Research On Electromagnetic Compatibility Based On Wavelet Analysis And FastICA Algorithm

With widely use of high-speed power switching devices in electrical and electronics products, a large number of conducted electromagnetic interference (EMI) problems generated by power electronic equipments n reality and have been a main restriction of progress of modern power electronic technology, people now attach importance to EMC problems over again. So, researches of diagnosis and mechanism of conducted EMI noise generated by power electronic equipments become more significant.According to the dynamic and the research status of electromagnetic compatibility at home and abroad, FastICA and wavelet analysis method are proposed in the application of the electromagnetic compatibility. First, this paper expounds the independent component analysis and wavelet analysis theory in detail, and use the Matlab simulation to verify the effectiveness of the proposed method. PI (Perfomance Index) performance indicators is put forward to measure FastICA, and Pearson correlation coefficient of correlation is put forward to show the effectiveness of the separation or extraction results.Taking example by Switch Mode Power Supply, this paper analyzes mechanism and transmitted route of conducted EMI noise, obtaining total noise in the power line by LISN, and get common mode component and differential mode component by the hard separation network, analyzing Pearson correlation coefficient by contrasting the hard separation network and the combination of FastICA and wavelet analysis method. The results show that the method of conducted EMI separation is effective, thus the modal separation process are simplified, this method greatly reduces the dependence on the hardware circuit, which reduces the cost and complexity. Secondly, noise source extraction model is established to obtain total noise by LISN. This paper not only demonstrates the effectiveness of the noise source extraction basing on FastICA through experiments, but also analyzes the advantage through wavelet analysis method and then FastICA by Pearson correlation coefficient. The results show the effectiveness of this method to conducted EMI noise source extraction, this method solved accurately and quickly the problem of identification of conducted noise source as the core of electronic device, which judging accurately electronic devices generating conducted noise signal and reducing electromagnetic interference contrapuntally. By this method, not only meeting the technical specifications and minimizing the dependence on hardware, which lead to save the economic cost, and realize the integration of technology and economy.