Research On Underwater Gravity Aided Positioning And Gravity Map Navigability Analysis

The inertial navigation system(INS)is the most important navigation system in the underwater vehicle.For the inherent characteristics of inertial sensors,the positioning error of the INS is accumulated and diverged over time,which makes it difficult to ensure the long-term positioning accuracy.Other methods are needed to correct the positioning errors of INS periodically.Compared with the commonly correction methods such as satellite positioning system,radio positioning and terrain aided positioning,the gravity measurement is a passivity process,and the gravity aided positioning could effectively protect the underwater vehicle’s concealment in the mission while providing reliable positioning information to correct the positioning error of INS.So,the gravity aided positioning has very important value in military missions.Gravity aided positioning system is mainly composed of INS,gravity anomaly measuring instrument,gravity anomaly map and matching algorithm,and so on and so forth.The performance of each component has an important influence on the performance of gravity aided positioning system.This paper takes the underwater vehicle working environment as the background,combined with the characteristics of the Earth’s gravity field and the application requirements of gravity aided positioning system,and studies the components that affect the preformation of gravity aided positioning system.Gravity anomaly map is the basis of gravity aided positioning.Both of gravity features and the resolution of gravity anomaly map will affect the performance of gravity aided positioning system.The paper expounds the advantages and disadvantages and data processing methods of common gravity anomaly data sources such as satellite gravity anomaly,ship gravity anomaly,digital elevation model(DEM)forward gravity anomaly and global gravity field model.Gravity anomaly from different sources reflects different information.The multi sources gravity anomaly fusion method based on discrete wavelet transform is studied in this paper,and the frequency coefficient fusion methods are designed respectively.The optimal wavelet basis function and decomposition level of wavelet decomposition are confirmed by experiments.Experiments show that the multi sources gravity anomaly fusion method based on discrete wavelet transform can effectively fuse the multi sources gravity anomaly.Matching algorithm is an important part of gravity aided positioning system,and its performance determines the positioning effect of the positioning system.This paper explains the principle of Iterative Closest Contour Point algorithm(ICCP)and real-time ICCP algorithm.The existing real-time ICCP has the problem that the length of the matching sequence is fixed and setting according to experience.That will can not obtain the optimal positioning accuracy on all tracks.In the process of searching for the closest contour point,there is a large amount of calculation.This paper proposes a real-time ICCP algorithm with optimized matching sequence length(OMSL-ICCP)to solve these problems.The optimal matching sequence length of OMSL-ICCP under the current measured gravity anomaly sequence is obtained by the golden section search algorithm.The Hausdorff distance is utilized to obtain the search range of the closest contour point,which can effectively shrink the search range of the closest contour point.In the simulation tests with different gravity sensor measurement noises,different INS positioning errors and different gravity map resolutions,the difference of positioning performance between the existing real-time ICCP and the OMSL-ICCP is compared.Under the same test conditions,the simulation results show that the positioning accuracy of OMSL-ICCP is higher than the positioning accuracy obtained by the existing real-time ICCP algorithm with the optimal sequence length.The intensity of the gravity anomaly in the gravity map is an important factor affecting the performance of the matching algorithm.In the area where gravity anomaly changes dramatically,the matching algorithm can achieve a good positioning performance.It is an effective method to improve the positioning accuracy of matching algorithm that the submarine pass through the high navigation suitable area as far as possible.This paper establishes a gravity map navigation suitable analysis model based on the projection pursuit model and heuristic optimization algorithm.In this paper,the gravity map is transformed into a grayscale map,and the global and local data features are extracted and analyzed by image texture analysis method,such as gray histogram complexity,Sobel operator and gray-level co-occurrence matrix.The projection pursuit model is applied to get the comprehensive evaluation of the navigation suitable of the gravity maps.According to the solar nebula disk model,a gravitation field algorithm(GFA)is proposed.The gravitation field algorithm is used to optimize the optimal projection direction of projection pursuit model.The optimal projection direction obtained by the gravitation field algorithm is most equally which achieve the purpose of comprehensive evaluation of the data features.The ICCP algorithm is used to verify the evaluation results of each matching area,the best positioning accuracy and matching track are obtained in the best evaluation area.The positioning accuracy of gravity aided navigation is closely related to the navigation direction of submarine in the matching area,that is,the matching positioning accuracy are differences when sailing in different directions.It may have a better navigation suitable direction in the matching area with ordinary navigation suitable.In this paper,the directional navigation suitable analysis of gravity map is studied from the perspective of image classification.The 5 scales and 8 typical directions directional navigation suitable feature maps of gravity map are obtained by Log-Gabor filter banks.On the basis of selective ensemble learning based on binary gravitation field algorithm and the optimized voting method,the gravity maps directional navigation suitable analysis model is obtained.The proposed method can effectively avoid the blindness in selecting the directional navigation suitable featrues and setting the directional navigation suitable evaluation thresholds.Compared with random forest and selective ensemble learning based on BPSO,the accuracy of the proposed method is highest.