Study And Implementation Of Key Technologies On Logistics Monitoring System

The goods theft in the process of logistics transportation is a big problem in the logistics industry.The traditional logistics monitoring system can only tracking the trucks by their GPS coordinates,which can not solve the problem of goods theft.The study found that the goods theft is strongly related to the long stay in the logistics process and the choose of anomaly logistics trajectories.Therefore,if we can detect the long time stay points and anomaly trajectories in the process of logistics through data mining,and set up an appropriate logistics monitoring system for these sensitive regions composed of long time stay points and anomaly trajectories,we can solve the problem of the goods theft in the process of logistics transportation effectively.In view of the above scheme,this thesis has studied and implemented two key technologies: anomaly trajectory detection,automatic generation and early warning of geofences.The specific work is as follows:First,this thesis proposes a data preprocessing scheme for GPS coordinates of logistics vehicle: Firstly,the missing GPS coordinates are processed by linear interpolation;And then the long time stay points of logistics vehicle are identified by the speed characteristics,with these long time stay points,the GPS coordinates are processed by data thinning;Finally,using the S-G filter to perform data smoothing on the GPS coordinates.The data preprocessing scheme can effectively improve the results of related algorithms in the following work.Second,this thesis proposes an anomaly trajectory detection algorithm: Firstly,the pre-processed trajectories are cut based on the geometric features.Then,an improved DBSCAN clustering algorithm based on MapReduce is used to cluster the cut trajectory segments.According to the clustering results,the anomaly trajectories are detected in conjunction with the characteristics of space and time.Third,this thesis proposes an algorithm for automatic generation of geofences.Firstly,the detected long time stop points and anomaly trajectories are converted into coordinate points,and then geofences are automatically generated based on the coordinate points.And,this paper also proposes a scheme of geofences` warning function based on the spatial index R-tree,which has low time complexity.Tests have shown that the algorithm and schemes proposed in this thesis can achieve good results under real data sets: They can effectively preprocess the trajectory data,can accurately identify the long time stay points in the logistics process,and can detect the anomaly trajectories in the trajectory data sets.They can automatically generate geo-fencing based on the long time stay points and anomaly trajectories,and provide real-time geo-fencing warning.