Analysis And Research Of Spatio-temporal Trajectory Clustering Based On Improved DBSCAN

The spatio-temporal trajectory is a set of sequences of moving objects about space and time.spatio-temporal trajectory is a very important and valuable data source for studying the characteristics of moving objects.Due to the rapid development of data acquisition technology,it is more and more convenient to collect spatio-temporal data.The clustering research on spatio-temporal trajectory plays an irreplaceable role in traffic management,logistics field,emergency evacuation management,human behavior analysis and geometric analysis.The study about spatio-temporal trajectory has a high theoretical line and feasibility.In this paper,we proposed a trajectory clustering algorithm based on the density-based clustering algorithm DBSCAN.We draw on the idea of segmentation and grouping,taking full advantage of the attributes about speed and direction in the trajectory data.Then the clustering of trajectory segments was simplified to the clustering of vector points.The originally complex measure of trajectory similarity was abstracted as a measure of similarity between vector points.Then we extracted the centroid vector of each cluster and linked them.In our research,we used the flight data of civilian planes after data cleaning,trajectory simplification as the experimental data.We selected six months flight data of one flight for clustering experiment.The results showed that this method could well achieve the effect of spatio-temporal trajectory clustering and laid a good foundation for future research of flight forecasting or anomaly detection.