Electromagnetic pulse recognition based on Tsfresh BP

DOI:	10.11809/bqzbgcxb2023.12.035
Keywords:	electromagnetic pulse; cable coupling; feature extraction; neural network; recognition
Abstract:	In order to classify different kinds of electromagnetic pulse, a BP neural network classification method based on Tsfresh feature extraction is proposed. The proposed Tsfresh BP method uses Tsfresh to statistically calculate and screen the features of the initial data set through the steps of aggregation features, feature significance testing, feature selection and others. Then the processed features are input into the portable BP neural network designed to realize the classification and recognition of electromagnetic pulse and cable types. The experimental results show that under the simulation model and scene of the electromagnetic pulse irradiated cable, nine types of cable coupling current simulation data sets are obtained for different electromagnetic pulse and cable types. The proposed method is verified on the simulation data set, the classification performance is better than that of decision tree and support vector machine, and the average classification accuracy is over 98%.
Issue:	Vol. 44 No. 12 (2023)
Published:	2023-12-28
PDF HTML