Simulink Model Construction And Fault Diagnosis Of Subway Train Auxiliary Inverte

With the continuous construction of our urban rail transit industry,metro trains with the advantages of punctuality,green and environmental protection have become the primary choice for most urban residents to travel,and the safety of rail transit trains has been widely concerned.As one of the indispensable high-power electrical equipment for rail transit vehicles,train auxiliary inverter has a high incidence of failure due to its complex structure and working in extreme environments such as high temperature and strong vibration.The occurrence of fault of auxiliary inverter is usually random.It is difficult to avoid fault during operation only through manual planned overhaul.The fault prone characteristic of auxiliary inverter requires good condition monitoring of the system during operation,so as to discover the weak fault in the process of vehicle operation in time,and ensure the safety and reliability of the whole auxiliary power supply system.In this thesis,train auxiliary inverter as the main research object,the research focus is divided into four parts:1.By analyzing the auxiliary inverter system of metro train,the circuit model of the main circuit of the auxiliary inverter is built by Simulink simulation tool according to the actual parameters.According to the actual situation,the structural faults at the level of circuit components are analyzed respectively,and then the soft and hard fault simulation is carried out in the circuit model built to find out the causes of transient faults at the output end which may lead to false positives of the system,and the mathematical model is built.2.In view of the characteristics that soft faults have little influence on the output voltage waveform of the system,which are weak faults and are easily overwhelmed by strong noise signals,on the basis of common fault diagnosis steps,the wavelet packet transformation(WPT)algorithm of the improved threshold algorithm is used to denoise the noisy output voltage signal.Denoising preprocessing can reduce the occurrence of modal confusion in the later modal decomposition process,so that the modal components can better reflect the real signal characteristics and retain the fault characteristics to the greatest extent.3.Since mode decomposition algorithms such as fully adaptive noise set Empirical Mode decomposition(CEEMDAN)have disadvantages of mode mixing and long computing time in fault feature extraction of auxiliary inverters,an improved fully adaptive noise set empirical mode decomposition(ICEEMDAN)algorithm is used in this thesis to decompose fault signals of auxiliary inverters.On the basis of this algorithm,the white noise after empirical mode decomposition(EMD)algorithm is added to reduce the mode aliasing phenomenon between the modal components and reduce the operation time of the system.In order to further improve the diagnosis effect of transient faults of auxiliary inverters,the IMF component generated by the decomposition of WPT-ICEEMDAN algorithm is directly used to obtain the eigenvalues by Hilbert(HT)transform,which improves the accuracy of the feature vector.4.Finally,the processed multi-class signal feature vectors are input into the unoptimized least square support vector machine(LSSVM),probabilistic neural network(PNN),and LSSVM and PNN models optimized by the improved Bat algorithm(IBA)for diagnosis.It can be seen from the results of simulation experiments that compared with the unimproved model and particle swarm optimization PSO-LSSVM and PSO-PNN model,the IBA-LSSVM and IBA-PNN algorithm models used in this thesis have great advantages in diagnosis speed and accuracy,and can distinguish several common faults of auxiliary inverters.