Research On Indexing And Query Algorithm Of Fishing Vessel Navigation Big Data For Trajectory Analysis

Big data of fishing vessel trajectory is a valuable resource to promote the development of modern marine economy,and the effective use of the corresponding data can be of great significance to improve the management level of China’s fisheries production and realize the modernization of fisheries.The use of similarity metric analysis of fishing vessel trajectories for route conflict collision warning and fishing intensity prediction is important for scientific and systematic management of fishery resources and production.However,unlike the common road traffic trajectory data with road network constraints,the fishing vessel navigation trajectory data has a series of problems such as high density of regional data points,random distribution of data points,irregular shape and non-uniform number of trajectory points of single trajectory data,which makes the traditional spatio-temporal trajectory indexing method unable to index and search the similarity of fishing vessel trajectory data well.To solve the above problems,the following work is conducted in this dissertation.(1)The similarity search algorithm of trajectory data based on Geohash and grid structure is proposed: Based on the characteristics of fishing boat trajectory data different from other traffic trajectory data,the similarity search algorithm of fishing boat trajectory based on Geohash and dynamic grid is proposed after the problems of slow index construction and query speed decrease caused by using the query algorithm built by traditional similarity index.The variable dynamic grid structure constructed by Geohash is used to query trajectories not by points but by the range of grid paths constructed by the uniqueness of Geohash,the other trajectory points contained in the grid path are counted directly,and the time dimension is added to ensure that the query trajectory,the queried trajectory are in the same time dimension;the number of trajectory points of a trajectory in the count result is compared with the total number of trajectory points of the trajectory.The ratio between the number of trajectory points of a trajectory and the total number of trajectory points of that trajectory is calculated,whether the trajectories are similar or not is determined by comparing the calculated results with the threshold value set by users.The experimental results using real fishing boat trajectory data show that the proposed method improves 52.33% in index construction speed and about 18.24% in range time query speed compared with the comparison method,which effectively improves the speed of obtaining similar fishing boat trajectory data and provides effective data support for further data analysis.(2)Proposed distributed index structure based on data community aggregation and columnar storage structure: In order to solve the problems such as long computing time of the algorithm in single-computer mode,difficulty of vertical expansion of single-computer,and slow search speed of sequence group similarity,a distributed index structure based on data community aggregation and columnar storage structure is proposed.For the data voids that may cause resource wastage in the composed data layer,the index is constructed by using the community aggregation possessed by the fishing boat track data to reduce or ignore the invalid data grid,and the invalid queries in data search are further reduced by using the columnar storage to achieve the purpose of reducing the search time.The experiments on the actual fishing boat track data show that the number of subqueries generated by the query is reduced by about 15.9% compared with the comparison algorithm,the response time of the algorithm is reduced by about 15.6%,the response time of the algorithm can be maintained more stable under the expansion of the data search range,and the problem of the proliferation of subqueries caused by the special data aggregation characteristics of indexed data is solved more effectively.