Identification of Broadband Electromagnetic Interference Sources Based on Joint Optimization of Feature Screening and Classifier

A broadband electromagnetic interference (EMI) source identification method for civil aviation communication and navigation is presented in this paper. The spectrum data of various broadband EMI sources are first preprocessed and segmented into several segments, and after that appropriate features are extracted based on the analysis of the waveform characteristics of the spectrum data. Then a novel feature screening algorithm, called center distance weighting (CDW), is proposed to perform further feature screening on the initial feature set. Finally, the random forest (RF) classifier is used to classify and identify the interference sources based on the screened feature dataset. To overcome the difficulty of parameters setting of the feature screening algorithm and classifier, and improve the identification accuracy and efficiency, adaptive particle swarm optimization (APSO) is designed to jointly optimize the parameters of CDW and RF algorithms. The test results with real data show that the proposed method can effectively identify the broadband EMI sources for civil aviation communication and navigation, and the overall accuracy of the proposed method reaches 98.73% for the identification of multiple different EMI sources.