Research On Electromagnetic Disturbance Analysis And Prediction Method Of High-speed Maglev On-board Power Supply System

With the rapid development of high-speed maglev system,the importance of electromagnetic compatibility(EMC)research of high-speed maglev system is gradually highlighted.As an important power system in the high-speed magnetic levitation system,the on-board power supply system is related to the power consumption and stable operation of the on-board equipment;and as a source of electromagnetic disturbance in the load system,the electromagnetic compatibility of the on-board power supply system also directly affects the passenger safety and train operation safety.In this paper,the main electromagnetic disturbance faced by the on-board power supply system of the 600 km/h high-speed maglev vehicle are systematically studied.Firstly,based on the structure of high-speed magnetic levitation on-board power supply system,the main electromagnetic disturbances and the propagation coupling mechanism are analyzed,and the key points of various electromagnetic disturbance research and the required electromagnetic compatibility technologies and methods are summarized.Secondly,the impact of long stator linear synchronous motors on the low-frequency magnetic field environment is modelled and characterized.Thirdly,a joint bilinear timefrequency analysis method is proposed to accurately locate and track the time-varying characteristics of the time-varying electromagnetic disturbances in the on-board power grid.Fourthly,a DC arc identification and real-time impact prediction method based on deep learning technology is proposed for the active protection idea of electromagnetic disturbance,and a DC arc test platform is built to evaluate the performance of proposed method.Based on the above research content,the main innovations of this paper are summarized as follows.(1)An equivalent modeling method for the radiation emission characteristics of three-phase motors is proposed to solve the problems of low efficiency of numerical simulation methods,complicated calculation of equivalent modeling methods,and limited solving capability for different working conditions.Based on the source reconstruction theory and electromagnetic field superposition principle,the radiation emission characteristics of the motor under the excitation of each phase cable are modeled independently,so that the model has the ability to analyze the characteristics under different operating conditions such as three-phase balance and phase-loss fault;a neural network is used to fit the complex inverse radiation problem in the source reconstruction process,which reduces the complexity of the equivalent modeling process while ensuring the accuracy of the model.(2)A time-frequency analysis method is proposed for the time-varying electromagnetic disturbance characteristics of the on-board power grid of high-speed maglev trains to solve the problems of low resolution and low accuracy of the existing time-frequency analysis methods in the time-frequency analysis of low sampling rate measurement results.Based on the high time-frequency resolution of the bilinear timefrequency analysis method,the cross-term interference in the bilinear time-frequency analysis process is filtered by the adaptive optimal kernel function,and the complete timefrequency characteristics are reconstructed by the compressed sensing technique to achieve the accurate description of the time-varying electromagnetic disturbance characteristics in the on-board power grid of high-speed maglev trains.(3)The DC arc harassment identification technology of the receiving boot-supply rail and its real-time prediction technology on the line impact are proposed to realize the accurate identification of the arc state and real-time prediction of the load line state.Using the rich features of electromagnetic disturbance in the time-frequency domain,the timefrequency features of arc disturbance and line state are extracted by using convolutional neural network in the vehicle power supply system,and the time-series correlation of each feature is fitted by using long and short-term memory network,so as to realize the identification of arc state and real-time prediction of load line state.The above innovative research results provide a technical means to comprehensively understand the EMC characteristics of the on-board power supply system of the 600 km/h high-speed maglev test vehicle,and have certain theoretical value and guidance significance for optimizing the EMC of high-speed maglev system.