Identification for Abnormal Electrical Phenomenon in Vehicle-grid Coupling System Based on Wavelet Packet Decomposition and Neural Network

Abnormal electrical phenomena(AEP) have a great impact on the normal operation of the vehicles in electrified railway. Therefore, it is particularly important to identify the AEP in vehicle-grid coupling system. At present, the AEP, including low frequency oscillation, short circuit, harmonic resonance and inrush current, have received much concern in electrified railway. In the existing identification methods, each method only identifies one type of AEP, which cannot cope with the frequent occurrence of various AEP types in practice. Therefore, this paper proposes an identification method for AEP in vehicle-grid coupling system based on wavelet packet decomposition and back propagation neural network (BPNN), which can identify the above four typical AEP while distinguishing the normal data. Firstly, wavelet packet global decomposition is used to accurately extract the feature information of non-stationary signal. Then, a three-layer BPNN is established to identify the signals. Finally, experimental results show that the proposed method has high identification accuracy for the typical AEP.