Radar target tracking algorithm based on CNN-Attention-LSTM deep learning network

In the article, the traditional target tracking algorithm has the problem of large error or dispersion of filtering when dealing with nonlinear and complex noise models, and proposes a target tracking algorithm based on CNN-Attention-LSTM deep learning network, which effectively integrates the local feature extraction capability of CNN-network, the temporal analysis capability of LSTM-network, and the focusing characteristic of Attention mechanism, so that it can effectively capture various complex motion states of the target and complete high-confidence tracking. The method effectively integrates the local feature extraction capability of CNN-network, the time-series analysis capability of LSTM network and the focusing characteristic of Attention mechanism, which enables it to effectively capture various complex motion states of the target and complete the tracking with high confidence. The experimental results show that the CNN-Attention-LSTM algorithm outperforms the traditional EKF algorithm in target tracking and filtering in different motion states, weakens the dependence on the accurate description of the motion model, and significantly improves the intelligent maneuver recognition ability, which has high value for engineering applications.