自主水下航行器海底地形与重磁场匹配方法OA

To overcome these issues,a multi-source fusion matching method based on a Bayesian estimation framework is proposed,unifying terrain and geomagnetic-gravity matching within a single probabilistic model.The approach employs the unscented Kalman filter(UKF)for nonlinear state estimation,enhanced with simulated annealing for online noise covariance adaptation.A triangle descriptor-based graph isomorphism method is designed for robust terrain correspondence extraction,and a fusion error band-constrained strategy is developed for optimal selection in gravity and magnetic matching.The method effectively improves robustness under large initial errors,reduces false associations caused by low overlap or ambiguous anomalies,and enables tight integration and consistent state updates across multi-source data.Simulation results show the proposed method suppresses navigation drift and achieves higher matching success rates and positioning accuracy in complex seafloor settings,offering a reliable solution for long-endurance,high-precision AUV navigation.