Gravity Matching Navigation Algorithm Based on Spatial-Frequency Multidimensional Feature Fusion

The gravity matching algorithm plays a crucial role in gravity matching aided underwater navigation. At present, most matching algorithms are constructed based on correlation calculation between numerical gravity measurement sequence with reference maps that the performance of this single spatial domain feature correlation mode may be constrained under rough marine environments in practical tests. In this paper, a novel gravity matching algorithm based on spatial-frequency multidimensional feature fusion (SFM) was proposed. In this algorithm three new frequency features were extracted to present the complete frequency domain characteristics of the gravity measurement sequence: Spatial-frequency Trend Concordance Degree, Spatial-frequency Joint-Correlation Comparison, and Spatial Directional Similarity. Then they are fused with spatial-domain numerical features to construct a multidimensional feature correlation evaluation index for gravity matching navigation. To verify performance of the SFM algorithm, simulation experiments were conducted across eight oceanic regions exhibiting diverse gravity distribution. The results demonstrated that under different conditions, SFM algorithm maintained much better positioning accuracy and stability than other algorithm. Furthermore, marine measured gravity data from the South China Sea was employed for verification experiments. The results indicated that the SFM algorithm consistently achieved a positioning accuracy better than 2.0 nautical miles along each survey line, while other algorithms suffered from emerging mismatches in segments with increased measurement errors. Benefit from the frequency features extraction and spatial-frequency multidimensional feature fusion, the proposed SFM algorithm shows the advantages of robustness and accuracy even under challenging conditions in both simulation and measured data experiments.