Intention Inference Method for Reentry Glide Vehicle Based on Double-Layer BPA Fusion

Aiming at the problem of difficulty in inferring vehicle attack intent due to the avoidance of no-fly zones by reentry glide vehicle, a reentry glide vehicle intent inference method based on the fusion of double-layer Basic Probability Assignment (BPA) is proposed. Firstly, based on the artificial potential field method and the pseudo heading deviation angle, intention cost functions are constructed in the distance dimension and angle dimension, respectively, and the likelihood probability is estimated by quantifying the cost of different intentions. Then, the first layer BPA fusion model is proposed, which utilizes the D-S evidence theory based on weighted averaging to fuse likelihood probabilities from different dimensions and avoid conflicts between evidence. Finally, the second layer BPA fusion model is proposed, which weights and fuses the posterior probability of the previous intention with the prior probability of the current intention to solve the problem of Bayesian recursive formula failure when continuously inferring vehicle intention. The simulation results show that compared with the latest method, the method proposed in this paper shortens the time to infer the true intention of the target by 44.62%, and increases the inference probability by 33.48% when the inference method is effective.