Radar signal modulation identification method based on time-frequency analysis and machine learning

Abstract

The invention relates to a radar signal modulation identification method combining time-frequency analysis and machine learning, and belongs to the technical field of radar signal sorting identification and machine learning. The method comprises the following steps: performing pulse processing based on a multi-phase filtering structure on a pulse flow signal, and estimating parameters such as pulse carrier frequency and arrival time; performing time-frequency analysis on the extracted pulse, and preprocessing the obtained time-frequency image; putting the preprocessed image into a residual network for training, and carrying out the primary recognition of an intra-pulse modulation mode; carrying out time-frequency diagram feature extraction on modulation modes such as multi-phase codes which are not easy to identify; sending the extracted features to a random forest classifier for training and testing; and obtaining a final intra-pulse modulation mode identification result. According tothe method, the intra-pulse modulation mode of the low-interception-probability radar signal can be accurately recognized under the condition of a low signal-to-noise ratio, and the method has the advantages of being good in noise immunity and high in correct recognition rate, has high adaptability to signal parameter changes and has certain application value.

Description

technical field [0001] The invention belongs to the technical field of radar signal sorting recognition and machine learning, and in particular relates to a radar signal modulation recognition method combining time-frequency analysis and machine learning. Background technique [0002] In modern warfare, information control has increasingly become a key factor in determining victory on the battlefield. As far as radar electronic reconnaissance is concerned, combat deployment depends on the capture of radar radiation signals and information acquisition. After measuring and analyzing the parameters of the intercepted radiation source signal, the reconnaissance system uses the collected parameter information to sort and identify the signal, make effective judgments on the information of the enemy radar, and interfere according to its modulation mode. Only by correctly identifying the signal can the initiative be gained. Identifying the intrapulse modulation of the radiation sour...

AI Technical Summary

Problems solved by technology

However, the correct rate of directly using the neural network for time-frequency image classification is related to the quality of the time-frequency image. If the effect of the time-frequency image is not good or the time-frequency images of the two modulation methods are relatively similar, it will affect the learning effect of the neural network and reduce the accuracy. Recognition rate

[0006] In summary, both methods have shortcomings. If only feature extraction is considered, the correct recognition rate will be affected when the feature selection is inappropriate, and neural network modulation recognition using only time-frequency images will also be affected by the similar time-frequency images of some signals. Correct recognition rate

Images