Aiming at the problem that existing interference identification algorithms rely on feature engineering to extract fea-tures,which is cumbersome and the identification accuracy is greatly affected by the value of signal-to-noise ratio,the interference type identification method of Beidou navigation satellite system(BDS)based on learning from time-frequency graph under different signal-to-noise ratio is proposed in this paper.Taking the B1I signal of the airborne BDS as the object,the original B1I signal and the B1I signal containing interference are subjected to short-time Fourier transform.The time-frequency graph obtained after the transform is used as the input vector of the support vector machine and convolutional neural network model to complete the detection and identification of interference types.The simulation results show that the average identification accuracy of both machine learning identification algorithms has reached over 99％,which is about 30％higher than that of traditional decision tree identification algorithms,solving the problem of existing interference identification algorithms that heavily rely on manually designed feature engineering to extract interference signal features and have low accuracy.The research results can provide prior information for subsequent interference suppression work and improve the safety of BDS in the aviation field.
维普
万方数据
