Fault Diagnosis Method For The Main Circuit Of Auxiliary Inverter In Urban Rail Train Based On Transforms And Computational Intelligence

In order to improve the level of diagnosis technology with the main circuit fault characteristics, this paper took the main circuit of auxiliary inverter in urban rail trains as diagnosis object, developed a fault diagnosis method based on transforms and computational intelligence.Firstly, this paper introduced the research background and significance, then summarized the fault diagnosis technology for the main circuit of inverter.Secondly, we discussed the structure, working principle and control technology. According to the analysis working mechanism of the auxiliary inverter, we establish and verified the simulation model of the auxiliary inverter by SIMULINK, analyzed the auxiliary inverter failure mode.Through the simulation model, we simulated the IGBT open circuit fault and got all the fault output voltage waveform as the research foundation of fault diagnosis.Thirdly, based on the simulated fault waveform, we studied the feature extracting methods by transforming the fault three-phase voltage with3/2transformation, then got the voltage trajectory for every fault, According to the characteristics of voltage trajectory, the gravity method which is used to extract the fault feature method has been clear. We extracted fault characteristics with FFT transform, analyzed the limitations of this method. Through the comparative analysis of the two methods, this paper presented the combination of two methods to extract the fault feature at the same time.Finally, based on fault feature extraction methods, fault identification methods were studied. By means of PSO-BP neural network to identify fault, cleared the fault codes and the method of building PSO-BP neural network, then the simulation data and field data demonstrated the effectiveness of its diagnostic method.The research results showed that the establishment of simulation model, fault simulation, fault feature extraction method and fault identification method are effective and valuable.