Improved Neural-Network-Based Source Reconstruction for Estimating the Emission of Linear Synchronous Motors in Vehicle Bodies

Abstract

To quickly and accurately characterize the effects of linear synchronous motors (LSMs) in the electromagnetic suspension (EMS) maglev system on the surrounding electromagnetic environment, the field-source equivalent model for electromagnetic emission with obstacles in the near zone was built using cascade neural network that integrates convolutional neural networks (CNNs) and backpropagation neural networks (BPNNs). The influence of the train body on the electromagnetic emissions of linear motors is equated with the magnetic field shielding effectiveness. Results by the proposed model were validated by measurement and full-wave simulation results.