The Improved Maneuvering Model Algorithm Based on Dynamic Feedback Neural Networks for Online Learning
Shuyi Jia, Liqiang Ren, Haipeng Wang, Tiantian Tang · 2023
Tracking of maneuvering targets plays an important role in sea battlefield situation awareness and threat assessment. To solve the problem of low prediction accuracy of the traditional prediction method and model, an hybrid filter algorithm based on dynamic feedback neural networks for online Learning is designed, which embedded the trained neural network with memory function into the state estimation step of the UKF filter to form a hybrid filter. Based on the input eigenvalues, the estimated error is predicted. This estimated error corrects the state estimation in real time and realizes the online monitoring of maneuvering. The simulation results show that the algorithm has a strong adaptability to the target maneuvering form, and has better performance in terms of convergence and filtering accuracy.