CHENG Lu1,2,3, WANG Xueqing1,2,3, LIU Yan3, CHENG Xianyou4, XU Xin1,2,3, JI Yicai1,2,3, FANG Guangyou1,2,3
JOURNAL OF VIBRATION AND SHOCK. 2023, 42(8): 275-281.
In order to accurately and efficiently identify nuclear electromagnetic pulse (NEMP) and lightning electromagnetic pulse (LEMP), a recognition and classification method based on adaptive signal decomposition and ensemble learning was proposed. First, for the problem of sample imbalance, data enhancement methods were used to preprocess the data set. In addition, Hilbert-Huang transform was applied to perform adaptive signal decomposition on NEMP and LEMP respectively. Then, features of the decomposed signal in the time domain, frequency domain and wavelet domain were extracted. Finally, the ensemble learning algorithm was used to identify and classify the extracted features. Experimental results show that the accuracy of the method on the measured data can reach more than 99.99%, and the false alarm rate of LEMP signals is less than one in ten thousand.