Research On Trajectory Clustering And Prediction Method Based On AIS Dat

In order to reduce maritime traffic congestion and the probability of accidents,and promote the intelligent and efficient operation of water traffic management,this thesis focuses on the theme of "Clustering and Prediction of Ship Trajectories",and takes engineering applications in the maritime field as the ultimate goal.With machine learning and deep learning theory as technical support,Python language is used to build models,and research is carried out by combining theoretical research and experimental analysis.The main work is as follows:(1)In the aspect of ship trajectory segment clustering,two improved trajectory similarity measurement algorithms are proposed,and combined with DBSCAN algorithm,two ship trajectory segment clustering models based on comprehensive similarity measurement are established.Then,taking the data of the Yangtze River estuary as the research object,the proposed model is compared with the trajectory segment clustering model based on Hausdorff and Frechet respectively.The lack of timing performance and poor noise immunity improves the clustering effect.For the clustering results,a typical trajectory extraction algorithm based on sector scanning area was successfully used to extract the typical trajectory of the clustering result,and the error was compared with the standard route.The results showed that the typical trajectory under the two models proposed was It has obvious advantages in error and can be used for subsequent engineering applications.(2)In the aspect of ship trajectory prediction,in view of the problems of poor practical application and low accuracy of the existing ship trajectory prediction models,a composite ship trajectory prediction model combined with AMS is proposed.By increasing the complexity,the model uses the BP network to correct the LSTM network error,and the model is initially improved,and the optimization algorithm is explored,and the improved algorithm AMS of the current mainstream ADAM optimization algorithm is successfully applied to the proposed ship trajectory prediction model,which is used to further optimize the model.Finally,for the proposed model,the model is verified based on AIS trajectory data.By comparing with a single LSTM network model in terms of multiple performance parameters and recursive application tests,the results show that the composite ship trajectory prediction model has high precision and high robustness.And other advantages,it can provide reference for the maritime ship traffic management department.