基于改进EWT的模拟电路故障诊断研究

Research on Analog Circuit Fault Diagnosis Based on Improved EWT

针对模拟电路故障信号非线性、非平稳性、容差性的特点,提出了一种基于经验小波变换(EWT)的模拟电路故障诊断方法。但信号分解需要设定分割模态个数,为实现经验小波变换中Fourier谱的自适应分割,提出了自适应无参的经验小波变换(APEWT)方法。经验小波变换能有效分离信号的调幅调频成分,将改进方法对Leapfrog benchmark电路不同故障的输出信号进行模态分解以及时频能量谱分析,采取人工智能算法进行实验分析,结果表明其分解的各分辨率模态具有相应的时域特征。并与希尔伯特黄变换的方法进行对比,表明所提出的方法不仅能有效地提取模拟电路软故障特征,诊断正确率高于后者,而且具有完备小波理论支撑,计算速度快,不存在虚假分量的特点。这为模拟电路故障在线诊断提供了新思路。

Aiming at the nonlinearity,non-stationary and poor component tolerances in extracting analog circuit fault signals features,we propose a new method based on EWT.However,it is difficult to set the number of modes in separating Fourier spectrum.To fulfill an adaptive separation of Fourier spectrum,we put forward an adaptive nonparametric EWT (APEWT) which can separate the amplitude modulation-frequency modulation components effectively.It has been applied to analyze the output signals of different faults in the Leapfrog benchmark circuit,to perform modal decomposition and time-frequency energy spectrum analysis.The experimental analysis carried by artificial intelligence algorithm shows that the resolution modes obtained by EWT have the corresponding time domain signal characteristics.By comparing with the method of HHT,the proposed method can not only extract features of soft fault features in analog circuit effectively,diagnose more accuracy than the latter,but it has complete wavelet theory,high calculation speed without no false mode. This will help to provide a new idea in extracting features in analog circuit soft fault diagnosis online.