A Fast Method for Calculating the High Altitude Electromagnetic Pulse Environmental Parameters Based on Neural Network

Abstract

High altitude electromagnetic pulse is a type of electromagnetic pulse with high field strength and wide coverage range. It can enter electronic devices or systems and cause serious damage to them. It's necessary to know how high-altitude electromagnetic pulses distribute on the ground. This helps in quickly evaluating their damage to electronic systems under different conditions. Due to the traditional calculation method requiring a calculation time of 2 h, real-time calculation and evaluation cannot be achieved. To solve the problem of real-time evaluation calculation, through research and analysis, only artificial intelligence methods can achieve rapid calculation by approximating physical processes. Neural network methods exhibit superior performance in nonlinear approximation of physical processes and functions. To make the established neural network model more accurate, a semi-supervised multilayer feedforward method is proposed, using a semi supervised strategy to find the optimal modeling parameters. The experimental results show that this method has a high approximation accuracy for this traditional calculation process. The field distribution calculation of high-altitude electromagnetic pulses based on the semi-supervised method only takes 0.09 s. Compared to the traditional calculation time of 2 h, 0.09 s can be used for real-time calculation and evaluation. This fast calculation method based on a semi-supervised neural network solves problems that traditional numerical calculation methods cannot solve. It can be truly applied to real-time evaluation and decision-making in the damage assessment of electromagnetic pulses.