Research On Ship Adaptive Trajectory Prediction And Application Based On GPR Model

With the implementation of the national “One Belt,One Road” strategy,the Yangtze River shipping conditions have been greatly improved.The development trend of rapid,specialized and large-scale operation of ships has become more apparent,and the export-oriented economy has developed more rapidly.However,the increase in the number of ships,especially large ships,will occupy larger navigable waters,affect the customary routes of other ships,and increase the complexity of the navigation environment of the waters.At the same time,drinking water of many provinces and cities directly comes from the Yangtze River.Once the collision between ships happens,the environmental pollution and economic losses cannot be estimated.In order to improve the navigation safety of the Yangtze River and ensure the healthy development of the Yangtze River shipping industry,this paper uses the AIS data of the Wuhan section of the Yangtze River Channel to conduct ship trajectory prediction and application research,aiming to provide ship-related information services for the port,maritime,search and rescue departments monitor and assess the dangers at sea to ensure the safety of navigation and operations.Firstly,it summarizes the theoretical basis of AlS related including AIS data types,and has errors and loss problems for ship AIS data.After performing error data cleaning and recognition operations,cubic spline interpolation is used to interpolate and restore missing data.Through the collection,analysis and processing of AIS data of Wuhan section channel,a ship trajectory database which can be quickly queried is established.Secondly,based on the comprehensive distance between ship trajectories,a density-based ship adaptive trajectory clustering model is established by comparing various trajectory similarity metrics and clustering methods.The model is applied to the pre-built trajectory library and different categories are obtained.The trajectory data set is used to serve subsequent ship trajectory prediction studies.Among them,the method of compressing data by Douglas-Peucker algorithm is proposed.Through multiple compression experiments,the compression thresholds which can retain more trajectory feature points are selected.The effectiveness of the proposed method is verified by the actual ship data in Wuhan waters.Then,considering the large amount of ship trajectory data,the current trajectory prediction algorithm is analyzed and improved,and the ship adaptive trajectory prediction model based on GPR and KRR is established.According to the basic principle of machine learning,the latitude,longitude,heading and speed of clustered trajectory set are used as the eigenvalues to train the model.The ship AIS data of the Wuhan section of the Yangtze River channel is taken as an example.The results show that the GPR-based prediction model predicts the ship's navigation behavior quickly and accurately within acceptable limits.Finally,a ship trajectory prediction application system based on AIS information is designed,including ship navigation trajectory query analysis,regional ship dynamic monitoring and ship future position prediction.On the one hand,through the excavation of the navigation trajectory of massive ships,the query and analysis of the ship's historical navigation data is realized,and the motion law of the ship is found.On the other hand,the GPR-based trajectory prediction model and the electronic chart are combined to realize the future position of the ship the forecast and custom monitoring of ships in specific areas.The research and design of ship adaptive trajectory prediction has created conditions for predicting the ship's next-time state,and has important reference value for the construction of maritime intelligent supervision system and ship navigation safety protection research.