Research On Clustering Method Of Trajectory Segments Based On AIS Data

With the increase in trade volume between countries in the world and the rapid development of the waterway transportation industry,the number of various types of ships is increasing day by day.Various researches on the data mining field of Automatic Identification System(AIS)also show an active trend.Data mining and analysis of AIS data can derive a large number of data characteristics of waterway transportation.At the same time,the accuracy of data mining analysis is closely related to the data quality of the data source.Only by rationally designing data quality evaluation methods to obtain high-quality data can more accurate data characteristics be mined.Based on the theory of data quality evaluation and data mining,this paper uses Python language to build the corresponding algorithm model,adopts a combination of theoretical research and experimental analysis,and focuses on the data quality evaluation method and trajectory clustering method based on AIS data.,The main work is as follows:1)In order to provide better data for subsequent data mining research,and to make up for the lack of a data quality evaluation method for the application characteristics of AIS data.Designed and proposed a ship data quality evaluation method based on data processing.By studying the application scenarios of AIS data,three data quality evaluation indicators of completeness,continuity,and timeliness are established;through data processing and data analysis of AIS data,a data quality evaluation model is constructed,and a large amount of data is analyzed to find data problems for follow-up The trajectory clustering research provides high-quality data.2)In order to verify the accuracy of the proposed ship data quality evaluation method based on data processing,the data quality evaluation model is used to process satellite AIS data and shore-based AIS data separately,and the comprehensive quality scores of these two types of data and the data visualization results are verified by analysis and comparison.The accuracy of the data quality evaluation method proposed in this article.3)In order to solve the shortcomings of the traditional trajectory clustering model's trajectory similarity measurement algorithm and the typical extraction trajectory algorithm.A clustering model of ship trajectory segments based on Curve Length(CL)distance is designed and proposed.By searching for characteristic points of heading change rate and speed change rate,the trajectory segment is compressed and divided;the defects of current common trajectory similarity measurement are studied and the traditional similarity measurement algorithm between trajectories is improved,which not only reflects the internal connectivity of the trajectory similarity measurement,It also improves the computational efficiency;builds a clustering model of trajectory segments and completes cluster analysis;improves the traditional typical trajectory extraction algorithm,so that the typical trajectory contains more complete features in the cluster.4)In order to verify the accuracy and superiority of the proposed clustering method for ship trajectory segments based on curve length distance.Taking a bifurcated route at the mouth of the Yangtze River as the experimental area,the accuracy of the algorithm is verified by comparing the specified channel of the electronic chart with the clustering results of the algorithm,and the clustering effect and execution efficiency of the algorithm and the current commonly used algorithms are compared to Verify its superiority.