Research On Modeling And Online Fault Diagnosis Methods Of Electrical Faults For Traction Induction Motors In Railway Transit

Electric traction systems are the key equipments in railway transit vehicles,and traction motors are the key components of electric traction systems,where their reliability directly affects the operation of the vehicles.Up to now,traction motors mainly adopt the method of regular maintenance,which suffers from “excessive maintenance” and “shortage of maintenance”,inevitably resulting in human and material wastes.This thesis is supported by the key project of High-speed Rail Joint Fund of Chinese National Nature Science Foundation “Safety Prediction and Control of High-Speed Railway Electric Traction Systems”,which focuses on the research of stator inter-turn short-circuit fault and rotor broken bar fault.Fault diagnosis techniques based on signal processing and model are discussed.A large number of simulation and experimental results are demonstrated to verify the proposal.The thesis is arranged as follows:Firstly,to overcome the limitation of traditional induction motor model in fault simulation,motor voltage and flux equations are derived based on the multiple coupled circuits,where the analytical expressions of mutual inductances between stator and rotor loops at any rotor position are derived based on the winding function approach.Furthermore,a magnetizing coefficient is introduced into the model to adjust the inductances,and thus a multiple coupled circuit model considering saturation is built for a healthy induction motor.Moreover,parameter variation characteristics caused by stator inter-turn short-circuit fault and rotor broken bar fault are analyzed,and then voltage and flux equations of these failures are derived,which develops the models of stator inter-turn short-circuit fault and rotor broken bar fault.The proposed models of the induction motor realize the fault simulation,which can provide available fault models for the vadiation of fault diagnosis methods in simulation.Then,characteristic frequencies produced by stator inter-turn short-circuit fault in instantaneous active power and torque are analyzed,where the power and torque signals are estimated through phase voltage reconstruction and close loop flux observer respectively.Accordingly,a stator fault diagnosis algorithm is proposed based on the characteristics in the power and torque spectra,which can detect stator inter-turn short-circuit fault of the induction motor in operation.Moreover,fault sidebands of fundamental,space magnetic motive force(MMF)harmonics and time current harmonics induced by rotor broken bar fault in motor current are analyzed,and then a rotor broken bar fault diagnosis algorithm is proposed based on these fundamental and harmonic characteristics.However,the sideband-based rotor fault diagnosis method depends on slip frequency,which will be limited when the slip frequency is close to zero.Consequently,a rotor broken bar fault diagnosis algorithm based on the low frequency signal injection method is proposed,which has the advantage of being independent on slip frequency,achieving rotor broken bar fault detection in idling operation.Finally,parameters variation mechanism of traction induction motors in railway transit is analyzed,and then an online parameter compensation strategy is proposed,where stator resistance is estimated through the DC bias injection method and magnetizing inductance is corrected through the magnetizing curve,which lays the foundation for followed model-based fault diagnosis method.Furthermore,a uniform expression of fault parameters for stator inter-turn short-circuit fault is derived,and a simplified mathematical model of stator inter-turn short-circuit is futher built,and then the current difference between the healthy and stator fault motors is analyzed,where it contains fault rate and location related information.Accordingly,an observer-based stator inter-turn short-circuit fault diagnosis algorithm is proposed by evaluating the current residual.Moreover,from the point of regarding rotor broken bar fault as an increase of rotor resistance,a uniform expression of the rotor resistance matrix caused by rotor faut is derived,and then a simplified mathematical model of rotor broken bar fault is built.On this basis,a rotor broken bar fault diagnosis alagorithm based on the parameter identification by using least square method is proposed,where the fault severity and location are unknow parameters to be identified.In conclusion,the proposed methods realize the estimation of stator and rotor fault severity and location.98 figures,4 tables,and 171 references.