Online Optimization and Feedback Elman Neural Network for Maneuvering Target Tracking

The uncertainty of maneuver model and nonlinear filtering, which are two difficult problems in practical application of maneuvering target tracking, are becoming the focus of research. Based on this, we propose an online maneuvering target tracking filter algorithm based on Elman neural network which can feedback while optimizing the estimation. Based on the Constant Acceleration (CA) model, the Elman neural network algorithm is used to obtain the size of the target maneuver and adaptive adjustment factor of noise covariance matrix, by online learning of the difference of the target state prediction and the optimal estimation, the innovation and the filter gain matrix, to real-time adjust optimal estimation and motion model. Mass of simulation experiments show that the proposed algorithm can effectively reduce the interference of the maneuvering of targets to the motion model during the target motion and improve the filtering performance. Under the condition of strong maneuvering, the tracking performance is far superior to Singer model, and also better than the IMM_ELM tracking filter algorithm.