Research On Matching Suitability For Underwater Geomagnetic Navigation

Underwater vehicles(UVs) are important tools for exploring the ocean. With the development toward deep ocean and long voyage, long-endurance and high-precision autonomous navigation has been a key technology of UVs. Geomagnetic navigation owns good merits such as all-weather, all-terrain, no-radiation, high-concealment and errorbounded, and integrating geomagnetic navigation with inertial navigation system(INS)may effectively meet with the requirement of UV navigation.The core of geomagnetic navigation is to achieve excellent matching precision, which is influenced not only by the matching algorithm but also the matching suitability of the geomagnetic map. The problem of matching suitability for underwater geomagnetic navigation(UGN) is investigated in this paper, which aims to select the suitable-matching areas and directions with high reliability and further provide support for underwater navigation and positioning.Firstly, the constraints of matching suitability for UGN are analyzed by considering the underwater environment and the characters of geomagnetic navigation, and further the basic frame of matching suitability analysis for UGN is built. Then following the frame, the premises of matching suitability analysis are discussed:(a) the size of candidate matching areas(CMAs) is modeled from the aspect of INS error propagation;(b)the database for matching suitability analysis is constructed by summarizing the basic suitable-matching features and giving the reasonable definition of matching probability.Afterwards the regional and directional matching suitability analysis methods are studied in detail.(1) A construction method of evolutionary synthetical features based on gene expression programming(GEP) is proposed.The correlation coefficient between the synthetical feature and matching probability is treated as the fitness function of GEP, and the synthetical feature can evolve by simulating the genetic process in biology. Then the optimal combination of basic suitablematching features, which are acquired by GEP search, will be taken as the evolutionary synthetical features. The proposed method may select the basic suitable-matching features adaptively and create diverse mathematical expressions, and therefore the obtained evolutionary synthetical features can effectively integrate the inherent advantage of the basic suitable-matching features. Experimental results show that the evolutionary synthetical features are robust, and the selected suitable-matching areas own high mean matching probability.(2) A self-organizing optimization method for classifying CMAs is proposed.The problem of classifying CMAs is studied from the viewpoint of pattern recognition, and a self-organizing optimization classification method based on genetic algorithm and support vector machine(GA-SVM) is proposed, where SVM is employed as the classier of CMAs, and GA is utilized for selecting the optimal feature subset and optimizing the SVM parameters to improve the classification performance. Meanwhile the one-against-one strategy is introduced to realize the multi-classification of CMAs. Experimental results show that the proposed method can greatly improve the classification accuracy of CMAs by feature selection and parameter optimization, and the classification results own low miscalculation risk.(3) A hierarchical decision-making scheme for directional matching suitability analysis is proposed.First the directional matching suitability of CMAs is analyzed form the aspects of frequency domain and spatial domain based on Gabor filtering and gray-level co-occurrence matrix(GLCM), respectively. Meanwhile the parameter settings of the above methods are also studied in order that the extracted directional features can be well consistent with directional matching probability. Then a complementary model between Gabor filtering and GLCM is established by adaptive neuro-fuzzy inference system(ANFIS), and further a hierarchical decision-making scheme for directional matching suitability analysis is designed. Experimental results show that the proposed scheme is effective, and its advantage lies in that it can select an appropriate directional matching suitability analysis method based on the characteristics of the given CMA and obtain more reliable analysis results than separately using Gabor filtering or GLCM.Finally, based on the theoretical fruits above, comprehensive simulation experiments of matching suitability analysis for UGN are implemented. On the one hand, the influence of downward continuation on matching suitability is analyzed, and it is found that the matching suitability may be improved after downward continuation; on the other hand,the suitable-matching areas and directions are selected by using the above-mentioned matching suitability analysis methods, and results show that the selected areas and directions own high matching probability, where excellent matching precision can also be got. Moreover a route planning strategy for UGN is proposed under matching suitability constraints, and the obtained route can successfully avoid the threats and traverse the suitable-matching areas following the motion constraints of UVs. The above experimental results further validate the effectiveness of the proposed methods in this paper, and the conclusions can provide beneficial guidance for geomagnetic matching and route planning.