| With the rapid development of mobile Internet technology and information technology,a large number of spatio-temporal trajectories of mobile objects have been produced.In these trajectories it hides its rich movement and characteristics.The data of mobile objects' temporal and spatial trajectories are mined to get the behavior characteristics of mobile objects,and then we can fully exploit and utilize the infinite value of these mobile objects' space-time trajectories.At present,the research of spatio-temporal trajectories of mobile objects has been paid much attention by scholars both at home and abroad.Prediction of moving object trajectories both have an important significance in the fields of scientific research and application in real life,for example,through the research of mobile trajectory data of the car,you can find the congested road,re planning of section of city,city traffic calming,alleviate road congestion problems;research by moving track of hurricanes and other natural disasters,mining the movement pattern,warning information to the people,make people do prevention work in the disaster comes,the losses caused by natural disasters to a minimum.By studying the route of bird migration and predicting the migration trajectory of birds,we can better protect birds,and provide more opportunities for bird researchers and enthusiasts to study and observe birds.According to the moving objects trajectories for sparse sampling points and sampling frequency instability of prediction algorithm and its application,this paper is through the actual application needs,to carry out research,through life trajectory data mining and bird migration,similar to that of spatio-temporal trajectory data sparse sampling points,complete this kind of moving object trajectory prediction.The main work includes the following aspects:1.Data preprocessing removes long-term stopover interference points that seriously hinder trajectory research according to the characteristics of migratory bird trajectory points,and proposes a Deleted Interference Points(DIP)algorithm.2.For a class of spatio-temporal trajectory point data with mobility,concentration,and arbitrary shape changes,an important position extraction algorithm for multi-level clustering combined with DBSCAN algorithm and STING algorithm is proposed.The number of clusters cannot be determined before clustering such spatio-temporal trajectory points.Density and Grid based clustering algorithm do not need to set the number of clusters in advance.The classical density-based DBSCAN algorithm meets the clustering requirements of this paper,but the running time of DBSCAN is O(NlogN).So it does not apply to higher-dimensional data.DBSCAN clustered bands do not show the trend of the trajectory in the clustering results.Therefore,choosing a clustering algorithm based on density and grid is the best choice for this paper.The STING grid algorithm is a clustering algorithm that uses a grid to divide downwards step by step.It can no longer form more accurate dense area borders in a short period of time.However,in the clustering process,density connectivity areas can be extracted and connected areas can be obtained.The center point.Therefore,a clustering algorithm based on the combination of DBSCAN and STING can effectively make the trajectory point data of the paper communicated with each other,find the coordinates of the center point of the liaison area,and connect multiple links in time to obtain multiple trajectories.3.In order to explore the influence of weather characteristics on the spatio-temporal trajectories of moving objects,the relationship between temperature and time-space arrival trajectory points of moving objects was studied.Find a temperature index.Furthermore,taking temperature as the influence factor of hidden state transition probability in Hidden Markov process,a temperature-influencing factor-based Hidden Markov Model trajectory prediction algorithm,referred to as THMM trajectory prediction algorithm,is proposed. |
PDF Full Text Request