Behavioral-Pattern Mining And Visualization Analysis For Different Dimensional Trajectory Data

The application of GNSS equipment,sensors,wireless communication and other technologies enables efficient acquisition of real-time trajectory data of mobile objects.These data contain attribute information of different spatial scales,different time scales and different degrees of complexity.Effectively process,mine and visualize it,and explore and analyze the inherent laws of moving objects.Currently,when mining and visualizing trajectory data behavior patterns,the following problems exist:(1)In low-dimensional space,a single trajectory is used to express multiple attributes,a trajectory set description attribute is single,a trajectory set attribute is over-drawn,and a multi-link view is visualized;(2)In high-dimensional space,the data projection algorithm is difficult to obtain information from high-dimensional space and associate it with a visually low-dimensional space.In the face of high-dimensional spatial data mining and visual modeling with complex and dynamic features,the theoretical basis and efficient algorithms need to be studied in more depth and extensively.To this end,this paper analyzes and studies the visualization theory and method of low-dimensional trajectory data,and realizes the information mining algorithm and visualization model of high-dimensional trajectory data related to mine car trajectory.Firstly,the widely used clustering algorithm,point-graphing algorithm and stacking algorithm are studied,and the shortcomings of mining and visualizing low-dimensional mining-truck trajectory data are found.The anomaly trajectory detection method based on edit distance and hierarchical clustering is improved.The trajectory is segmented by similarity measure and historical marks,and the computation method based on the position and number of stay points of point stack is redesigned.Based on this,the characteristics of the similarity trajectory are detected,the behavior pattern of the abnormal trajectory is evaluated,the geographic point map of the staying point is drawn,and the mining model of the semantic trajectory is defined.Based on the stacking trajectory,the moving object-based trajectory is established.The regression elevation model implements multi-attribute correlation analysis based on static visualization and dynamic visualization.Secondly,in terms of the high-dimensional trajectory data,the interactive trajectory star coordinate i-tStar and the interactive three-dimensional trajectory star coordinate i-tStar(3D)are designed.The effectiveness of the method for processing high-dimensional trajectory data is verified by visualizing actual mining transportation data,and i-tStar(3D)has higher degree of freedom than i-tStar.Reviewing the full text,there are the following innovations:(1)Designed a 3D point stack form to draw the geographic point map of the stay point to better display the granularity and local features of the data;(2)Applying the stacking technology,an elevation regression model suitable for the mining vehicle trajectory is established to reflect the elevation characteristics of the stacked trajectory;(3)The i-tStar and i-tStar(3D)models are optimized and applied to the visualization of high-dimensional trajectory data,showing the spatial characteristics of dynamic multidimensional data.