Research On Fault Diagnosis Of Auxiliary Power Supply System Of CRH3 EMU

With the development of economy,the high-speed rail in China has reached a new level,while the number of operational EMUs exceeds 3,800 standard groups,with a mileage of38,000 kilometers so far.High-speed rail has become the most important one among many transportation modes,among which CRH3 EMU has the longest running mileage with the largest number of EMUs.Since 2007,the first batch of EMUs has been in operation for more than ten years.As the service life of EMUs continues to increase,the failure rate also increases year by year,while the corresponding maintenance pressure also increases.However,the power supply quality and reliability of the auxiliary power supply system is the most direct factor that affects the safety of EMUs operation and passenger comfort.During the operation of EMU,it is of top priority to ensure the safety of passengers is "safe and secure" and provide them with a safe and comfortable riding experience.Therefore,this thesis takes the auxiliary inverter,which is the key component of the auxiliary power supply system of CRH3 EMU as the research object,various IGBT open-circuit situations that may occur are researched,while different fault diagnosis algorithms are adopted and compared to achieve effective and accurate fault diagnosis of the auxiliary inverter.(1)The current status of domestic and international research on the fault diagnosis of auxiliary power supply system and key components of EMU are analyzed through elaborating the background of the thesis and the significance of the research.The operation principle of auxiliary power supply system is studied,and the topology of auxiliary power supply system is also drawn.Based on statistics,the key component of auxiliary power supply system is obtained as the auxiliary inverter,and the MATLAB/Simulink simulation model of auxiliary inverter is also established by deep profiling of the major circuit diagram,modulation strategy and filtering parameters of auxiliary inverter.(2)Through analyzing the basic structure and failure mechanism of IGBT,different open-circuit faults of auxiliary inverter IGBT is classified to obtain three-phase current output waveforms in various states.Based on the features of its output current waveforms,further research and analysis are conducted to obtain the preliminary connection between current waveform features and different faults.(3)Through studying the basic principles of wavelet transform,the ideal db3 wavelet decomposition is adopted to reconstruct the current output waveform and obtain data consisting of high & low frequency energy values,which are processed and reconstructed into energy spectrum,so that the fault feature vector of auxiliary inverter is extracted as the input data for the following fault diagnosis algorithm.(4)The feature vectors extracted by wavelet decomposition are summarized as input,while the faults are coded as output;BP and GRNN neural networks are used for fault identification,L-M algorithm is used by BP for iteration,and the cross-validation method is used by GRNN for repeated training to find the best SPREAD value.Both methods can achieve fault diagnosis and conduct comparative analysis.Finally,the operation with db3 wavelet basis is determined based on detailed analysis,and the resulting fault feature vector can better present the fault information in it,as well as verifying the correctness of fault feature vector.The two neural network algorithms used can effectively complete the fault diagnosis of auxiliary inverter,among which the training time of BP neural network doesn’t have much difference from that of GRNN,but the GRNN neural network has significantly improved the recognition accuracy with more accurate diagnosis results than BP.