Method For Predicting Electromagnetic Response Of Sensitive Equipment Based On Neural Network

With the rapid development of high-speed railway systems,the electromagnetic environment inside and outside the Electric Multiple Units(EMUs)becomes more complicated.Ensuring that the on-board communication system and train control system equipment can work normally in the complex electromagnetic environment is a key link for the reliable operation of the high-speed EMUs.Taking the LTE-R roof antenna as a sensitive equipment,this thesis considers using electromagnetic data to focus on analyzing and predicting the radiated disturbance of the pantograph-catenary coupled to the antenna port.The main and difficult points of the research are analyzed separately.First,the data processing method of the neural network model is used to identify and warn the signal of the off-line discharge from the pantograph-catenary under specific conditions,and the cascaded neural network model is used to predict and analyze the coupling characteristics of the disturbance transmission.The specific research content and results are as follows:(1)Firstly,the data of the off-line discharge from the pantograph-catenary are obtained of the field measurement and simulation modeling under specific conditions,and the effectiveness of the simulation model is verified by comparing and analyzing the characteristics of field measurement data.and verified the effectiveness of the simulation model by providing the electromagnetic data for the signal identification using the neural network method.The electromagnetic simulation model provides electromagnetic data for the identification of signals using a neural network method.(2)Preprocess the electromagnetic data obtained from the off-line discharge from the pantograph-catenary to extract the characteristic eigenvector,and build a BP?RBF neural network model to realize the identification and early warning of the off-line discharge from the pantograph-catenary.The simulation results show that the sample data under specific conditions are simulated using the above two network models,and the recognition rate is above 85%.The study of the off-line discharge from the pantograph-catenary is the prerequisite for predicting the electromagnetic response of sensitive equipment.(3)According to the numerical theoretical calculation of electromagnetic field to study the longitudinal distribution law of electromagnetic field with frequency and distance.To compare and verify the longitudinal distribution law of the radiation electric field intensity of the pantograph-catenary model through electromagnetic simulation model,it provides theoretical support for studying the nonlinear relationship between the defined field strength characteristic parameters and the spatial position and frequency.(4)The electromagnetic simulation model is used to provide data for the analysis and definition of the field strength characteristic parameter prediction,and the neural network model is constructed and the cascade neural network model is improved.According to this method,the prediction of the disturbance intensity of the sensitive equipment port is realized.Three kinds of neural network models are used for simulation,and the results show that the error of prediction results based on the PSO-BP cascaded neural network model is within 5%,and the field strength of the antenna port can be accurately predicted based on the electric field strength of the window,and the cascade model improves the prediction accuracy well.(5)The predicted electric field intensity is added to the LTE-R wireless channel to analyze the impact of the disturbances on the block error rate of the LTE-R system.The signal-to-noise ratio is in the range of 0?16dB,the poor channel environment increases the block error rate of the LTE-R system,but the order of magnitude is still 10^-1.The increased block error rate will lead to the system throughput decline,thereby affecting the system transmission efficiency.This study utilizes the data samples of electromagnetic data of field measurement and electromagnetic simulation to study the electromagnetic response of high speed EMUs sensitive equipment through quantitative and digital representation of data,which provides a new method for studying electromagnetic compatibility of on-board sensitive equipment to some extent.