Reference composite features database construction method based on track segmentation for gravity matching aided navigation

Abstract

The reference database or maps is the most important foundation of the gravity matching aided navigation system, which can directly affect the navigation performance of the system and determine the level of matching positioning accuracy. The matching algorithm is usually designed according to the form of database, and its principle is closely related to the structure of database. In this paper, a reference composite features database construction method based on track segmentation (TS-RFD) is proposed. The TS-RFD method mainly can extract texture features from gravity reference data, and determines 8 kinds of features such as index feature and matching feature to establish a reference composite features database. Combined with the search structure of key-value pairs, index feature is used as keys and matching features as values to propose the track segmentation search strategy for matching, and the final positioning result is obtained. Compared to the traditional numerical gravity reference databases or maps which briefly constructed from just gravity anomaly values, with the proposed TS-RFD method various matching features derived from gravity distribution and change are composited to construct a reference composite features database. When the new TS-RFD based database is loaded for gravity matching aided navigation system instead of traditional numerical database, it can significantly enhance efficiency and accuracy, and maintains robust stability. Six subregions in the South China Sea with different gravity distribution and changes were chosen for simulation experiments to verify the performance of proposed database construction. In these six subregions, matching positioning tests along 4200 simulated tracks were carried out. Simulation results show that, with gravity features database constructed from TS-RFD method, both matching accuracy and stability of gravity matching navigation are distinctly improved that obviously superior to the gravity matching with traditional pure numerical database. Moreover, the gravity matching navigation tests were conducted utilizing real measured gravity data in the South China Sea. Measured data test results indicate that after 150 times matching positioning along the track, the average positioning accuracies of matching navigation experiment with traditional numerical database are more than 1.45 nautical miles. In contrast, the average positioning accuracy with TS-RFD database can achieve 1.19 nautical miles, the stability and success rate are still significant advantages.