Underwater Navigation Methods Based On Gravity And Environmental Features

Inertial navigation system (INS) can meet the requirements of autonomous and covert navigation system for submarines and other underwater vehicles over long periods in blue water without exposure, however, the precision of INS is not satisfactory due to its well-known errors which increase with time. As a result, other aiding navigation systems are needed to correct such errors. Due to the characteristics of high precision and no need to surface, the marine geophysical navigation could make up for the aforementioned INS deficiencies in positioning precision, thus is the ideal underwater aiding navigation means for submarines and other underwater vehicles. Therefore, the dissertation will focus on the study of underwater geophysical navigation methods based on gravity and environmental features.Since there are no measured gravity gradient data in our country nowadays, the right rectangular prism method is used to construct the gravity gradient map from marine terrain map, and two gradient maps are generated from actual marine terrain maps to be used as the study and test basis of gradient matching algorithm. Since most of current gravity matching algorithms failed in the event of large INS position error, based on the further study of the principle of probabilistic neural network (PNN), a PNN based gravity gradient matching algorithm is proposed to correct large INS position errors by taking advantages of fine classification ability of PNN. When taking different combinations of gradient components as network input, the influence on the probabilities of successful matchings is also discussed. Simulation results show that the presented gradient matching algorithm can effectively reduce the positioning errors of navigation system without the help of other INS information, and can be used in the event of large positioning errors.From the viewpoint of gravimeter measure principle, the truth that large positioning error will lead to quite large observation error for matching navigation system during the normal gravity computation process is further analyzed, and a strategy of changing both the gravimeter measurement and gravity anomaly into gravity for matching process is carried out to eliminate such observation error. By taking advantage of short-term high precision characteristics of INS, a contour matching algorithm adaptive to large positioning error is proposed, which takes the trajectory most similar to INS indicated one as the matching trajectory. Simulations performed on synthetic gravity map and actual gravity map show that, the contour matching algorithm can achieve a pretty high precision even under operating sea condition and with a moderate-grade INS, thereby adapting to underwater correction of large INS positioning error.Furthermore, to realize the correction of velocity and attitude errors for INS, a gravity/INS integrated navigation system is constructed on the basis of position reset information provided by gravity matching navigation system. The integrated system takes INS error model as the state equation and the positional difference between matching algorithm and INS as observation, and achieves a precise estimation of position, velocity and attitude errors of INS by the use of Kalman filter. The relationship between INS position, heading errors and gyro drift is analyzed, and the triple location correction method is applied to estimate the constant drifts of gyro. The feasibility and validity for estimating gyro drifts are proved via simulation experiments using the results of gravity aided navigation.When underwater vehicles cruise in some area without any gravity data or with a featureless gravity field, gravity aided navigation system will not work. At this time, the simultaneous localization and mapping (SLAM) algorithm based on marine environmental features can substitute for gravity aided navigation system to reduce the increasing rate of navigation errors. The basic principle of SLAM is studied, and the systemic prediction, observation and state augmentation process are discussed according to the nonlinear model of underwater vehicles, and the algorithm is implemented with Extended Kalman Filter techniques. The simulated experiment results acquired in straight sailing pattern and local searching pattern prove that, EKF based SLAM algorithm can efficiently slow down the error increase and improve the positioning precision for submarines and other underwater vehicles. Since current data association methods have some deficiencies in computational complexity and association stability, a modified data association method is put forward. Simulation experiment results show that the presented method could achieve a satisfactory association result with a small computational complexity, thus suitable to on-line data association for underwater SLAM implementations.