Novel convolutional neural network and Wigner-Ville distribution combined radar signal classification method

Abstract

Traditional radar signal classification methods usually manually analyze and extract various low-intercept probability radar signal features and then carry out classification by utilizing the extracted features, so that the classification correctness in practical application is relatively low. The invention discloses a pseudo Wigner-Ville distribution analysis and novel convolutional neural network model combined method for classifying low-intercept probability radar signals. The method comprises the specific steps of: 1, intercepting an LPI radar signal; 2, carrying out Wigner-Ville distribution processing on the intercepted radar signal to obtain a radar signal image; 3, normalizing the processed radar signal image; 4, classifying the radar signal image on the basis of a novel convolutional neural network; and 5, outputting an LPI radar signal classification result. The radar signal classification method provided by the invention is capable of automatically extracting radar signal features, along with the increase of collected radar signal data, the classification correctness is enhanced, so that the method has self-adaptation ability and has important significance for enhancingthe electronic countermeasure ability of the country.

Description

technical field [0001] The invention relates to a technique in the field of electronic countermeasures, in particular to a novel radar signal classification method. Background technique [0002] In order to improve the reconnaissance and survivability of modern radars in electronic countermeasures, more and more radars currently use Low Probability Intercept (LPI) signals, which use special transmission waveforms to prevent non-cooperative receivers from intercepting, detected signal. In order to effectively jam radars with LPI signals, after intercepting LPI radar signals, the jammer must quickly and accurately classify them, judge the signal type of the enemy radar, and take targeted jamming measures. Therefore, accurate classification of radar signals is an important basic work in radar countermeasures, and has important practical significance and application value in radar countermeasures. [0003] Convolution neural network (CNN) has been widely used in image classifi...

AI Technical Summary

Problems solved by technology

[0006] If you use existing CNNs, such as AlexNet, VGG, Inception, etc., although no training is required, it is found through simulation experiments that the classification accuracy is not ideal, and the stability is poor, which is quite different from the accuracy requirements of the real environment.

Images