Research on Jamming Recognition Based on Time-Frequency Domain Weighted Fusion and Attention Mechanism

The jamming recognition of target detection aims to achieve rapid judgment and effective response to jamming by analyzing the target echo signals. This paper addresses the shortcomings of the existing methods in terms of jamming recognition capabilities and practical effectiveness and conducts research on jamming recognition based on time-frequency domain fusion and attention mechanism. First, by analyzing the principles of target detection and jamming effects, a multi-terrain random fluctuation model for ground detection is established. Second, the time-frequency domain weighted fusion method is proposed. Taking multi-period time domain + time-frequency domain as jamming recognition information, combined with the attention mechanism, the jamming recognition model based on time-frequency domain weighted fusion and the attention mechanism (TFWF-AM) is established. Then, the single jamming and compound jamming sample sets are established by superimposing the beating signals of target echo and multi-jamming. Finally, the accuracy of the TFWF-AM jamming recognition model is compared with that of existing method models, and the effectiveness of multi-period time domain + time-frequency domain information is verified. The results show that the TFWF-AM jamming recognition model has the highest accuracy for both single jamming and compound jamming, reaching 99.92% and 99.56%, respectively, which is 10.42% and 52.81% higher than that of the feature fusion model. This research holds huge significance for the perception and decision-making of target detection systems in complex electromagnetic environments.