Fault Diagnosis For Sensors Of PMSM Drive System In Metro Train

With the acceleration of urbanization process in China, Rail transit is playing an important role in economic development and improving people’s livelihood. How to improve maintenance efficiency of Metro train has become an increasingly pressing issue. Traction system is one of the nine key systems of Metro train, And the online monitoring for its working status attract more and more attention. Though subway train commonly used asynchronous traction system, Permanent magnet synchronous motor(PMSM) traction system as a new traction system, with small size, light weight, high efficiency, has greater potential for development, and major equipment manufacturers are competing to develop a PMSM traction systems. This paper is working on sensor fault diagnosis of PMSM drive system in metro train, and the detail contents are illustrated as following:Firstly, the common types and working principles of position sensor, current sensor and voltage sensor in PMSM drive system were summed. And three fault sensors were modeled. Effect of sensor failure on PMSM vector control was also theoretical analyzed, and the theory was verified by simulation using Matlab/Simulink.Secondly, based on the mathematical model for PMSM, full order sliding mode observer was built. On the base of that, it presented a fault diagnosis algorithm for sensors using the estimated value of extended back-EMF and the rotor position angle. Unfortunately, the algorithm is difficult to identify which current sensors is fault. Then an open-loop observer was used to achieve the current sensor diagnosis and localization with a precondition that position sensor and voltage sensor work properly. With the introduction of diagnostic logic and diagnostic variables, the paper synthesized the advantage and disadvantage of two algorithms, and founded a new fault diagnosis algorithm. The new diagnosis algorithm can realize the diagnosis and localization of sensor faults.Thirdly, since method based on the model is greatly influenced by the parameters’ change and traditional PMSM model (d/q model) cannot reflect the variation of d/q axis inductance, a finite element model of subway PMSM was established using Jmag software. The main characteristics of PMSM under rated conditions were analyzed, and the variation law of inductance along with currents was also given. Furthermore, finite element model of PMSM was compared with d/q model based on the vector control algorithm under Matlab/simulink.Finally, Hardware-In-the-Loop (HIL) experiment platform with DSP TMS320F28335 as controller and RT-LAB as virtual load was set up. The main circuit model of electrical traction and drive system is load to RT-LAB simulator, and the vector control algorithm for PMSM was programmed and load to DSP TMS320F28335. The algorithm for sensor fault diagnosis was validated by HIL experiment using finite element model instead of d/q model.