Discrimination of nuclear explosion and lightning electromagnetic pulse using S-transform

In this paper, the S-transform is used to analyze the nuclear explosion and lightning electromagnetic pulses for NEMP and LEMP signals based on their non-stationary and non-linear characters. The region features of S-transform time-frequency spectrums of NEMP and LEMP signals are calculated to discriminate them. Then, the discrimination of NMEP is studied using nearest neighbor classification. At last, the discrimination result is analyzed based on the different mechanisms of the nuclear explosion and lightning. The experiment results indicated that, using the nearest neighbor pattern classification method, the average discrimination rate of NEMP and LEMP is over 85% based on angle relative to lateral axis region features of time-frequency spectrum. Finally, we get a conclusion that the image region features of the S-transform spectrum are effective features to discriminate NEMP and LEMP, which is possible to provide technical support for real-time, fast and effective monitoring of foreign nuclear test, and provide basis for decision making for the intelligence agencies.