HEMP EXCITED SHIELD RESIDUAL ELECTRIC FIELD MODELING METHOD BASED ON NARX NEURAL NETWORK

To predict the residual electric field inside an electromagnetic (EM) shield under illumination of different HEMP waveforms, a method based on NARX neural network is proposed in this paper. The model can be established from input-output data of EM shield without knowing enclosure and internal structural details. To evaluate the precision of the prediction method, two error criteria based on energy and field amplitude are provided in this paper. As a numerical example, the double exponential pulse with 10% to 90% rise time of 2.5 ns, the pulse width at half maximum of 23 ns, and the corresponding residual electric field are taken as the training data. The EM simulation is used to establish the model of residual electric field inside the shield. The NARX neural network is then built and trained. Other double exponential pulses, with different rise times and pulse widths, and their residual field are taken as the checking data. The results show that the error of the prediction method is sufficiently small for actual use.