Numerical Study of HEMP Coupling With Overhead Lines in Large Range by Using Deep Neural Networks

The study of high-altitude electromagnetic pulse (HEMP) coupling with overhead power lines is crucial for assessing potential damage to power systems. This paper presents a method that utilizes two deep neural networks in conjunction with theoretical analysis to rapidly compute HEMP coupling with overhead lines in the range of thousands of kilometers. This approach accounts for the varying characteristics of HEMP at different observation points and takes approximately 19.64 seconds on a notebook computer to compute the spatial distribution of the coupled voltage, which is about 2.88 million times faster than numerical methods. This efficiency reduces computational demands and demonstrates significant potential for applications in conventional nuclear physics. The results indicate that the spatial distribution of the peak coupled voltage resembles a "double egg yolk”, different from the smiley face of the HEMP. The results also show that the relationships between coupled voltage and blast parameters differ significantly from those observed in HEMP scenarios.