Reliability Evaluation Of Dc-link Capacitor Banks In The High-speed Train Traction Drive System

In the high-speed railway traction drive system,due to the complex operating,physical structures and environmental conditions,and the strict requirements for safe operation,the reliability of the DC-link capacitor banks faces the huge challenges.This paper presents a DC-link capacitor banks reliability evaluation method suitable for traction drive systems.It mainly studies the analysis of the electro-thermal stress of the capacitor banks,the evaluation of the hot spot temperature and the lifetime bottleneck,and the optimal selection and design of the capacitor banks.Combined with the multi-operation conditions of high-speed train,more complex mission profiles need to be handled in the traction drive system.In the analysis of electrical stress,a method of obtaining the power loss curve by combining the discrete Fourier analysis with the sliding window group and the neural network is proposed,obtaining more realistic electrical stress profiles.In the thermal stress analysis,the actual physical structure of the capacitor banks is characterized by multiphysics modeling,and a bidirectional thermal model is established to estimate the thermal stress under dynamic power loss profiles.In addition,the effects of thermal capacity,thermal coupling,and air cooling are considered.The aging of the capacitor further affects the change of the capacitor parameter.Based on the analysis of the electro-thermal stress of different capacitor parameters,under the stability analysis and the rated temperature limit,the capacitor failure standard that meets the requirements for safe operation of trains is proposed.In different operating timescales,the lifetime distribution change with the dynamic changes of the complex mission profile.The lifetime bottleneck of the DC-link capacitor banks is evaluated by Weibull distribution and linear superposition law.In addition,the experimental platform of prototype is used to verify the rationality of electro-thermal stress modeling.Based on the above analysis,the DC-link capacitor optimized lifetime evaluation model considering the influence of power module failure is further analyzed.In terms of electrical stress,four fault models of power modules aree established based on fault location,and the electrical and thermal stresses of capacitors corresponding to different power modules are calculated.Based on the lifetime loss rate of the fault,the impact of the power module fault on the capacitor reliability is quantified.An active design scheme of thin DC-link capacitors for traction drive system based on deep learning is also provided.Based on the collection of electrical and performance parameter data related to film capacitors,the deep neural networks(DNN)are established.It can be used to quickly map performance parameters to the volume,cost,and lifetime expectancy of DC-link capacitor in the traction drive system.At the same time,the performance parameters of the capacitor can be used as a black box to directly select and design the DC-link capacitor based on volume,cost and lifetime.In this case,two different test cases are set up,and the performance comparison of the two schemes is given.