Fault feature extraction of power electronic circuits based on sparse decomposition
Jing Hou, Yuan Wang, Tian Gao, Yan Yang · 2016
A fault feature extraction approach is presented based on sparse decomposition theory for power electronic circuits. An overcomplete dictionary including discrete cosine transformation (DCT), Heaviside, and wavelet packet dictionaries is constructed according to the characteristics of different faults in power electronic circuits, then N maximum sparse decomposition coefficients and their indexes which can well distinguish the different fault types are extracted as the fault characteristics, finally a support vector machine (SVM) classifier is employed to perform fault diagnosis and classification. A Buck power circuit simulation is used to verify the effectiveness of the proposed approach. Simulation results show that using the feature vector extracted by the sparse decomposition approach and SVM classifier can achieve a high probability of correct classification of faults.