Research On The Method Of Maritime Route Network Construction Based On Ship Trajectory Data

With the increasing progress of maritime traffic supervision and technologies,the availability,timeliness,and accuracy of various maritime traffic data have been improved,laying a good foundation for the analysis,application and knowledge discovery of maritime traffic data.Using data mining methods to study maritime transportation has gradually become an important approach in the field of maritime transportation.The maritime route network is an abstract network structure formed by massive ship trajectories,which is of great significance for applications such as ship trajectory prediction,ship abnormal behavior detection and ship route planning.However,at present,the research on maritime route network in the field of maritime transportation is relatively lagging behind,especially the related research on the construction method of maritime route network is less.To extract new methods are urgently needed to construct maritime route networks.Ship trajectory data,as the most important maritime traffic data,contains a large number of rules formed by ships sailing at sea.Therefore,it is a feasible method to mine the relevant information of the marine route network and construct the marine route network through the ship trajectory data.This paper mainly focuses on the construction method of maritime route network,and proposes the extraction and connection method of route network nodes based on ship behavior characteristics,constructing a multi-level route network model and proposed related applications based on the multi-level route network model.The main research contents and results of this paper are as follows:(1)Construction of route network node extraction method.In order to extract the basic element of building the route network—the route network node,this paper adopts the method of feature modeling,and combines the density clustering and the convex hull algorithm to extract the route network node based on the ship trajectory data,Firstly,different ship behavior characteristics are constructed to identify various trajectory feature points in the ship trajectory data;then the feature points are clustered by combining the density clustering method;finally,the convex hull algorithm is used to extract the convex polygons of each cluster to form the waypoint area,and calculate the geometric center of each waypoint area as a topological node.(2)Construction of route network node connection method.In order to correctly connect the nodes of the route network to form the route network,this paper constructs the connection matrix of the route network nodes through the spatial calculation method,and combines the trajectory clustering and contour extraction methods to generate the routes connecting the waypoint areas.Firstly,the spatial topological relationship between the ship’s trajectory points and the waypoint area is calculated,and the ship’s trajectory sequence is converted into a waypoint area sequence to extract the adjacent waypoint area numbers,to construct the route network node connection matrix;then the sub-sections between the connected waypoint areas are extracted The sub-trajectories are divided into different clusters by trajectory clustering;finally,the boundary lines of each sub-trajectory cluster are extracted,and the polygonal areas contained in the boundary lines are used as routes to connect adjacent waypoint areas.(3)Construction of multi-level route network model.In order to realize the multilevel representation of the route network,this paper proposes a method to construct a multi-level route network model based on the connection matrix.Firstly,the topological route network model is constructed by connecting topological nodes according to the connection matrix;then the geometric route network model is constructed by connecting the waypoint areas by routes;finally,define and identify the key nodes of the route network and construct the sub-network model inside the key nodes.(4)Research on the application of multi-level route network model.In order to verify the high availability of the multi-level route network model constructed in this paper,this paper proposes and validates several application scenarios based on the multi-level route network.Firstly,the ship path planning is carried out based on the topological route network model;then the ship yaw detection is carried out based on the geometrical route network model;finally,the ship intersection anomaly detection is carried out based on the key node sub-network model.The research results of this paper can effectively mine the information of the marine route network from the ship trajectory data,and have strong guiding significance and application value for the research on the construction method and application method of the marine route network.