Adaptive Sampling Methods Of Underwater Vehicles Based On Ocean Forecasting

The modeling and forecasting of ocean environment is important for the research of marine science,and the high-precision and high-resolution environmental parameters also play a very important role in the research of underwater acoustic propagation,target detection,underwater acoustic communication,etc.Since the ocean environment is complex and changeable,and the ocean area is vast,it has become one of the key and difficult issues to efficiently collect useful spatio-temporal samples of ocean environmental parameters to improve the accuracy of environmental forecasting in the field of ocean environmental monitoring depending on the limited observation resources.Observation data of ocean environment can be obtained by fixed nodes,such as buoys,or mobile nodes with sensors.Compared to the high cost of deploying fixed nodes in a large area in the ocean,mobile nodes have the advantages of flexible observation and wide coverage.Therefore,in recent years,adaptive sampling techniques based on underwater vehicles such as autonomous underwater vehicles and underwater gliders have gradually become the focus of research.The adaptive sampling problem of underwater mobile nodes can be modeled mathematically as a constrained optimization path planning problem.The solution of the problem can optimize the best observation path based on the given observation task and limited platform resources on the basis of the environmental forecasting field,so as to achieve the best use of the observation resources under the relevant criteria and reduce the error of the environmental forecasting field.Based on the prediction output of the ocean dynamic model,this paper studies the path planning methods for adaptive sampling of underwater mobile nodes so as to improve the accuracy of the posterior estimation of the environmental field.The details are as follows.The optimization goal of adaptive sampling is proposed based on the temperature field output by the ocean dynamic model.Then the path planning methods based on ant colony algorithm and genetic algorithm are designed according to the optimization goal and the movement characteristics of underwater vehicles.To verify the effectiveness of the path planning algorithm,the high-resolution multisource multistage spectral interpolation method is applied to assimilate the local high-resolution sampling data of mobile nodes and the low-resolution temperature field of the ocean dynamic model.To solve the problem of adaptive sampling of vertical and horizontal temperature profiles,the data of the environmental forecast field in the South China Sea is deployed.The simulation verifications of two-dimensional path planning methods are on the basis of the ant colony algorithm and the genetic algorithm.The adaptive sampling effect of the two methods is analyzed and compared according to the correlation between the sampling data and temperature gradient,the error between the sampling data and the real data,the high-resolution data assimilation results,etc.The results show that the ant colony algorithm has better adaptive sampling performance under the current ocean dynamic model.Through the three-dimensional high-resolution interpolation method,the three-dimensional adaptive sampling effect of the ant colony algorithm is analyzed and verified.In addition,this paper analyzes the effect of sampling data at different temperature gradient resolutions on the error of high-resolution interpolation results,and proposes the optimal gradient resolution in the modeling environment.Considering the time-varying feature of the ocean environment while the underwater mobile nodes sailing during adaptive sampling,a four-dimensional adaptive path planning algorithm based on ant colony algorithm is proposed,and the time-varying ocean forecasting environmental field is used for simulation verification.The simulation uses three underwater vehicles and each of them plans optimal paths.The three-dimensional high-resolution interpolation method is applied to perform data assimilation on the high-resolution sampling data and the low-resolution temperature field of the ocean dynamic model in different time periods.The results of data assimilation show the effect of the adaptive sampling path on the accuracy of the environmental field,and the effectiveness of the fourdimensional path planning algorithm is verified.