Research On Fault Diagnosis For Traction Motor In Closed-loop System Of High-speed Railway

As one of the most important transport modes for passenger and freight,the safety issue of high-speed railway draws attention,accompanied by its rapid development,more rigorous safety and reliability requirements should be satisfied.It is of great importance to ensure the safety by fault diagnosis or security measures.Induction motor provides power as part of traction system of vehicle.In the event of failure,the train will be late or even stop and cause an adverse effect,even an affair.Therefore,this study aims to traction motor in the closed-loop system,confronting high-order harmonic interference,proposing fault diagnosis approaches to detect sensor faults.Firstly,combined with background of research,this study describes the status situations,tendencies and existing problems of fault diagnosis for traction motor of closed-loop system.On this basis,considering the sensor faults and disturbances,the necessity and significance for motor fault diagnosis are given,as well as the main work and arrangement.The fundamental mathematic knowledge,including theories about linear matrix inequality(LMI)and robust analysis are introduced in this study for observer convergence and robustness analysis in later chapters.For traction motor closed-loop system,from input to output signals of motor loop,the feedback control is equivalent to the stator voltage control by proportional-integral(PI)control laws to achieve constant speed objective.In this equivalent control system,the dynamic differential equations of stator and rotor currents are established under static two-phase coordinate.In order to facilitate the design and digital controller,the digital model is given by discretization.The disturbances of 5th harmonic interference,temperature and zero drift sensor faults in the condition of high temperature variable environment are proposed based on motor operations.To solve sensor fault diagnosis issues with harmonic interference,the augment system extended by sensor faults is introduced.The estimated stator,rotor currents and sensor faults can be achieved by the designed observer,which is decoupled with sensor faults.The Lyapunov stable theory demonstrates the convergence of observer.And its gain can be calculated by the linear matrix inequality method.The robustness of observer with disturbances is carried out by error equations.This paper presents two fault diagnosis methods based on observer estimations.One method is to construct residual formulated by faults estimated,whose threshold is calculated by robust analysis of observer.The other method is to design residual by plant outputs and observer outputs,and its threshold is formed by the impulse transfer function from disturbance to residual.Through theoretical analysis and simulation experiments,the applicability and limitations of the two method are compared.The method based on faults estimation calculates less,but with smaller fault diagnosis available region.The other one requires a large amount of calculations by(Linear Matrix Inequation)LMI solver but has larger detection range.The reconstructed output based on estimated fault is to reduce the propagation in closed-loop system and give guidance to the precise feedback signal,which is of great significance for fault isolation.At last,the paper concludes and gives prospects for the study.The innovations and limitations are proposed for further investigations.