Research On Storage Optimization And Fishing Behavior Identification Technology Of Vessel Trajectory

The safety of fishery production has a bearing on the vital interests of most fishermen,which has become a concern for the government.And the Vessel Monitoring System(VMS)combines the technology of sensor,Internet of Things,satellite positioning and GIS,and has become a vessel intelligent service system which is able to provide real-time monitoring and management services for fisheries sector.With the acceleration of China's marine information construction,a large number of fishing vessels have been incorporated into VMS.The storage,retrieval and analysis of the massive,high-concurrency,continuous and spatio-temporal trajectory data generated by those vessels has become an enormous challenges.Therefore,this paper mainly focus on the storage,query,and analysis techniques for the trajectory of fishing vessels.For the big storage redundancy and low query efficiency of the massive trajectory data,this paper first establishes a moving objects spatio-temporal trajectory model(MOST)and proposes the update strategy for the high-concurrency trajectory data,based on the characteristics of spatio-temporal trajectory data.And this strategy updates the trajectory data by setting the speed and direction thresholds.Only the positioning points that exceed the threshold will be updated into the database,which can dramatically reduce the amount of records in database.Then combined with the storage characteristics of Mongo DB,we design the storage model for the trajectory of moving objects.This storage model can organize the trajectory data of the same fishing vessel into one document for storage,thereby enhancing the locality of storage and improving query efficiency.For the big cost of spatio-temporal query,this paper first considers the distributive characteristics of trajectory data,and designs a spatio-temporal trajectory index based on Geo Hash algorithm.This index divides the map into many grids,and each cell grid has a globally unique code corresponding to it.In addition,the codes of adjacent grids are also adjacent in lexicographical order.Then,we implement this index strategy on the distributed Mongo DB database,and propose the spatio-temporal query algorithm for trajectory.The experimental results show that the proposed method can effectively reduce data redundancy and improve the efficiency of spatio-temporal query.For the problem that VMS data basically consists of sequentially recorded positions and do not directly indicate whether a vessel is fishing or not,this paper propose an automatically learning and discovering human fishing behaviors scheme based on multi-step clustering algorithm(MSC-FBI).First,a temporal-spatial distance model is established;then,an improved multi-step clustering algorithm is used to identify the behaviors of the fishing vessels;and finally,many experiments on different fishing trajectory data were implemented compared with a traditional identification method based on the Gaussian Mixture Model(GMM-FBI).The experimental results illustrate the proposed model's superior performance.For the inefficient recognition of fishing vessel behaviors,this paper builds a recognition model of fishing behavior on the basis of multi-step clustering through the Fisher Discriminant Analysis method,which can automatically learn and extract the patterns of different fishing behaviors from the results of MSC-FBI clustering and then achieve online discrimination of fishing behaviors.The experimental results show that the recognition model can accurately identify various behaviors of vessel based on fishing trajectory.