Real Time Clustering Method And System Implementation Of Ship Trajectory Based On AGA-TraClus

With the advent of the era of maritime information and big data and the establishment of automatic identification system（AIS）network and trajectory database,a large number of AIS ship historical trajectories can provide data support for exploring intelligent supervision algorithms,so as to identify ship motion patterns,so as to improve the supervision efficiency of the maritime department.However,AIS cannot discover the ship navigation features in real time and cannot solve the shortcomings of massive trajectory data clustering.In the thesis,a trajectory clustering algorithm based on AGA-Tra Clus is proposed.Combined with Storm distributed real-time computing platform for large-scale data clustering analysis,a real-time clustering system for ship track based on AGA-Tra Clus is designed and implemented.Firstly,in order to solve the sensitive problem of parameter selection in clustering algorithm,this thesis proposes a trajectory clustering algorithm based on AGA-Tra Clus combined with adaptive genetic optimization algorithm.The advantages,basic principles and implementation process of the algorithm are explored in depth.The algorithm can adapt to the parameter selection,so as to reduce the workload and realize the clustering effect better.Then the advantages and disadvantages of distributed cloud computing platform are compared and analyzed.Storm is selected as the real-time processing framework of big data,combined with Kafka message queue,build a powerful clustering environment,effectively avoid the system crash caused by high concurrency of data,so as to realize the item by item reading of big data.By comparing the data latency,throughput capacity and memory usage of Storm cluster with Java stand-alone environment,it is proved that Storm cluster is in good running condition and can process big data flow efficiently in real time.Finally,a ship trajectory clustering analysis system is developed and implemented by combining algorithm,framework platform and Web end.The AIS data of a cargo ship in Bohai Sea area for one month is taken as an example for experiment.The experimental results show that this system can realize real-time and efficient clustering analysis of ship trajectory data,and the clustering results are good and consistent with reality,which has certain research significance for future maritime traffic planning and maritime supervision.