Adaptive Tracking Methods Of Marine Environment Based On AUV Platform

Marine environmental observation is the basis for studying the ocean,developing the ocean,and using the ocean.The detection,analysis and tracking of marine environmental characteristics have always been the key research objects of ocean observation.The medium and small-scale marine environment,which is volatile in time and space,is susceptible to interference from the external environment,which poses a major challenge for forecasting and tracking marine features.The observation data of the marine environment can be observed by scientific research ships,buoys and submersible targets.However,the measurement range of these methods is limited and the autonomy is poor,which makes the long-term marine scientific examinations costly and inefficient.In contrast,with the development of underwater unmanned systems,small and flexible underwater robots can quickly sample a designated area and obtain high-resolution environmental information of the target area,and efficiently collect key spatiotemporal samples of marine environmental parameters.But underwater mobile nodes still face the contradiction between limited observation resources and coverage for a long time.Therefore,we hope that the underwater mobile observation platform can autonomously track and detect the feature area of interest with limited resources and unmanned intervention.Through real-time processing of the environmental data that are sampled by itself,which changes with space and time,the underwater mobile observation platform adaptively adjusts the sampling path according to the target feature distribution to achieve autonomous sampling of key space-time information.Fronts and thermoclines have a significant impact on the propagation of sound waves in the ocean,and the study of these two ocean features is of great significance in marine science,hydroacoustic engineering,and military applications.Aiming at the detection,tracking and forecasting requirements of small and medium-scale ocean features,this paper selects two representative ocean features,temperature front and thermocline,as the targets of adaptive sampling and tracking,and focuses on relevant real-time feature tracking algorithms,mainly including the following work content.To solve the problem of rapid environmental changes and complex morphology in frontal real-time tracking,the front search strategy and constraint conditions of the two-dimensional front tracking algorithm based on Gaussian Process Regression are improved.In the static and dynamic frontal environment fields,the simulation comparison of the traditional zigzag front tracking algorithm,the front tracking algorithm based on Linear Least Square Fitting method,and the front tracking algorithm based on Gaussian Process Regression is carried out respectively.Using multiple performance indicators to evaluate the tracking performance of each algorithm,the results show that the stability and adaptive tracking ability of the front tracking algorithm based on Gaussian Process Regression are better.In addition,considering the characteristics of the front in the vertical direction,the Vertical Temperature Homogeneity Index is introduced,and the two-dimensional algorithm is extended to the three-dimensional front tracking.In view of the high computational complexity of existing thermocline tracking algorithms,a thermocline tracking algorithm based on tracking isotherms is designed.The feasibility of the algorithm is verified by sea trial data,and compared with the thermocline tracking algorithm based on the maximum temperature gradient,and the advantages and disadvantages of the related algorithms are analyzed in combination with numerical indicators.In order to improve the limitation of the isotherm tracking algorithm,based on the isotherm-based thermocline tracking algorithm combined with the global path planning algorithm,a local and global adaptive sampling algorithm is proposed.