Condition Monitoring And Lifetime Estimation Method Of Dc-link Capacitors In High-speed Train Traction Converters

With the increasing mileage of high-speed railways,the high-speed trains will face more complex operating environment and more changeable working conditions.In addition,with the increasing operating time of high-speed trains,the problems such as fatigue cumulative damage and failure rate increase of key components will be inevitable,which will threaten the safe operation of high-speed trains.Therefore,it is of great significance to carry out reliability research for key components of high-speed trains.This paper takes the DC-link capacitors and resonant capacitors of the traction converter as the research object,and focuses on their condition monitoring and lifetime estimation methods.The specific research work of this paper is as follows.Firstly,the topology of the traction converter and the functions of its components are outlined,the control methods of the traction rectifier and the traction inverter are introduced.In addition,the roles of the pre-charging and fast-discharging stages of the traction converter are analyzed to lay the foundation for circuit analysis and model building in this paper.Moreover,the physical structure of the metalized film capacitor is introduced,the equivalent circuit model of the capacitor is established,and the impedance characteristics,temperature characteristics and frequency characteristics of the metalized film capacitor are analyzed,and the failure modes of the metalized film capacitor are summarized.Secondly,the failure mechanism of metalized film capacitors shows that the reduction of capacitance can reflect capacitor’s ageing process.Therefore,combining with the equivalent circuit model of the capacitor,this paper analyzes and builds the model of pre-charging stage of traction converters.Then,a second-order resonant circuit is constructed by specific control method,and an analytical model of the DC-link voltage in the resonant state is derived.According to the measurable information such as DC-link voltage and grid-side current,a circuit model-based condition monitoring method is proposed in this paper.The circuit model-based condition monitoring method requires additional operating state which should be designed before the normal operation of the traction converter,it is not convenient to practical application.Therefore,this paper analyzes and builds the model of fast discharge stage of traction converter,and the characteristic parameters in the DC-link voltage are extracted by non-linear fitting algorithm.Then,the mapping relationship is constructed according to the strong correlation between the characteristic parameters and the capacitance of DC-link capacitor,the capacitance of resonant capacitor and the inductance of resonant inductor.Finally,a BP neural network-based condition monitoring method is proposed to monitor the capacitance of DC-link capacitors and the resonant capacitors without additional operating states.Finally,in order to provide a theoretical basis for the maintenance schedule of capacitors in traction converter,a mission profile-based lifetime estimation method is researched.To address the shortcomings of the method which doesn’t consider the degradation of the parameters during the ageing of the capacitor,the electro-thermal analysis process of the capacitor is modified by the ageing model of the metallized film capacitor,and the lifetime of the capacitor is evaluated based on the life estimation model,the cumulative damage model and the weibull distribution,then the life estimation and analysis are carried out by taking the DC-link capacitor as an example.