Electromagnetic Emissions Measurement And Data Analysis Of High-speed Train

With the rapid development of high-speed railway system,the electromagnetic environment of high-speed train is becoming more and more complicated while the electromagnetic compatibility performance is directly related to the safety,reliability and usability of train operation.Electromagnetic environment of the high-speed train is obviously particular which is closely related to the complicated and variable operation conditions,thus,the analysis of train electromagnetic environment faces too many challenges,for example,the undetermined causal relationship,overmany parameters.Traditional statistical analysis bases on the rigorous theoretical reasoning,through the actual situation to simplify the construction of the analysis model,which makes the classical statistical analysis does not apply to high-speed train electromagnetic compatibility research.Data mining can specialize in large data whose information density is very low to extract the knowledge information.Therefore,this thesis draws on the method of data mining to analyze the electromagnetic emission measurement data of high speed train.The contents and innovations of this thesis mainly include the following two aspects:(1)In order to clear the mining target,this thesis first carries on the thorough theoretical analysis to the application domain.This thesis discusses the disturbance source,coupling path and sensitive equipment of high-speed train from the three elements of electromagnetic compatibility measurement.About the disturbance source this thesis mainly discusses the low frequency magnetic field disturbance caused by the traction and inverter system,the high frequency radiation disturbance generated by pantograph part off line and the rich harmonic waves of train traction current.Meanwhile,this thesis discusses the coupling path includes conduction coupling,radiation coupling and conduction coupling between cables.Then,this thesis discusses the anti-jamming of vehicle-sensitive equipment includes signal transmission and information processing equipment,control signal operation equipment and other electrical equipment by their working principles.Finally,according to GB/T24338(1-5)developed a field measurement program and carried out field experiments,completed the initial data collection and collection work.(2)The thesis builds analytical models bases on the BTM and SPD time domain measurement data,frequency domain measurement data,train GPS data and operation condition recording data in the original measurement data.Firstly,the thesis uses ARMA timing analysis model to analyze the time-domain measurement data in order to realize the prediction of electromagnetic compatibility data,then uses CART decision tree,BP neural network and RBF radial basis function neural network to analyze the BTM frequency domain measurement data in order to realize the classification of train operation conditions.Meanwhile,the thesis adopts ROC diagram computational rate as well as CPU occupancy rate to evaluate the models.The results shows that the CART can be used when there are only two classification types and the RBF should be used when the classification types are more than two.Finally,the thesis adopts outlier analysis model based on clustering algorithm to analyze the frequency domain measurement data of BTM so as to realize the disturbance judgment about sensitive equipment.