Radar active jamming identification method based on CWD and improved AlexNet

Abstract

The invention provides a radar active interference identification method based on CWD and improved AlexNet. The radar active interference identification method is used for identifying radar active interference signals in communication signals. The method comprises the following steps: firstly, modeling different types of radar active interference signals, and superposing noise of a communication signal receiving site on the various radar active interference signals; performing CWD time-frequency analysis on the various radar active interference signals to generate a two-dimensional CWD time-frequency graph; and an improved AlexNet model is adopted to identify the CWD time-frequency diagrams of the various radar active interference signals. According to the radar active interference identification method based on the CWD and the improved AlexNet, an interference identification algorithm combining a CWD time-frequency analysis method and an improved AlexNet model is adopted, the characteristics of radar interference signals can be more obvious through time-frequency analysis, and the characteristics of the radar interference signals can be more obvious through the improved AlexNet model by changing an activation function and an optimizer of the improved AlexNet model. Therefore, the method still has a good interference signal recognition rate under the condition of a communication signal with a low interference-to-noise ratio.

Description

technical field [0001] The invention relates to the technical field of radar active jamming signal identification, in particular to a radar active jamming identification method based on CWD and improved AlexNet. Background technique [0002] With the continuous improvement of the level of science and technology, the modern battlefield environment is increasingly complex, and higher requirements are put forward for the identification of radar jamming signals. When the noise completely covers the interference signal, the feature parameters extracted at this time can only reflect the law of the noise, and have no identification significance. The extraction process of feature parameters is also very complicated. For an interference signal, multiple feature parameters are often extracted to reflect the characteristics of interference signals in different domains. This process of feature parameter extraction consumes a lot of computational costs. [0003] Aiming at the problem of...

AI Technical Summary

Problems solved by technology

[0006] However, in the early stage of BP neural network training, it is necessary to extract the features of the interference signal. The feature parameters extracted by the existing feature extraction algorithm are relatively fixed, and the low SINR still has a great impact on it. , the recognition rate of BP neural network is often low

Images