Return to Session D3 Next Abstract

Session D3: Marine Vehicle Navigation

A Combinational Underwater Aided Navigation Algorithm Based on ICCP and Shannon Entropy Based Adaptive Fusion Particle Filter
Chao Liu, Wei Gao, Jiachong Chang, Ya Zhang, Harbin Institute of Technology, China
Location: Galleria I/II
Alternate Number 1

The Inertial Navigation System(INS) is widely used in underwater navigation because of its completely autonomous working mode and the advantage of being immune to weather change and geography limit. However, the error of the INS diverges with time, cannot meet the navigation requirements of the underwater carrier for long-term navigation, and compensation measures must be taken to suppress the divergence of navigation errors. The gravity assisted inertial navigation system can solve the problems due to their passivity and stability of gravity information. The gravity assisted inertial navigation system consists of four modules: INS, gravity measurement system, gravity reference information database and matching algorithm. The accuracy of the inertial navigation system, the accuracy of the background image of the ocean gravity field and the accuracy of the gravity measurement system are determined by the accuracy of the device, and the improvement of the matching positioning algorithm can improve the navigation positioning accuracy from the system level.
The matching positioning algorithm determines the best matching sequence or matching point by comprehensively analyzing the information provided by the inertial navigation system, the gravity real-time measuring system and the gravity field background image, so as to obtain the estimation of the carrier navigation information. The most common matching algorithms are the maximum correlation algorithm and Iterated Closest Contour Point (ICCP) algorithm, Sandia Inertia Terrain-aided Navigation (SITAN) algorithm with recursive filtering. The maximum correlation algorithm does not have excessive requirements on the initial trajectory accuracy, but it needs to be performed after the acquisition of the measurement sequence, so it is difficult to meet the real-time requirements of the navigation system; in addition, the matching result is greatly affected by the initial data error . The ICCP algorithm requires high accuracy of the indicated trajectory and high accuracy of the gravity abnormality measurement information. However, these two conditions are difficult to realize in practical application. In addition, the limitation of the ICCP algorithm is that the rigid transformation ignores the error variation of the indicated trajectory in a short time, and the limitation of the number of matching points is likely to cause a matching error. Due to the nonlinearity of the measurement equation, the SITAN algorithm uses the random linearization technique to obtain the gravity change rate in real time. The SITAN algorithm has good real-time performance, good working performance and allows for large maneuverability. However, the algorithm needs to have a more accurate initial error, which is sensitive to linearization of nonlinear observation models, and low linearization accuracy will lead to filter divergence.
In view of the problems of the above methods, this paper introduces the concept of Shannon entropy and proposes an ICCP algorithm based on Shannon entropy-based adaptive fusion particle filter. The value of information entropy is the amount of information that characterizes the probability system. The entropy value can better reflect the variation of the gravity field in the latitude and longitude direction. The larger the entropy value, the more balanced the gravitational field changes in both directions. The gravity map matching method established by local entropy has good ability to resist geometric distortion. In addition, since the local gravity map contour value has little effect on entropy. And the normalization processing can smooth the noise, so the local entropy is not sensitive to noise. The algorithm firstly obtains the estimated value of the real position of the carrier in real time by the particle filter estimation method with a large initial position error. Then the local gravity difference probability and the local gravity difference entropy are calculated by using the underwater gravity difference value. The true position of the carrier is estimated according to the minimum variance of the underwater gravity difference entropy of the specific area of a certain sea area and the actual measured seabed gravity difference entropy. The point set is used to indicate the trajectory transmission to the ICCP algorithm with a sliding window. The ICCP algorithm makes the indicated trajectory continuously approach the projection point on the gravity anomaly contour by rigid transformation and iterative loop until the condition satisfies the iteration termination. This combination mode can effectively suppress the ICCP algorithm matching failure. The simulation results show that the algorithm has the advantages of good robustness, small matching error and no linear transformation compared with the traditional matching method, and it is less affected by the initial error. The estimation of the position in the ICCP algorithm enables high-precision navigation information to be obtained. Therefore, this algorithm can meet the navigation requirements of underwater carriers for long-term navigation.



Return to Session D3 Next Abstract