Compound Fault Diagnosis Of Traction Motor Drive Systems In The High-speed Train

The high-speed train operation system is a large complex engineering system.Due to factors of multiple closed-loop,high coupling,strong interference and multi-source uncertainties,the traction system is prone to compound failure and stability decreasing during the operation of the train.In this paper,traction motor drive system of CRH2 EMU is taken as the research object.According to the characteristics of different types of compound fault in induction motor drive system,the compound fault diagnosis scheme is studied.Considering the requirement of CSR Zhuzhou Electric Locomotive Research Institute,this paper studies the following three types of composite fault diagnosis of traction motor drive system:(1)asynchronous motor compound fault of stator turn-to-turn short circuit and rotor breakage;(2)the composite fault of gain and offset about the current sensor;(3)composite fault of degradation of resistive properties and IGBT open fault.The diagnosis details are as follows:There is a method to diagnosis the complex fault of inter-turn short circuit and rotor breakage.Considering the influence of noise to fault diagnosis,this paper adopts wavelets analysis to reduce the effect of noise.The extended Park vector is used to fusion of stator current to get Park vector module,which can solve the problem that the characteristic frequency of complex fault is easily inundated by the fundamental wave component.Then,the spectrum of this module is analyzed by Fast Fourier Transform to extract the features of the compound fault,which are used to train BP neural network.Because the training speed is slow,this paper adopts the additional momentum method and the variable step method to improve the training speed of BP neural network.The composite fault diagnosis scheme is verified in Zhuzhou experimental platform.A compound fault diagnosis scheme based on average current signal is proposed to the compound fault of current sensor gain and bias of asynchronous motor drive system.Because the signal noise will lead to misjudgment by affecting the composite fault signal,the wavelet analysis is used to reduce the noise of the current signal.The weighted average method is used to convert three-phase current to the average current signal,which can remove symmetric components.Analysis and extract the feature in time domain,and then Fisher classifier is used to diagnose the type of fault and curve fitting method is adopted to estimate the severity according to the extracted features.The composite fault diagnosis scheme is verified in Zhuzhou experimental platform.A composite fault diagnosis method based on Hilbert modulus is proposed for diagnosising the complex fault of resistance performance degradation and insulated gate bipolar transistor(IGBT)open fault.This paper use correlation coefficient to select useful information of empirical mode decomposition and reduce noise.The Hilbert transformation is used to transform the current signal into Hilbert modulus,which can identify fault frequency clearly.Then,analysis and extraction of Hilbert features.According to the nonlinearity of fault feature samples and difficulty of obtaining the global optimal parameters,the genetic algorithm is used to optimize support vector machine,which can diagnosis the fault type.The composite fault method is verified on the Central South University traction drive system Fault injection platform.