Research On The Fault Diagnosis For Inverter Of Electric Vehicle Charging Station Based On Empirical Wavelet Transform And SVM

The electric vehicle industry,as a basic energy supply facility,is developed rapidly nowadays.The electric vehicle charging stations are not only the cornerstone of building a complete charging service network,but also an important prerequisite for promoting the progress of the electric vehicle industry.With the development of the charging station,the current commercial value of traditional AC charging stations is declining and cannot effectively meet the increasingly large charging needs of consumers.With the application of large-capacity lithium-ion batteries and ternary batteries,DC fast charging stations have become the mainstream commercial charging mode due to their efficient charging rate.The inverter is the core part of the DC charging station which is easily cause failures by the frequent operation of internal IGBT.There are no research reports on fault detection and diagnosis of IGBT in charging stations on both china and abroad.However,power electronic devices inside charging stations are closely related to the grid,and the daily use scenarios are complex.So a practical and suitable fault detection and diagnosis method is required.Since the IGBT short-circuit fault in inverter can be simply cut off by the protection circuit,the IGBT open-circuit fault diagnosis is the main concern of the current research.In addition,the charging station is often in a non-Gaussian noise environment in the actual environment,which interferes with the output signal waveforms,causes difficulties in the extraction of fault characteristics.Therefore,this article is based on the open-circuit fault reaserch of IGBT in the DC fast charging station of electric vehicles,and the work on the fault feature extraction and diagnosis under the background of non-Gaussian noise are done.The main work and research results are as follows:(1)The foundation for fault research lies on modeling an electric vehicle charging station model which meets the national standard,and analyzing the IGBT open-circuit fault of the inverter.Therefore,the topological structure and working principle of electric vehicle charging stations are analyzed.The key parts are divided into five modules,three-phase power grid and transformer,LCL filter,AC/DC inverter,DC charger and power battery.The key parameters of this modules are calculated.Second,the voltage and current closed-loop control strategy based on voltage feedforward is proposed.The charging strategy were also designed.The model was simulated and tested under different working conditions,which verified the correctness of the proposed model.Thirdly,This paper analtzed different types and characteristics of IGBT open-circuit faults,and the influence of non-Gaussian noise on that is studied.(2)The empirical wavelet transform theory is studied.Aiming at the lack of adaptability by the original EWT,the adaptive division of the frequency band by the scale space theory is proposed.On this basis,correlation analysis is used to filter the subs signal components decomposed by EWT.Since EWT is more sensitive to non-Gaussian noise,cyclic correntropy and cyclic correntropy spectrum to reduce the impact of non-Gaussian noise is introduced,and fault characteristics of the IGBT open-circuit fault characteristics under noisy environments is collected.Finally,an IEWT-CCE method for IGBT open-circuit faults is proposed,simulation verification of multiple types of fault conditions has been carried out.(3)The theory of support vector machine is studied,and the semiparametric SVM is used to diagnose the faults of IGBT open-circuit.The semiparametric SVM combines the advantages of linear and nonlinear SVMs,which reduces the influence of artificial setting of parameters and improves efficiency.However,the key issues lies in the selection of the basic element set and the calculation of the weights.Therefore,the sparse greedy matrix approximation algorithm is used to select the best basic element set,and the iterative re-weighted least squares are used to calculate the convergence weight.Based on this method,the fault diagnosis experiment of IGBT open-circuit is done.The effectiveness and practicability is verified by comparative experiments with other SVM methods.This paper aims at providing new method and idea for the fault diagnosis of IGBT open-circuit in the EVCS inverter.