Maneuvering Target Tracking Based on LSTM for Radar Application
Shixiang Cai, Shengli Wang, Mingjie Qiu · 2023
Aiming at the problem of model mismatch in traditional model-based target tracking methods when the target undergoes maneuvers, an adaptive tracking method based on Long Short-Term Memory (LSTM) is proposed. The algorithm replaces traditional filtering algorithms with LSTM and introduces the error between the measurement and the predicted value during training. This enables the network to simultaneously acquire information about the target position and measurement error, and during prediction, use the measurement error to directly modify the network's prediction results to obtain the final tracking results. Simulation results demonstrate that this method effectively solves the problem of describing the motion state of maneuvering targets, and has higher tracking accuracy and stronger adaptability compared to traditional filtering algorithms.