Prediction of Electromagnetic Pulse Irradiation Response Based on Support Vector Machine

Abstract

Due to the complexity of electromagnetic simulation method based on the the mechanism analysis, this paper studies the feasibility of the system identification modeling method based on support vector machine for the prediction of electromagnetic pulse response of electronic equipment, and introduces the genetic algorithm for the model parameter selection. First, we use this method to verify the response of cable irradiated by the EMP. The results show that this method can accurately predict the cable response under different pulse front. Then, we use the experiment data of an UAV electronics system irradiated by the electromagnetic pulse to model and test. The results show that when using the low-amplitude field-strength data as the model training set, the identification model can accurately predict the pulse-coupling waveform under high field strength. The successful application of this method provides a simple and efficient simulation technique for the prediction of electromagnetic pulse response of electronic equipment, which has certain application prospect.