Research On Iterative Smoothing AUV Multi-source Navigation Based On Factor Graph

Autonomous underwater vehicle(AUV),as a kind of highly intelligent ocean equipment,plays a great role in both the search and rescue of civil resources and the anti-submarine and detective in the military field.In the development trend of the AUV,long-endurance is one of the important research directions in the future.In the long endurance AUV task execution process,with the change of navigation area,the availability of the AUV auxiliary navigation system will also change with the change of the environment,to ensure the long endurance AUV navigation and positioning system to ensure the long duration safe navigation and smooth and efficient task execution needs to put forward high requirements.At present,the commonly used integrated navigation of AUV is usually set for a specific environment.With the increase of underwater navigation sensors and the influence of complex environments on navigation sensors,it cannot meet the requirements of long-term navigation accuracy and navigation safety.In this context,this paper takes the long endurance AUV multi-source navigation information fusion method as the research object,takes advantage of the flexible configuration characteristics of the factor graph probability model in the navigation system,and mainly studies the factor graph multi-source fusion algorithm.Finally,an AUV multi-source navigation information fusion algorithm based on factor graph iterative smoothing is proposed.The main research contents include:First of all,the overall structure of the long endurance AUV multi-source navigation system is designed,and the key technologies are analyzed.Then,the inertial navigation solution module of the whole structure is analyzed in detail.Finally,the strapdown inertial navigation update algorithm described in this paper is simulated,which verifies the effectiveness of the algorithm and the short-time high-precision characteristics of inertial navigation,laying a foundation for the iterative smoothing steps of inertial navigation in the following paper.Then,the paper studies the availability analysis and decision algorithm of navigation sensors used by the long-endurance AUV in the navigation and positioning of multiple sea areas(here divided into the shallow sea,deep sea,and polar area).To comply with the inevitable trend of long endurance AUV,on the one hand,taking the AUV power supply as the reference side,the optimal decision-making process between sensors of AUV is designed within the acceptable range of navigation accuracy.On the other hand,with the current Marine environment of AUV as the reference side,the chi-square detection based on residual error is introduced to judge the availability of navigation sensors.When a certain threshold value is exceeded,the problem sensors are eliminated.Finally,the multi-source navigation decision algorithm is derived by synthesizing these two aspects.Then,the navigation scenario of long endurance AUV is constructed and the decision algorithm is simulated and verified.Thirdly,the multi-source navigation algorithm based on the factor graph is emphatically studied,including modeling the factor graph of the common navigation sensor of long-endurance AUV and constructing the navigation topological structure based on INS/BDS/DVL/USBL/MOD multi-sea area factor graph.Then the standard factor graph fusion algorithm is compared with the traditional federal Kalman filter algorithm to analyze the advantages and disadvantages.Finally,the feasibility and effectiveness of the factor graph algorithm in long-endurance AUV are verified by comparing the two algorithms.Finally,an AUV multi-source navigation algorithm based on iterative smoothing of factor graph was designed to increase the navigation accuracy due to the time-varying characteristics of observation noise of sub-sensors and the real-time distribution of measurement information of standard factor graph and global optimization under complex underwater environment.The algorithm firstly adds the belief function of standard factor graph algorithm,to suppress the child navigation sensor measurement noise and time-varying problem,and then to improve the factor graph algorithm with sliding window,ensure the real-time performance,finally according to characteristics of high precision inertial navigation in a short time,within the sliding window on the marginalized nodes for inertial navigation iteration,improve the navigation accuracy.Finally,the improved factor graph algorithm is simulated and verified according to the sensors selected by AUV’s two power decision processes proposed in Chapter 3.