is the highest.