FRFT域能量谱模拟电路故障特征提取方法

故障特征提取是模拟电路故障诊断的关键技术之一,为了提高故障特征的可诊性,提出 1种基于分数阶傅里摘叶变换(Fractional Fourier Transform,FRFT)域能量谱的模拟电路故障特征提取方法。首先,采集测试节点电压信号并将其映射到不同的 FRFT域空间中(p从 0变化到 1)；然后,计算所有 FRFT域空间中的能量谱峰值并将其作为故障特征；最后,将归一化后的特征用于训练最近邻分类器进行诊断验证。与现有的 FRFT故障特征提取方法相比,该方法减少了计算量,且提取的特征能够在所有 FRFT域中更全面地反映不同故障响应信号的细微差异,有利于提高故障特征的可分性。在仿真和物理电路上进行了验证,实验结果表明:所提方法能提高故障诊断准确率,且时间复杂度有明显改善。

Fault Feature Extraction Method of Analog Circuits Based on FRFT Domain Energy Spectrum

Fault feature extraction is one of the key technologies in analog circuit fault diagnosis, in order to improve the di.agnosability of fault features, a method to extract the fault feature based on the energy spectrum of fractional Fourier trans.form (FRFT) domain is proposed. First, the voltage signals of test nodes are sampled and mapped to different FRFT domainspaces (the p changes from 0 to 1). Second, the peak of the energy spectrum in all FRFT domains are computed as a fault fea.ture. Finally, the normalized features are used to train the nearest neighbor classifier for diagnosis and verification. Com.pared with the existing methods, the proposed method reduces the computation cost, and the extracted features can be morecomprehensive to reflect the subtle difference of different fault responses in all FRFT domains, improving the distinguish.ability of different fault features. The simulated and physical experiments validated that the proposed method can improvethe fault diagnosis accuracy and the time complexity is significantly improved.