Fault Diagnosis Of IGBT And Sensors In Traction Inverter System

With the rapid development of China’s railways, the reliability of AC electric traction drive system has a vital role on trains’safe operation, which is due to its conducting as a power source on the train. However, in the main circuit of traction drive system the insulated gate bipolar transistor (IGBT) modules and sensors often fail, which could lead to trains’ abnormal operation, even stopping and some other unexpected accidents. In order to locate the fault quickly, reduce the operation cost of maintenance and replacement, it is very important to have much research on traction drive system’s online fault diagnosis.Firstly, the IGBT modules, current sensors and speed sensor in traction inverter system were studied in this paper, and the causes of their failure were briefly described. Then the single IGBT fault and double IGBT fault were respectively analyzed in detail, as well as the fault of current sensors and speed sensor, and the impact on system caused by faults was also explored, which could provide the basis for real-time fault diagnosis.Secondly, through comparing three-phase voltages and currents in the inverter under normal and faulty conditions, the fault characteristics corresponding to different types of IGBT fault were obtained. Then the fault diagnosis method based on residual voltages as well as the fault diagnosis method based on residual currents was introduced in this paper. So as to locate the fault easily, the diagnosis rule tables of two methods were designed respectively. In the simulation environment the above diagnosis methods had been verified, and the diagnosis periods were less than one fundamental, then their strengths and weaknesses were compared based on simulation results.Thirdly, concerning current sensors fault and speed sensor fault, this paper presented a fault diagnosis method based on the full-order observer used to estimate motor speed and three-phase currents online. Then those estimated values were compared with the corresponding variables measured by sensors, and the residual vectors were obtained. Combining with the diagnosis rule table that had been designed, it was convenient to distinguish the fault. For the purpose of avoiding the possible misdiagnosis caused by the IGBT fault, the algorithm also took advantage of the feature that the sum of three-phase currents was zero. The simulation results showed that the diagnostic algorithm presented was valid.Finally, based on the experiment platform consisted of TMS320F28335 controller and RT-LAB real-time simulator, the control method in C style was programmed, and on this platform, the above IGBT fault diagnosis algorithms and sensors fault diagnosis algorithm were carried out successfully. The experimental results were consistent with simulation results, which further demonstrated the feasibility of the diagnostic algorithms.