Fault Diagnosis Of Traction Inverter System Based On Signal Processing

High-speed railway is a major national development strategy,which plays a significant role in promoting national economy,and the international status.As the power source of high-speed trains,the traction drive system needs high-reliabity to maintain the safety of the trains.However,due to the harsh working environment and complex working conditions,the traction drive system has a high fault rate of some key components,which cause to sudden accidents.Therefore,the research on fault diagnosis technology of the high-speed train traction drive system has a great significance to improve the trains reliability and reduce the maintenance cost.In this dissertation,the traction inverter is used as the research object.Aiming at the faults of the insulated gate bipolar transistor(IGBT),current sensor and asynchronous motor in the traction inverter system,the corresponding fault diagnosis algorithms are proposed.Firstly,the working principle of the traction inverter system is introduced briefly.And then,the fault feature analysis of IGBT open-circuit fault(single switch,double switch),current sensor fault(open circuit,bias,gain)and asynchronous motor fault(short circuit between stator turns)are completed.And the influence of different faults on the electrical quantity of the whole system are analyzed.Secondly,aiming at the problem that enormous fault types of IGBT and current sensor can affect the diagnosis accuracy,a fault diagnosis method combining signal processing and classifier is proposed.Firstly,different optimization algorithms are used to select variational mode decomposition(VMD)parameters adaptively to improve the decomposition effects.The optimized VMD decomposition is used to separate the frequency band of each stator current.Then,the false components obtained by decomposition are filtered out.The fault feature is extracted according to the difference between the probability entropy and time domain feature in different faults type.The optimized extreme learning machine is used for fault identification.Finally,the simulation results are used to verify the feasibility of the studied fault diagnosis algorithm.Afterwards,the weakness of fault characteristic frequency is a troublesome problem in the the short-circuit fault between stator turns,which may lead to degrade the diagnostic accuracy.To address this,a fault diagnosis method combining multiple signal decomposition and reconstruction based on the DC side current signal is proposed.Firstly,the fault characteristic frequency in the DC side current when the stator turns short circuit fault is derived,and the effective extraction of the fault characteristic frequency is realized through simulation experiments.Finally,the hardware-in-the-loop(HIL)test platform is used to collect data.And then the diagnosis effect of the algorithm is verified.The results show that the proposed method can diagnose the IGBT,current sensor and stator inter turn short circuit faults accurately.