A ship experience route extraction system and method based on AIS historical data

Abstract

The present invention discloses a ship experience route extraction system and method based on AIS historical data, the system includes a semantic route extraction system, a channel boundary extraction system and a hottest route extraction system; the semantic route extraction system is used to obtain semantic routes in a region through AIS track start and end point clustering, feature point extraction and feature point clustering; the channel boundary extraction system is used to extract boundary information for each semantic route; the hottest route extraction system is used to construct a channel triangulation network using a route boundary point set, and extract the hottest route in the channel. Based on AIS historical data, the present invention comprehensively considers the ship route mode and the track density distribution under each route mode, and realizes the extraction of high-precision hottest routes from complex AIS data.

Description

Technical Field

[0001] The present invention relates to the technical field of ship AIS data mining, and in particular to a ship experience route extraction system and method based on AIS historical data. Background Art

[0002] With the development of economy, maritime traffic is getting busier and busier, and the huge ship traffic flow increases the difficulty of maritime traffic management. The extraction of ship experience routes helps to understand the behavior patterns of ships in the region, provide effective information for decision makers, help understand the distribution pattern of ship traffic flow, and find common ports, regional entry and exit points, route turning points and other information. In addition, experience routes can use historical experience to monitor abnormalities, predict behavior, and plan routes for existing ships. Therefore, in maritime traffic management, it is very important to obtain ship experience routes in advance. However, some traditional route extraction methods are not suitable for marine environments. For example, manual measurement is limited by factors such as high cost, slow update speed, poor safety, and susceptibility to environmental factors; high-precision remote sensing imaging technology requires satellites to monitor ship movement data in real time, which is costly and difficult to promote; methods based on image recognition technology have good applications on land, but it is not easy to distinguish between waterways and non-waterway areas at sea; although electronic charts recommend waterways for ships, the navigation of ships in the waters is not restricted by waterways, so the ship's experience routes cannot be accurately represented.

[0003] The application and development of the Automatic Identification System (AIS) provides reliable ship trajectory information. Massive AIS data potentially reveals the behavioral characteristics of ships, ports and other information. In addition, AIS data has the characteristics of low cost, strong real-time performance and large volume, and is an ideal information source for reconstructing ship trajectories and extracting empirical route knowledge. How to extract high-precision empirical routes from complex AIS data has become a hot topic in the current data mining field. Many studies have done a lot of work in this regard and proposed some feasible methods. For example, a Chinese patent with publication number CN112164247A disclosed on January 1, 2021, a ship route prediction method based on ship trajectory clustering, which matches the current route trajectory of a specified ship with the characteristic route trajectory clustered and extracted after extracting the historical route, effectively overcomes the problems of low prediction accuracy and long matching time, and can greatly improve the accuracy of prediction compared to existing solutions. For example, the Chinese patent with publication number CN112906830A, which was disclosed on June 4, 2021, discloses a method for automatically generating optimal ship routes based on AIS big data. The method first pre-processes massive historical AIS data, then introduces ship static data, adds the ship draft to the distance measurement to measure the similarity of the AIS data, and then uses an improved adaptive clustering algorithm to cluster the ship trajectory clusters until the sub-cluster is single. Finally, the main navigation route extraction method is used to extract the main route from the clustered sub-cluster, so as to realize automatic generation of the optimal route for the ship.

[0004] At present, the shipping network extraction technology based on AIS data has made great progress, but how to comprehensively consider the ship route pattern and the trajectory density distribution within the channel has become an urgent problem to be solved. Summary of the invention

[0005] In order to overcome the above-mentioned deficiencies of the prior art, the present invention provides a system and method for extracting ship experience routes based on AIS historical data, and mines ship experience route knowledge from AIS historical data based on an unsupervised learning method.

[0006] According to one aspect of the present invention, there is provided a system for extracting ship empirical routes based on AIS historical data, including a semantic route extraction system, a channel boundary extraction system and a hottest route extraction system; the semantic route extraction system is used to obtain semantic routes within a region by clustering AIS track start and end points, extracting feature points and clustering feature points; the channel boundary extraction system is used to extract boundary information for each semantic route; the hottest route extraction system is used to construct a channel triangulation using a route boundary point set and extract the hottest route within the channel.

[0007] The above technical solution first extracts the route pattern based on the AIS historical data to obtain all the semantic routes in the target area; then extracts the boundary point set of each semantic route, constructs a channel triangulation based on the extracted boundary point set information of each semantic route, and extracts the hottest route in each route pattern in the constructed channel triangulation, so as to achieve the purpose of mining the ship's empirical route knowledge from the AIS historical data and then determining the hottest route; based on the AIS historical data, the technical solution comprehensively considers the ship's route pattern and the trajectory density distribution under each route pattern, so as to achieve the extraction of high-precision hottest routes from complex AIS data, thereby improving the route extraction accuracy.

[0008] As a further technical solution, the semantic route extraction system further comprises: firstly, using the DBSCAN clustering algorithm to cluster the starting point and the ending point of the AIS data respectively; then, using the improved DP compression algorithm to extract the characteristic points of each ship AIS track, including: the starting point, the ending point, and the waypoint; finally, using the DBSCAN algorithm to cluster the characteristic points of all tracks, and the track set with the same characteristic point cluster is defined as a semantic route;

[0009] The improved DP compression algorithm further comprises:

[0010] First, each trajectory point is normalized using the following normalization method:

[0011]

[0012] Then, the first and last vertices of the ship's trajectory are connected by an imaginary straight line, and the synchronous Euclidean distance (SED) from the remaining vertices to the straight line is calculated. The largest one is selected and compared with the specified tolerance. If it is less than or equal to the tolerance, all points between the two ends of the straight line are deleted; if it is greater than the tolerance, the vertex with the largest distance from the straight line is retained, and the curve is divided into two parts based on this. The above method is repeated for these two parts until no further compression can be made.

[0013] P i =(x i ,y i , t i ,S i ) and P j =(x′ j , y′ j , t′ j ,S′ i ) can be calculated by the following formula,

[0014]

[0015]

[0016]

[0017]

[0018] In the above technical solution, the ship AIS track is a collection of historical track points of ships with the same MMSI code, which can be represented by T i =(P1, P2, ..., P n ), where P i =(x i ,y i , t i , S i ), x i ,y i , t i , S i They represent longitude, latitude, time, and speed over the ground respectively.

[0019] As a further technical solution, the waterway boundary extraction system includes a kernel density estimation module and a boundary point extraction module; the kernel density estimation module is used to calculate the kernel density estimation distribution of the trajectory point set of each semantic route; the boundary point extraction module is used to extract the boundary point set of the semantic route according to a set threshold, wherein the threshold is obtained by observing the kernel density estimation distribution of the trajectory point set and combining the actual application requirements for the boundary to select a value suitable as a boundary point, and the threshold is adapted to the kernel density estimation distribution of the trajectory point set;

[0020] Wherein, the kernel density estimation module further includes:

[0021] The trajectory point set of each semantic route is converted into the corresponding point (x1, x2, ..., x n ), and use the following formula to calculate the kernel density estimate distribution:

[0022]

[0023] Among them, K(x) represents the kernel function, h represents the smoothing parameter, and K(x) meets the following conditions:

[0024] K(x)≥0(-∞ <x<+∞)

[0025]

[0026] K(-x)=K(x).

[0027] As a further technical solution, the hottest route extraction system includes a channel triangulation construction module and a hottest route extraction module;

[0028] The channel triangulation construction module is used to construct the channel triangulation of the semantic route using the boundary point set extracted from the semantic route based on the Delaunay triangulation method;

[0029] The hottest route extraction module is used to calculate the centroid coordinates of each triangle in the channel triangulation network, filter the triangles outside the channel, extract the coordinate points corresponding to the maximum value of KDE in each triangle, connect these points in order according to the direction of the semantic route, and use the Savitzky-Golay filter to smooth the obtained path to obtain the most popular route in the semantic route, that is, the hottest route.

[0030] As a further technical solution, a channel triangulation network of the semantic route is constructed based on the Delaunay triangulation method using the boundary point set extracted from the semantic route, which further includes: taking the extracted boundary point set as a finite point set, and the closed line segment connecting two points in the point set as an edge. At the same time, these edges must satisfy the following conditions: there are no intersecting edges; except for the endpoints, the edge does not contain any point in the point set; all faces in the plane are triangles, and the union of all triangles is the convex hull of the finite point set, thereby constructing a channel triangulation network.

[0031] According to another aspect of the present invention, there is provided a method for extracting ship experience routes based on AIS historical data, comprising:

[0032] Get AIS historical data;

[0033] The DBSCAN clustering algorithm is used to cluster the starting points and ending points of the AIS data;

[0034] The improved DP compression algorithm is used to extract the characteristic points of each ship AIS track, including: starting point, end point, and waypoint;

[0035] The DBSCAN algorithm is used to cluster the feature points of all trajectories, and the set of trajectories with the same feature point clusters is defined as a semantic route;

[0036] Calculate the kernel density estimation distribution of the trajectory point set of each semantic route, determine a threshold value suitable for the kernel density estimation distribution of the trajectory point set by observing the kernel density estimation distribution of the trajectory point set and combining the requirements of the actual application for the boundary, and extract the boundary point set of the semantic route according to the determined threshold value;

[0037] Construct a channel triangulation network using the boundary point set of the semantic route;

[0038] Calculate the coordinates of the center of gravity of the triangle, filter out the triangles outside the channel, extract the coordinate points corresponding to the maximum KDE value in each triangle, and obtain the maximum KDE point set;

[0039] The points in the maximum KDE point set are connected in order according to the direction of the semantic route, and the obtained path is smoothed using the Savitzky-Golay filter to extract the hottest route.

[0040] In the above technical solution, the AIS historical data of the target area is first obtained, and the semantic routes with directionality in the target area are obtained through start and end point clustering, feature point extraction and feature point clustering; then the trajectory point set of each semantic route obtained is subjected to kernel density estimation distribution calculation to extract the channel boundary point set; based on the extracted boundary point set, the channel triangulation is constructed to obtain the maximum KDE point set, that is, the maximum kernel density distribution point set, and the hottest routes in the target area are finally extracted by orderly connecting the points in the maximum KDE point set and processing the connected paths. This technical solution is based on AIS historical data, comprehensively considers the ship route mode and the trajectory density distribution in each route mode, and realizes the extraction of ship experience routes through semantic route extraction, channel boundary extraction and hottest route extraction, and realizes the extraction of high-precision hottest routes from complex AIS data.

[0041] As a further technical solution, the feature points of each ship AIS track are extracted using the improved DP compression algorithm, further comprising: connecting the starting and ending points of the ship's track into a line as the initial baseline, then calculating the distance from each point to the baseline, and finding the point with the largest distance; secondly, connecting the starting and ending points with the point to form two baselines, and then calculating the distance from each point in the first segment to the baseline and the distance from each point in the second segment to the baseline, and finally finding the point with the largest distance in the first segment and the point with the largest distance in the second segment; and repeating this cycle until the maximum distance is less than the distance threshold parameter D max When , the loop ends; thus, the feature point set of each ship AIS trajectory is obtained.

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] (1) The present invention provides a system, which first extracts route patterns based on AIS historical data to obtain all semantic routes in a target area; then extracts a boundary point set of each semantic route, constructs a channel triangulation based on the extracted boundary point set information of each semantic route, and extracts the hottest route under each route pattern in the constructed channel triangulation, thereby achieving the purpose of mining ship experience route knowledge from AIS historical data and then determining the hottest route; based on AIS historical data, the system comprehensively considers ship route patterns and trajectory density distribution under each route pattern, thereby achieving high-precision extraction of the hottest route from complex AIS data, thereby improving the route extraction accuracy.

[0044] (2) The present invention provides a method, which first obtains the AIS historical data of the target area, obtains the semantic route with directionality in the target area through start and end point clustering, feature point extraction and feature point clustering; then performs kernel density estimation distribution calculation on the trajectory point set of each semantic route obtained, and extracts the channel boundary point set; based on the extracted boundary point set, constructs a channel triangulation, obtains the maximum KDE point set, that is, the maximum kernel density distribution point set, and finally extracts the hottest route in the target area by orderly connecting the points in the maximum KDE point set and processing the connected paths. Based on the AIS historical data, the method comprehensively considers the ship route mode and the trajectory density distribution under each route mode, realizes the extraction of ship experience routes through semantic route extraction, channel boundary extraction and hottest route extraction, and realizes the extraction of high-precision hottest routes from complex AIS data. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 FIG. 4 is a system block diagram according to an embodiment of the present invention.

[0046] Figure 2 (a)-(d) are structural block diagrams of a semantic route extraction system according to an embodiment of the present invention.

[0047] Figure 3 Schematic diagram of DBSCAN clustering of trajectory starting points according to an embodiment of the present invention.

[0048] Figure 4 A schematic diagram of extracting trajectory feature points using the improved DP algorithm according to an embodiment of the present invention.

[0049] Figure 5 4 is a structural block diagram of a waterway boundary extraction system according to an embodiment of the present invention.

[0050] Figure 6 4 is a structural block diagram of a hottest route extraction system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0051] The following will clearly and completely describe the technical solutions of various embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0052] The present invention provides a ship experience route extraction system and method based on AIS historical data, which comprehensively considers the ship route mode and the trajectory density distribution in the channel, calculates the trajectory density distribution of different routes, and reasonably selects the appropriate boundary threshold to obtain the channel boundary according to the trajectory kernel density estimation distribution; selects the point with the largest trajectory point density on the route and connects it to obtain the hottest route, which can effectively obtain the shipping network model comprehensively.

[0053] Example 1

[0054] This embodiment provides a ship experience route extraction system based on AIS historical data, such as Figure 1 , including semantic route extraction system, channel boundary extraction system and hottest route extraction system.

[0055] The semantic route extraction system obtains the semantic route in the area through trajectory start and end point clustering, feature point extraction and feature point clustering.

[0056] The waterway boundary extraction system extracts boundary information for each semantic route.

[0057] The hottest route extraction system constructs a channel triangulation network using a route boundary point set, and extracts the most popular route path in the channel, namely, the hottest route.

[0058] Furthermore, the semantic route extraction system proposed in this embodiment has the following extraction steps:

[0059] like Figure 2 As shown in (a)-(d), first, the DBSCAN clustering algorithm is used to cluster the starting point and the ending point of the AIS data, then the improved DP compression algorithm is used to extract the intermediate waypoints, and finally the DBSCAN clustering algorithm is used to cluster all the feature points to obtain active waypoints. These active waypoints can represent ports, entrances and exits, offshore platforms, turning points and other areas in the study area. A cluster of trajectories with the same starting waypoint, intermediate waypoint and ending waypoint is defined as a semantic route. Based on the obtained waypoints, the AIS data set can be classified to obtain the semantic route distribution of the sea area or region.

[0060] The DBSCAN clustering algorithm mentioned in this embodiment is a density-based spatial clustering method, which contains two parameters: the distance threshold eps and the trajectory point number threshold min_samples. Taking the trajectory starting point clustering as an example, if there are min_samples other trajectory points with a distance eps from the current trajectory point, we consider the current trajectory point to be the core point, and the other min_samples samples are regarded as the neighbors of this trajectory point. A starting waypoint is composed of a series of core points and included boundary points, while other trajectory points not included in the cluster are considered to be noise, such as Figure 3 As shown in Figure 2, the algorithm divides trajectory points with sufficient density into clusters and can discover active waypoints from a large number of trajectory points.

[0061] The core idea of ​​the improved Douglas-Peucker (DP) algorithm mentioned in this embodiment is to use a set of points w = {w1, w2, ..., w i} to find a point set w′ that represents the overall characteristics, which contains a synchronous Euclidean distance (SED). The specific process is: connect the starting and ending points of the trajectory into a line as the initial baseline, then calculate the distance from each point to the baseline and find the point with the largest distance; then connect the starting and ending points with the point to form two baselines, and then calculate the SED distance from each point in the first segment to the baseline and the SED distance from each point in the second segment to the baseline, and finally find the point with the largest distance in the first segment and the point with the largest distance in the second segment; repeat this cycle until the maximum distance is less than D max The loop ends when . Thus, we get the feature point set w′={w2,w4,w5,w6,w 11},like Figure 4 The system performs DBSCAN clustering on the feature point sets of all trajectories in the same semantic route to find the feature waypoints of the semantic route, which will serve as a reliable input for semantic route decomposition.

[0062] P i =(x i ,y i , t i ,S i ) and P j =(x′ j , y′ j , t′ j ,S′ i ) can be calculated by the following formula

[0063]

[0064]

[0065]

[0066]

[0067] The waterway boundary extraction system proposed in this embodiment is as follows: Figure 5 As shown in the figure, for each semantic route, the kernel density estimation (KDE) distribution of its trajectory point set is calculated, and the boundary point set of the semantic route is extracted by setting a threshold suitable for the trajectory distribution density, which will serve as a reliable input for constructing the channel triangulation network.

[0068] The core of the kernel density estimation (KDE) mentioned in this embodiment is to use a smooth peak function to fit the data points, thereby simulating the real probability distribution. Therefore, when the spatial distribution of the trajectory is unknown, the KDE distribution of the AIS data point set can well describe the spatial distribution characteristics of the trajectory. The calculation process is as follows: First, the trajectory point set of the semantic route is converted into the corresponding points in the grid (x1, x2, ..., x n ), and use the following formula to calculate the kernel density estimate distribution:

[0069]

[0070] Among them, K(x) represents the kernel function, h represents the smoothing parameter, and K(x) meets the following conditions:

[0071] K(x)≥0(-∞ <x<+∞)

[0072]

[0073] K(-x)=K(x).

[0074] The hottest route extraction system proposed in this embodiment is as follows: Figure 6 As shown in the figure, firstly, the channel triangulation of the semantic route is constructed based on the Delaunay triangulation method using the boundary point set extracted from the semantic route, and the triangles outside the channel are filtered by calculating the centroid coordinates of the triangles. Then, the coordinate points corresponding to the maximum value of KDE in each triangle are extracted. Finally, these points are connected in order according to the direction of the semantic route, and the obtained path is smoothed using the Savitzky-Golay filter to obtain the most popular route within the semantic route, that is, the hottest route.

[0075] The Delaunay triangulation method mentioned in this embodiment includes: taking the obtained channel boundary points as a finite point set, and the closed line segments connected by two points in the point set as edges. At the same time, these edges must satisfy: there are no intersecting edges; except for the endpoints, the edges do not contain any points in the point set; all faces in the plane are triangles, and the union of all triangles is the convex hull of the finite point set, thereby constructing a channel triangulation network. The channel triangulation network is unique, and there will be no other trajectory points within the circumscribed circle of any triangle in the Delaunay triangulation network.

[0076] Example 2

[0077] This embodiment provides a method for extracting ship experience routes based on AIS historical data, including:

[0078] Get AIS historical data;

[0079] The DBSCAN clustering algorithm is used to cluster the starting points and ending points of the AIS data;

[0080] The improved DP compression algorithm is used to extract the characteristic points of each ship AIS track, including: starting point, end point, and waypoint;

[0081] The DBSCAN algorithm is used to cluster the feature points of all trajectories, and the set of trajectories with the same feature point clusters is defined as a semantic route;

[0082] Calculate the kernel density estimation distribution of the trajectory point set of each semantic route, determine a threshold value suitable for the kernel density estimation distribution of the trajectory point set by observing the distribution and combining the actual application requirements for the boundary, and extract the boundary point set of the semantic route according to the determined threshold value;

[0083] Construct a channel triangulation network using the boundary point set of the semantic route;

[0084] Calculate the coordinates of the center of gravity of the triangle, filter out the triangles outside the channel, extract the coordinate points corresponding to the maximum KDE value in each triangle, and obtain the maximum KDE point set;

[0085] The points in the maximum KDE point set are connected in order according to the direction of the semantic route, and the obtained path is smoothed using the Savitzky-Golay filter to extract the hottest route.

[0086] In the above technical solution, the AIS historical data of the target area is first obtained, and the semantic routes with directionality in the target area are obtained through start and end point clustering, feature point extraction and feature point clustering; then the trajectory point set of each semantic route obtained is subjected to kernel density estimation distribution calculation to extract the channel boundary point set; based on the extracted boundary point set, the channel triangulation is constructed to obtain the maximum KDE point set, that is, the maximum kernel density distribution point set, and the hottest routes in the target area are finally extracted by orderly connecting the points in the maximum KDE point set and processing the connected paths. This technical solution is based on AIS historical data, comprehensively considers the ship route mode and the trajectory density distribution in each route mode, and realizes the extraction of ship experience routes through semantic route extraction, channel boundary extraction and hottest route extraction, and realizes the extraction of high-precision hottest routes from complex AIS data.

[0087] The use of the improved DP compression algorithm to extract the characteristic points of each ship AIS track further includes: connecting the starting and ending points of the ship route track into a line as the initial baseline, and then calculating the distance from each point to the baseline to find the point with the largest distance; secondly, connecting the starting and ending points with the point to form two baselines, and then calculating the distance from each point in the first segment to the baseline and the distance from each point in the second segment to the baseline, and finally finding the point with the largest distance in the first segment and the point with the largest distance in the second segment; and repeating this cycle, when the maximum distance is less than the distance threshold parameter D max When , the loop ends; thus, the feature point set of each ship AIS track is obtained.

[0088] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the technical solutions of the embodiments of the present invention.

Claims

1. A ship experience route extraction system based on AIS historical data, characterized by: It includes a semantic route extraction system, a channel boundary extraction system and a hottest route extraction system; the semantic route extraction system is used to obtain semantic routes in the area through AIS track start and end point clustering, feature point extraction and feature point clustering; the channel boundary extraction system is used to extract boundary information for each semantic route; the hottest route extraction system is used to construct a channel triangulation network using a route boundary point set and extract the hottest route in the channel; The semantic route extraction system further comprises: firstly clustering the starting point and the ending point of the AIS data respectively using the DBSCAN clustering algorithm; then extracting the characteristic points of each ship AIS track using the improved DP compression algorithm, including: the starting point, the ending point, and the waypoint; finally clustering the characteristic points of all tracks using the DBSCAN algorithm, and the track set with the same characteristic point cluster is defined as the semantic route; The improved DP compression algorithm further comprises: First, each trajectory point is normalized using the following normalization method: ， Then, the first and last vertices of the ship's trajectory are connected by an imaginary straight line, and the synchronous Euclidean distance (SED) from the remaining vertices to the straight line is calculated. The largest one is selected and compared with the specified tolerance. If it is less than or equal to the tolerance, all points between the two ends of the straight line are deleted; if it is greater than the tolerance, the vertex with the largest distance from the straight line is retained, and the curve is divided into two parts based on this. The above method is repeated for these two parts until no further compression can be made. and The SED can be calculated by the following formula, ； The hottest route extraction system includes a channel triangulation construction module and a hottest route extraction module; The channel triangulation construction module is used to construct the channel triangulation of the semantic route using the boundary point set extracted from the semantic route based on the Delaunay triangulation method; The hottest route extraction module is used to calculate the centroid coordinates of each triangle in the channel triangulation network, filter the triangles outside the channel, extract the coordinate points corresponding to the maximum value of KDE in each triangle, connect these points in order according to the direction of the semantic route, and use the Savitzky-Golay filter to smooth the obtained path to obtain the most popular route in the semantic route, that is, the hottest route.

2. The ship experience route extraction system based on AIS historical data according to claim 1 is characterized in that: The waterway boundary extraction system includes a kernel density estimation module and a boundary point extraction module; the kernel density estimation module is used to calculate the kernel density estimation distribution of the trajectory point set of each semantic route; the boundary point extraction module is used to extract the boundary point set of the semantic route according to a set threshold, wherein the threshold is obtained by observing the kernel density estimation distribution of the trajectory point set and combining the actual application requirements for the boundary to select a value suitable for the boundary point; Wherein, the kernel density estimation module further includes: Convert the trajectory point set of each semantic route into the corresponding point in the grid , use the following formula to calculate the kernel density estimate distribution: ， in, represents the kernel function, represents the smoothness parameter, Meet the following conditions: 。 3. The ship experience route extraction system based on AIS historical data according to claim 1 is characterized in that: The channel triangulation network of the semantic route is constructed based on the boundary point set extracted from the semantic route based on the Delaunay triangulation method, which further includes: taking the extracted boundary point set as a finite point set, and the closed line segment connecting two points in the point set as an edge. At the same time, these edges must satisfy the following conditions: there are no intersecting edges; except for the endpoints, the edge does not contain any point in the point set; all faces in the plane are triangles, and the union of all triangles is the convex hull of the finite point set, thereby constructing the channel triangulation network.

4. A method for extracting ship experience routes based on AIS historical data, characterized in that: include: Get AIS historical data; The DBSCAN clustering algorithm is used to cluster the starting points and ending points of the AIS data; The improved DP compression algorithm is used to extract the characteristic points of each ship AIS track, including: the starting point, the end point, and the waypoint; the improved DP compression algorithm is used to extract the characteristic points of each ship AIS track further including: connecting the starting and ending points of the ship's track into a line as the initial baseline, and then calculating the distance from each point to the baseline to find the point with the largest distance; secondly, connecting the starting and ending points with the point to form two baselines, and then calculating the distance from each point in the first section to the baseline and the distance from each point in the second section to the baseline, and finally finding the point with the largest distance in the first section and the point with the largest distance in the second section; this cycle is repeated, and when the maximum value of the distance is less than the distance threshold parameter When , the loop ends; thus, the characteristic point set of each ship AIS track is obtained; The DBSCAN algorithm is used to cluster the feature points of all trajectories, and the set of trajectories with the same feature point clusters is defined as a semantic route; Calculate the kernel density estimation distribution of the trajectory point set of each semantic route, determine a threshold value suitable for the kernel density estimation distribution of the trajectory point set by observing the kernel density estimation distribution of the trajectory point set and combining the requirements of the actual application for the boundary, and extract the boundary point set of the semantic route according to the determined threshold value; Construct a channel triangulation network using the boundary point set of the semantic route; Calculate the coordinates of the center of gravity of the triangle, filter out the triangles outside the channel, extract the coordinate points corresponding to the maximum KDE value in each triangle, and obtain the maximum KDE point set; The points in the maximum KDE point set are connected in order according to the direction of the semantic route, and the obtained path is smoothed using the Savitzky-Golay filter to extract the hottest route.

Images