Method, apparatus, electronic device, and readable storage medium for constructing a maritime traffic network

Abstract

The present invention provides a method, apparatus, electronic device and readable storage medium for constructing a maritime traffic network. The method includes preprocessing AIS data, including filtering abnormal data by specifying a research scope, screening ship AIS data, segmenting trips, and calculating and complementing the SOG value; analyzing the trajectory set to extract static and dynamic feature points, identifying major category clusters through a clustering algorithm, and calculating the convex hull to determine the regional boundary; projecting the feature points into a grid for filtering, retaining important feature points, constructing flow directions, and finally forming a complete maritime traffic network. Through the present invention, by preprocessing AIS data, extracting feature points, clustering analysis, regional division, and flow direction construction, an accurate maritime traffic network is generated, providing decision-making support for maritime management.

Description

Technical Field

[0001] This application relates to the technical field of data processing, and particularly to a method, device, electronic device and readable storage medium for constructing a maritime traffic network. Background Art

[0002] As an important part of international trade and the global economy, maritime transportation bears more than 80% of the international cargo trade volume. Maritime transportation not only supports the large-scale cargo flow of international trade, but also plays a crucial role in the efficient operation of the global supply chain. Due to its large transportation volume, diverse shipping routes, wide coverage and other characteristics, maritime transportation has a lower cost advantage than other transportation modes. Especially in the long-distance transportation of bulk goods, more costs can be saved through ocean transportation. It also has high safety and the flexibility to adapt to different scales of transportation needs. However, with the growth of globalized trade and the expansion of the international market, maritime transportation is facing more and more challenges, such as environmental protection, shipping safety, and congestion problems in waterways and ports. The extraction of the maritime traffic network is not only related to the optimization of shipping routes, but also related to maritime safety and environmental protection. Through the extraction of a high-precision maritime traffic network, the shipping routes of ships can be effectively planned, the navigation paths can be optimized, thereby reducing the navigation time and fuel consumption, lowering the transportation cost and environmental pollution. In addition, accurate route information helps to avoid ships entering environmentally sensitive or navigationally dangerous areas, such as coral reef areas and narrow straits, thereby reducing the potential threat to the marine ecosystem and significantly improving navigation safety.

[0003] Currently, China is vigorously developing smart maritime affairs, and at the same time strengthening the construction of the Maritime Silk Road internationally. Strengthening the construction of maritime transportation is of great significance for China to become a major maritime country and a transportation power. In recent years, the development of digital and automated technologies has brought new opportunities for the development of maritime transportation. According to the regulations of the International Maritime Organization (IMO), ships with a deadweight of more than 300 tons and all passenger ships need to install the Automatic Identification System (AIS). The application of AIS and other navigation safety technologies has improved the safety and efficiency of ship navigation. AIS data includes: current time, Maritime Mobile Service Identity (MMSI), longitude and latitude, the actual speed of the ship (or vessel) relative to the ground (SOG, i.e., Speed Over Ground), course angle (COG), navigation status and other information. The extraction of the maritime traffic network based on AIS data can better understand the flow mode of maritime traffic, and this process is crucial for fields such as shipping safety, maritime traffic management, environmental monitoring and shipping logistics.

[0004] Currently, the methods for extracting marine traffic networks are mainly divided into: (1) vision-based methods, which visualize AIS data and identify waypoints in the visualized trajectories through computer image processing techniques; (2) vector-based methods, which identify marine traffic networks by extracting representative waypoints and waterways, usually by calculating the angle of trajectory change to identify waypoints; (3) grid-based methods, which divide the sea area into spatial grids and analyze the traffic characteristics in each cell to establish a marine traffic flow map.

[0005] As described above, the existing methods for extracting marine traffic networks do not mention much about data preprocessing techniques. At the same time, the fixed-size area division of marine traffic network nodes cannot well adapt to the spatial heterogeneity unique to marine traffic. Summary of the Invention

[0006] The main technical problem to be solved by the present invention is to provide a method, device, electronic device and readable storage medium for constructing a marine traffic network, which effectively solves the technical problems that the existing methods lack in-depth data preprocessing when dealing with AIS data and the area division of marine traffic network nodes is not flexible enough, and achieves the technical effect of more accurately reflecting the spatial heterogeneity and dynamic characteristics of marine traffic, thereby improving the accuracy and efficiency of marine traffic management and analysis.

[0007] To solve the above technical problem, a technical solution adopted by the present invention is: to provide a method for constructing a marine traffic network, the method includes: obtaining AIS data, preprocessing the data within the specified research range of the AIS data to obtain a set of preprocessed ship trajectories; analyzing the set of preprocessed ship trajectories to extract feature points, clustering and dividing the area of the feature points to identify and group important feature points to form area boundaries, and establishing the spatial boundaries of the marine traffic network; wherein, the feature points include static feature points and dynamic feature points; analyzing the set of preprocessed ship trajectories to extract the feature points of the ship trajectories, identifying and recording the order in which the ship passes through the feature points to form the edges of the traffic network, and constructing the flow direction of the marine traffic network according to the edges of the traffic network; and

[0008] Constructing according to the spatial boundaries of the marine traffic network and the flow direction of the marine traffic network.

[0009] Among them, to obtain AIS data and preprocess the data within the specified research scope of the AIS data, specifically including: obtaining AIS data, filtering abnormal data in the AIS data through the specified research scope, and screening vessel AIS data from the AIS data through vessel identification to obtain the vessel AIS data after data cleaning; determining that the AIS data between two consecutive recording points in the vessel AIS data after data cleaning is an AIS data segment with data missing if the time interval between the two consecutive recording points exceeds the maximum segmentation time threshold, and performing trip segmentation processing on the AIS data segment with data missing to obtain a segmented data set; calculating the SOG for consecutive AIS data points within each segment in the segmented data set, and filling the calculated SOG values into the corresponding AIS data points to obtain AIS data points with speed information; and filtering drift points in the AIS data points with speed information.

[0010] Among them, the drift points include independent drift points and continuous drift points; filtering drift points in the AIS data points with speed information specifically includes: filtering independent drift points according to the SOG in the AIS data points with speed information being less than the SOG threshold; and clustering the AIS data points within each segment in the segmented data set by adjusting the neighborhood radius and the minimum number of samples parameters to form clusters, identifying and retaining the cluster with the largest number as the main category cluster, and filtering out the data points that do not belong to any main category cluster as noise or outliers to filter continuous drift points.

[0011] Among them, clustering the AIS data points within each segment in the segmented data set by adjusting the neighborhood radius and the minimum number of samples parameters to form clusters, identifying and retaining the cluster with the largest number as the preprocessed vessel trajectory set, and filtering out the data points that do not belong to any preprocessed vessel trajectory set as noise or outliers to filter continuous drift points, specifically includes: obtaining the segmented data set and setting the neighborhood radius and the minimum number of samples; traversing all data points in the segmented data set; selecting any unmarked object in the segmented data set p and marking the object p as visited; if the number of data points within the neighborhood radius centered on the object marked as visited p is less than the minimum number of samples, execute marking the object marked as visited p as a noise point and re-traversing all data points in the segmented data set; if the number of data points within the neighborhood radius centered on the object marked as a noise point p is not less than the minimum number of samples, execute adding the object marked as a noise point p to a new cluster C and all the objects marked as noise pointsp Points within the neighborhood radius form a neighborhood sub-dataset; traverse all data points in the neighborhood sub-dataset; select any unmarked object from the neighborhood sub-dataset q , and mark the object q as visited; if the number of data points within the neighborhood radius centered on the object marked as visited q is less than the minimum sample number, execute traversing all data points in the neighborhood sub-dataset; if the number of data points within the neighborhood radius centered on the object marked as visited q is not less than the minimum sample number, execute adding all objects within the neighborhood radius of the object marked as visited q to the neighborhood sub-dataset, and adding the object that does not belong to any cluster q to the cluster C ; obtain the clustering result by traversing all data points in the neighborhood sub-dataset, calculate the number of data points in different clusters in the clustering result, retain the cluster with the largest number as the preprocessed ship trajectory set, and regard the data points that do not belong to any preprocessed ship trajectory set as noise or outliers for filtering to filter continuous drift points.

[0012] Among them, analyze the preprocessed ship trajectory set to extract feature points, perform clustering and regional division on the feature points to identify and group important feature points to form regional boundaries, and establish the spatial boundary of the maritime traffic network, specifically including: analyzing the preprocessed ship trajectory set, and marking data points with a SOG value less than the first threshold as static feature points; randomly sampling from the preprocessed ship trajectory set to obtain a subset, processing each ship trajectory in the subset according to the maximum deviation threshold, and extracting key dynamic feature points by checking the influence of the data points forming the ship trajectory on the shape of the ship trajectory to obtain a set of dynamic feature points; wherein, the ship trajectory includes multiple trajectory segments, and the maximum deviation threshold defines the maximum deviation allowed for the distance between a data point and the corresponding trajectory segment; analyzing the local density of each feature point in the set of dynamic feature points to construct a clustering hierarchy, extracting the clustering result according to the stability of the clustering hierarchy, clustering and grouping the static feature points and the dynamic feature points according to the clustering result to obtain important feature points included in a clustering group with high stability; dividing the feature points of the preprocessed ship trajectory set into different grids according to geographical location for clustering and regional division, projecting the static and dynamic feature points of the clustering group onto the grid, filtering the grid, retaining the grids containing important feature points, and removing the grids containing noise or insignificant feature points; wherein, each grid is assigned a corresponding category according to the category label of the feature points inside it; from the filtered grids, according to the category labels of the feature points inside the grid, compile the feature points belonging to the same category into a set of feature points; calculating the convex hull for each set of feature points of each category to obtain a polygon, providing a boundary for the feature point region of each category, and forming the spatial boundary of the maritime traffic network.

[0013] Among them, analyze the preprocessed ship trajectory set, identify and record the order in which the ship passes through the feature points to form the edges of the traffic network, and construct the flow direction of the maritime traffic network according to the edges of the traffic network, specifically including: analyzing the preprocessed ship trajectory set, traversing each ship trajectory, and identifying and extracting the feature points of the ship trajectory; determining and recording the order in which each ship passes through the feature points, connecting the traffic nodes in sequence to form the edges in the maritime traffic network diagram, and constructing the flow direction of the maritime traffic network.

[0014] To solve the above technical problems, another technical solution adopted by the present invention is: to provide a device for constructing a maritime traffic network, the device comprising: a data preprocessing unit, configured to obtain AIS data and preprocess the data within a specified research scope of the AIS data to obtain a set of preprocessed ship trajectories; a node establishment unit, configured to analyze the set of preprocessed ship trajectories to extract feature points, perform clustering and regional division on the feature points to identify and group important feature points to form regional boundaries, and establish the spatial boundaries of the maritime traffic network; wherein, the feature points include static feature points and dynamic feature points, which are used to represent the nodes in the maritime traffic network; a traffic network construction unit, configured to: analyze the set of preprocessed ship trajectories to extract the feature points of the ship trajectories, identify and record the order in which the ships pass through the feature points to form the edges of the traffic network, and construct the flow direction of the maritime traffic network according to the edges of the traffic network; and construct the maritime traffic network according to the spatial boundaries of the maritime traffic network and the flow direction of the maritime traffic network.

[0015] Wherein, the data preprocessing unit includes: a data cleaning module, configured to obtain AIS data, filter out abnormal data in the AIS data through a specified research scope, and screen ship AIS data from the AIS data through ship identification to obtain ship AIS data after data cleaning; a trajectory segmentation module, configured to determine, according to the time interval between two consecutive recording points in the ship AIS data after data cleaning exceeding the maximum segmentation time threshold, that the AIS data between the two consecutive recording points is an AIS data segment with missing data, and perform trip segmentation processing on the AIS data segment with missing data to obtain a segmented data set; a data completion module, configured to calculate the SOG for each continuous AIS data point within each segment in the segmented data set, and fill the calculated SOG value into the corresponding AIS data point to obtain AIS data points with speed information; a drift point filtering module, configured to filter out drift points in the AIS data points with speed information; wherein, the drift points include independent drift points and continuous drift points.

[0016] Wherein, the drift point filtering module includes: an independent drift point filtering sub-module, configured to filter out independent drift points according to the SOG in the AIS data points with speed information being less than the SOG threshold; a continuous drift point filtering sub-module, configured to cluster the AIS data points within each segment in the segmented data set by adjusting the neighborhood radius and the minimum number of samples parameters to form clusters, identify and retain the cluster with the largest number as the set of preprocessed ship trajectories, and filter out the data points that do not belong to any set of preprocessed ship trajectories as noise or outliers to filter out continuous drift points.

[0017] Among them, the continuous drift point filtering sub-module is specifically used for: obtaining the segmented data set, and setting a neighborhood radius and a minimum number of samples; traversing all data points in the segmented data set; selecting any unlabeled object in the segmented data set p , and marking the object p as visited; if the number of data points within the neighborhood radius centered on the visited object p is less than the minimum number of samples, then mark the visited object p as a noise point, and re-traverse all data points in the segmented data set; if the number of data points within the neighborhood radius centered on the object marked as a noise point p is not less than the minimum number of samples, then add the object marked as a noise point p to a new cluster C , and form a neighborhood sub-data set for all points within the neighborhood radius of the object marked as a noise point p ; traverse all data points in the neighborhood sub-data set; select any unlabeled object in the neighborhood sub-data set q , and mark the object q as visited; if the number of data points within the neighborhood radius centered on the visited object q is less than the minimum number of samples, then traverse all points in the neighborhood sub-data set; if the number of data points within the neighborhood radius centered on the visited object q is not less than the minimum number of samples, then add all objects within the neighborhood radius of the visited object q to the neighborhood sub-data set, add the objects that have not been assigned to any cluster q to the cluster C , and traverse all data points in the neighborhood sub-data set; obtain the clustering result by traversing all data points in the neighborhood sub-data set, calculate the number of data points in different clusters, retain the cluster with the largest number as the preprocessed ship trajectory set, and filter out the data points that do not belong to any preprocessed ship trajectory set as noise or outliers to filter continuous drift points.

[0018] Among them, the static feature point extraction module is used to analyze the preprocessed ship trajectory set and mark the data points with SOG values less than the first threshold as static feature points; the dynamic feature point extraction module is used to randomly sample a subset from the preprocessed ship trajectory set, process each ship trajectory in the subset according to the maximum deviation threshold, and extract key dynamic feature points by checking the influence of the data points forming the ship trajectory on the shape of the ship trajectory to obtain a dynamic feature point set; among them, the ship trajectory includes multiple trajectory segments, and the maximum deviation threshold defines the maximum deviation allowed for the distance between the data point and the corresponding trajectory segment; the feature point clustering module is used to analyze the local density of each feature point in the dynamic feature point set to construct a clustering hierarchy, extract the clustering result according to the stability of the clustering hierarchy, and cluster and group the static feature points and the dynamic feature points according to the clustering result to obtain important feature points included in the clustering group with high stability; the feature point area division module is used to: divide the feature points of the preprocessed ship trajectory set into different grids according to geographical locations for clustering and area division, project the static feature points and dynamic feature points of the clustering group onto the grid, filter the grid, retain the grids containing important feature points, and remove the grids containing noise or insignificant feature points; among them, each grid is assigned a corresponding category according to the majority category label of the internal feature points; from the filtered grids, according to the category labels of the feature points in the grid, the feature points belonging to the same category are compiled into a feature point set; and calculate the convex hull for each category of feature point set to obtain a polygon, providing a boundary for each category of feature point area and forming the spatial boundary of the maritime traffic network.

[0019] Among them, the traffic network construction unit is specifically used to: analyze the preprocessed ship trajectory set, traverse each ship trajectory, identify and extract the feature points of the ship trajectory; and determine and record the order in which each ship passes through the feature points, connect the traffic nodes in sequence to form the edges in the maritime traffic network diagram, and construct the flow direction of the maritime traffic network.

[0020] To solve the above technical problems, another technical solution adopted by the present invention is: to provide an electronic device, including: a processor and a memory, the memory is used to store computer program code, the computer program code includes computer instructions, and when the processor executes the computer instructions, the electronic device executes the steps of the above-mentioned maritime traffic network construction method.

[0021] To solve the above technical problems, another technical solution adopted by the present invention is: to provide a readable storage medium, in which a computer program is stored, the computer program includes program instructions, and when the program instructions are executed by a processor of an electronic device, the processor is caused to execute the steps of the above-mentioned method for constructing a maritime traffic network.

[0022] A method, device, electronic device and readable storage medium for constructing a maritime traffic network provided by the present invention filter abnormal data by specifying a research scope, screen vessel AIS data, perform trip segmentation processing, and calculate and complete the SOG value, ensuring the quality and accuracy of the data; by analyzing the SOG value to extract static feature points and extracting dynamic feature points based on the maximum deviation threshold, the key nodes and routes of maritime traffic can be accurately identified, providing important spatial information for network construction; by clustering the feature points using a density-based clustering algorithm, the main category clusters are effectively identified and retained, while noise and outliers are filtered out, and the region boundary is determined by calculating the convex hull of the set of feature points, adapting to feature point regions of different sizes and shapes, and more finely representing the maritime traffic node regions; further, by projecting the feature points into different grids and filtering, the grids containing important feature points are retained, reducing the influence of noise and enhancing the adaptation to spatial heterogeneity; by identifying and recording the order in which vessels pass through the feature points, the flow direction of the maritime traffic network is constructed, and combined with the spatial boundary and flow direction information, a complete maritime traffic network is constructed; based on the original AIS trajectory generated during the vessel's travel, a maritime traffic network is generated end-to-end, which is helpful for subsequent maritime traffic flow prediction and trajectory prediction, helps to understand the complexity of maritime traffic, and also provides a powerful tool for maritime management, route planning and safety monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 is a schematic flowchart of a method for constructing a maritime traffic network in an embodiment of the present invention;

[0024] Figure 2 is Figure 1 a schematic flowchart of the specific implementation method of step S10 in

[0025] Figure 3 is Figure 2 a schematic flowchart of the specific implementation method of step S102 in

[0026] Figure 4 is Figure 2 a schematic flowchart of the specific implementation method of step S103 in

[0027] Figure 5 is the effect diagram of the improved DBSCAN algorithm for filtering trajectory noise points;

[0028] Figure 6 is Figure 1 a schematic flowchart of the specific implementation method of step S11 in

[0029] Figure 7 a schematic diagram of clustering static feature points and dynamic feature points in the research area through step S112;

[0030] Figure 8 a schematic diagram of the maritime traffic network in a part of Denmark established by a method for constructing a maritime traffic network in an embodiment of the present invention;

[0031] Figure 9 a schematic diagram of the functional structure of a maritime traffic network construction device in an embodiment of the present invention;

[0032] Figure 10 is Figure 9 a schematic diagram of the functional structure of the data preprocessing unit shown;

[0033] Figure 11 is Figure 10 a schematic diagram of the functional structure of the data completion module shown;

[0034] Figure 12 is Figure 10 a schematic diagram of the functional structure of the drift point filtering module shown;

[0035] Figure 13 is Figure 9 a schematic diagram of the functional structure of the node establishment unit shown;

[0036] Figure 14 a schematic diagram of the hardware structure of an electronic device in an embodiment of the present invention.

[0037] The descriptions of the reference numerals involved in the above drawings are as follows:

[0038] 30, maritime traffic network construction device;

[0039] 31, data preprocessing unit; 310, data cleaning module; 311, trajectory segmentation module; 312, data completion module; 3120, SOG calculation sub-module; 3121, information completion sub-module; 313, drift point filtering module; 3130, independent drift point filtering sub-module; 3131, continuous drift point filtering sub-module;

[0040] 32, node establishment unit; 320, static feature point extraction module; 321, dynamic feature point extraction module; 322, feature point clustering module; 323, feature point region division module;

[0041] 33, traffic network construction unit;

[0042] 4. Electronic device; 41. Processor; 42. Memory; 43. Input device; 44. Output device; 45. Computer program; Detailed implementation manners

[0043] To describe in detail the possible application scenarios, technical principles, specific implementable solutions, achievable objectives and effects of this application, etc., the following will be described in detail with reference to the specific examples listed and in conjunction with the accompanying drawings. The embodiments described herein are only used to more clearly illustrate the technical solutions of this application, so they are only used as examples and cannot be used to limit the protection scope of this application.

[0044] Referring to "embodiments" in this article means that the specific features, structures or characteristics described in combination with the embodiments can be included in at least one embodiment of this application. The term "embodiment" appearing in various positions in the specification does not necessarily refer to the same embodiment, nor does it particularly limit its independence or relevance to other embodiments. In principle, in this application, as long as there is no technical contradiction or conflict, the technical features mentioned in each embodiment can be combined in any way to form corresponding implementable technical solutions.

[0045] Unless otherwise defined, the meanings of the technical terms used in this article are the same as those generally understood by those skilled in the technical field to which this application belongs; the use of relevant terms in this article is only to describe specific embodiments and is not intended to limit this application.

[0046] In the description of this application, the phrase "and / or" is an expression used to describe the logical relationship between objects, indicating that there can be three relationships. For example, A and / or B means: there is A, there is B, and there is both A and B at the same time. In addition, the character " / " in this article generally represents an "or" logical relationship between the associated objects before and after.

[0047] In this application, terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual quantity, primary-secondary or order relationship between these entities or operations.

[0048] Without more limitations, in this application, the open expressions such as "including", "comprising", "having" or other similar expressions used in the statement are intended to cover non-exclusive inclusion. These expressions do not exclude that there may be other elements in the process, method or product including the said elements, so that the process, method or product including a series of elements may not only include those defined elements, but also include other elements not explicitly listed, or also include elements inherent to this process, method or product.

[0049] Similar to the understanding in the "Examination Guidelines", in this application, expressions such as "greater than", "less than", and "exceeding" are understood to exclude the present number; expressions such as "above", "below", and "within" are understood to include the present number. In addition, in the description of the embodiments of this application, the meaning of "multiple" is two or more (including two). Similar expressions related to "many" are also understood in this way, such as "multiple groups", "multiple times", etc., unless otherwise clearly and specifically defined.

[0050] In the description of the embodiments of this application, the spatially related expressions used, such as "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "perpendicular", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", "axial", "radial", "circumferential", etc., the indicated orientation or positional relationship is based on the orientation or positional relationship shown in the specific embodiment or the drawing. It is only for the convenience of describing the specific embodiments of this application or for the reader to understand, rather than indicating or implying that the device or component referred to must have a specific position, a specific orientation, or be constructed or operated in a specific orientation. Therefore, it cannot be understood as a limitation to the embodiments of this application.

[0051] The efficiency and safety of maritime transportation have a significant impact on the stable operation of the global supply chain. In the embodiments of the present invention, a method for constructing a maritime traffic network is provided, which can handle various noise point situations in the trajectory, accommodate feature point regions of different sizes, and better adapt to complex maritime scenarios. By using the method for constructing a maritime traffic network provided in the embodiments of the present invention, based on the original AIS trajectory generated during the ship's voyage, a maritime traffic network is generated end-to-end, laying a foundation for subsequent maritime traffic flow prediction, trajectory prediction, maritime safety area identification, etc., and contributing to the development of intelligent maritime affairs.

[0052] Thus, please refer to Figure 1 , which is a schematic flowchart of a method for constructing a maritime traffic network in the embodiments of the present invention. The method includes:

[0053] Step S10, obtain AIS data, and preprocess the data within the specified research range of the AIS data to obtain a set of preprocessed ship trajectories.

[0054] AIS data is usually automatically generated by devices and reported manually. There are often abnormal data during the data transmission process. Therefore, it is necessary to preprocess the AIS data before analyzing it. Specifically, the AIS data preprocessing tasks include: data cleaning, trajectory segmentation, data completion, and outlier filtering algorithms.

[0055] Please also refer to Figure 2, Step S10, obtain AIS data and preprocess the data within the specified research scope of the AIS data, specifically including:

[0056] Step S100, obtain AIS data, filter out abnormal data in the AIS data through the specified research scope, and screen out vessel AIS data from the AIS data through vessel identification to obtain the vessel AIS data after data cleaning.

[0057] Among them, the specified research scope is the scope divided according to the specified longitude and latitude.

[0058] During the reporting and transmission process of AIS data, there will be abnormal situations such as error data, null values, and duplicate data. For such abnormal data, a method of dividing the research scope is adopted to filter the data within the specified longitude and latitude range, that is, the data within the specified research scope, remove the data with too many abnormal data items, and delete duplicate data.

[0059] In this embodiment, the "screen out vessel AIS data from the AIS data through vessel identification to obtain the vessel AIS data after data cleaning" is specifically: use the MMSI coding rule to screen out vessel AIS data from the AIS data to obtain the vessel AIS data after data cleaning.

[0060] Since some aircraft, coastal stations, navigation aids, etc. will also be equipped with fixed identification systems and report AIS data, this will interfere with the research of vessel AIS data. MMSI is the maritime mobile server identification in AIS data, which is a nine-digit code used to uniquely identify vessel radio communication equipment. According to the MMSI coding rule, the first digit 2 - 7 is assigned to vessels of different countries; therefore, vessel AIS data can be filtered and retained through this identification.

[0061] Step S101, based on the time interval between two consecutive record points in the vessel AIS data after data cleaning exceeding the maximum segment time threshold, determine that the AIS data between the two consecutive record points is an AIS data segment with data missing, and perform trip segmentation processing on the AIS data segment with data missing to obtain a segmented data set.

[0062] Among them, each segment contains a series of consecutive AIS data points without a time interval exceeding the maximum segment time threshold. The record point is the time point in the AIS data that records the vessel position information, and the time point includes at least the longitude and latitude coordinates, timestamp, speed, and course information of the vessel.

[0063] In this embodiment, the maximum segment time threshold is 30 minutes.

[0064] The AIS data reported by the same ship may have missing data due to reasons such as weak signals. For the AIS data segments with missing data, the maximum segmentation time threshold is applied for segmentation. If the interval between two consecutive points exceeds this threshold, a journey segmentation is performed between these two points. According to experience, the time threshold applied in this method is half an hour.

[0065] Step S102: Calculate the SOG for consecutive AIS data points within each segment in the segmented dataset, and fill the calculated SOG value into the corresponding AIS data points to obtain AIS data points with speed information.

[0066] Please refer to Figure 3 simultaneously. The "calculating the SOG for consecutive AIS data points within each segment in the segmented dataset" specifically includes:

[0067] Step S1020: For consecutive AIS data points within each segment in the segmented dataset, determine the longitude and latitude coordinates between every two AIS data points.

[0068] Step S1021: Calculate the great circle distance between the two AIS data points (i.e., the shortest path distance between two points on the earth's surface).

[0069] Step S1022: Determine the time difference between the two AIS data points (i.e., the difference in timestamps).

[0070] Step S1023: Divide the great circle distance between the two points by the time difference between the observations of these two points to obtain the SOG.

[0071] Specifically, as described in formula (1):

[0072] SOG = great circle distance / time difference Formula (1)

[0073] Furthermore, the "filling the calculated SOG value into the corresponding AIS data points to obtain AIS data points with speed information" specifically includes: Analyzing the AIS data points within each segment in the segmented dataset to determine the missing attribute information, obtaining the corresponding ship information from the ship database through ship identification, and extracting the missing attribute information from the ship database and filling it into the corresponding AIS data points to obtain AIS data points with complete attribute information. Among them, the attribute information at least includes ship type category, width, and length information.

[0074] Among them, the MMSI code in the AIS data is used as the unique identifier, and the MMSI code is matched with the ship database to obtain the corresponding ship information.

[0075] There is a lack of speed values in some AIS data. For the segmented trajectories, according to the longitude, latitude and time difference between every two points, calculate the SOG (Speed Over Ground, the actual speed of the ship on the sea surface, not affected by external factors such as water flow and wind speed) and complete the data. At the same time, there are missing data such as ship type category, width, and length in the process of reporting AIS data. As described in step S102, use the method of cross-data completion to match the MMSI of the AIS data with the ship database and complete the data such as type and size.

[0076] Step S103, filter the drift points in the AIS data points with speed information.

[0077] Among them, the drift points include independent drift points and continuous drift points.

[0078] The noise data in AIS data is mainly manifested as drift points on the trajectory. Among them, the drift points include independent drift points and continuous drift points; the independent drift points will show abnormal speed between adjacent two points in the data, and the continuous drift points will show outliers on the trajectory during the visualization process.

[0079] Step S103, that is, filter the drift points in the AIS data points with speed information, specifically including: filtering the independent drift points according to the SOG in the AIS data points with speed information being less than the SOG threshold; and by adjusting the neighborhood radius ε and the minimum number of samples MinPts parameters, cluster the AIS data points in each segment of the segmented data set to form clusters, identify and retain the cluster with the largest number as the preprocessed ship trajectory set, and regard the data points that do not belong to any preprocessed ship trajectory set as noise or outliers and filter them to filter the continuous drift points.

[0080] In this embodiment, the SOG threshold is 40 nautical miles per hour.

[0081] In this embodiment, apply the DBSCAN algorithm to each segment of the AIS data points (that is, each segmented ship trajectory) in the segmented data set to filter the continuous drift points.

[0082] Specifically, the j trajectory of the ship is defined by the observation value sequence where the data point is defined by formula (2):

[0083] Formula (2)

[0084] In formula (2), i is the index of a total of m points. Let MMSIj Represented as a ship j of MMSI , 、 and respectively represent the SOG, COG (Course Over Ground, the navigation direction of the ship on the sea surface) and navigation status of the ship j , and are respectively the latitude and longitude positions of the ship j at time .

[0085] The DBSCAN algorithm is a density-based clustering algorithm that divides data points into different clusters (categories) according to their distances and densities. A cluster is defined as the largest set of density-connected samples derived from the density reachability relationship. By setting the neighborhood radius ε and the minimum number of samples in the neighborhood MinPts , all high-density data sets are identified, and then clustering results of arbitrary shapes are discovered. By definition, the sub-data set in the data set D whose distance from the data point X is not greater than ε is called the neighborhood sub-data set N ε(xj) , calculated by formula (3):

[0086] Formula (3)

[0087] Different neighborhood sub-data sets N are grouped into different clusters, and other points that are not calculated into any cluster are grouped into outliers, that is, noise points.

[0088] Among them, in DBSCAN, a cluster is defined as a set of the largest density samples connected to each other through the density reachability relationship, that is, if there are enough points (determined according to MinPts ) in the neighborhood of a data point, then these points and the points in their neighborhoods can be grouped into the same cluster. Density reachability specifically means: If point A is within the neighborhood radius ε of point B, and there are enough points (at least MinPts ) in the neighborhood of point B, then point A is considered to be density-reachable from point B. If point B is also density-reachable from point C, and point C is within the neighborhood radius ε of point A, then point A and point C are also considered to be density-connected. The neighborhood radius ε defines how many points must be in the ε neighborhood to consider this neighborhood to be high-density and thus form the core of a cluster. A high-density data set refers to the data within the neighborhood radius εThere are at least MinPts data point sets with points, and these sets are considered to be part of the clusters.

[0089] Please also refer to Figure 4 "By adjusting the neighborhood radius ε and the minimum number of samples MinPts parameters, cluster the AIS data points in each segment of the segmented data set to form clusters, identify and retain the cluster with the largest number as the pre-processed ship trajectory set, and regard the data points that do not belong to any pre-processed ship trajectory set as noise or outliers for filtering to filter continuous drift points", specifically including:

[0090] Step S200, obtain the segmented data set D and set the neighborhood radius ε and the minimum number of samples MinPts ;

[0091] Step S201, traverse all data points in the segmented data set D ;

[0092] Step S202, determine whether there are unmarked objects in the segmented data set D ; if so, go to step S203; otherwise, go to step S212;

[0093] Among them, marking means accessing, classifying or identifying data points.

[0094] Step S203, select any unmarked object D in the segmented data set p and mark the object p as visited;

[0095] Step S204, check whether the number of data points within the neighborhood radius p centered on the object marked as visited is less than the minimum number of samples ε ; if so, go to step S205; otherwise, go to step S206; MinPts ;

[0096] Step S205, mark the object marked as visited p as a noise point, and then return to step S201;

[0097] Step S206, add the object marked as a noise point p to a new cluster C , and form a neighborhood sub-data set p for all points within the neighborhood radius ε of the object marked as a noise point N ;

[0098] Step S207, traverse all data points in the neighborhood sub - dataset N ;

[0099] Step S208, determine whether there is an unmarked object in the neighborhood sub - dataset N ; if so, go to Step S209; otherwise, return to Step S201;

[0100] Step S209, select any unmarked object from the neighborhood sub - dataset N , and mark the object q as visited; q

[0101] Step S210, check whether the number of data points within the neighborhood radius q centered on the object marked as visited is less than the minimum number of samples ε ; if so, return to Step S207; otherwise, go to Step S211; MinPts

[0102] Step S211, add all objects within the neighborhood radius q of the object marked as visited to the neighborhood sub - dataset ε , and add the objects not assigned to any cluster N to the cluster q ; then, go to Step S212. C

[0103] It should be noted that if an object q has not been assigned to any recognized cluster before and is now considered through the neighborhood sub - dataset of the object p , then under the following conditions, the object N will be added to the currently - being - constructed cluster q : C

[0104] (1) The number of data points in the neighborhood set of the object q (i.e., the set of data points within a distance not greater than the neighborhood radius q from the object) is not less than the minimum number of samples ε , which indicates that the object MinPts is a core point or a boundary point, rather than a noise point. q

[0105] (2) The object q is connected to the cluster C by the density - reachable relationship, that is, the object p is already part of the cluster C , and the object​​​​​q In the neighborhood of the object p .

[0106] Step S212, determine whether all the data points in the neighborhood sub-dataset N have been traversed; if so, go to step S213; otherwise, return to step S207.

[0107] Step S213, obtain the clustering result ( C 1, C 2, ……, C k ), calculate the number of data points in different clusters, retain the cluster with the largest number as the pre-processed ship trajectory set, and regard the data points that do not belong to any pre-processed ship trajectory set as noise or outliers for filtering to filter continuous drift points.

[0108] Specifically, calculate the number of data points in different clusters according to formula (4), and use argmax c | C | to identify the cluster with the largest number of data points, retain the identified cluster as the pre-processed ship trajectory set, and consider it to represent a normal trajectory pattern; regard the data points in other clusters as noise or outliers and filter them from the dataset.

[0109] Formula (4)

[0110] Wherein, lat j , lon j respectively represent the latitude and longitude coordinates of the j th data point, C i is the cluster obtained after clustering, and there are k different clusters in total; for each cluster C i , calculate the number of data points inside it, that is, | C i |.

[0111] Please also refer to Figure 5 , the effect diagram of the improved DBSCAN algorithm for filtering trajectory noise points. The left side is the visualization of the original data, and the right side is the visualization of the data filtered by using this algorithm. It can be seen that this algorithm can handle the noise points existing in the trajectory well.

[0112] Most of the existing ship AIS data preprocessing methods are based on speed filtering, that is, setting a threshold to remove noise points, and calculating the average speed by the distance and time difference between two adjacent data points. If the average speed is greater than the threshold, it means that the subsequent point is a noise point and is filtered. However, this method has defects: (1) When the initial point of the trajectory is a noise point, all subsequent normal points will be filtered out and the noise point will be retained, resulting in errors; (2) If there are continuous drift points, only the first noise point can be filtered out, and the subsequent noise points will not be filtered out due to the speed threshold because they are closer to the first noise point. The processing of noise points is not perfect.

[0113] The data preprocessing method described in steps S200 to S213, the DBSCAN algorithm used to filter trajectory noise points can deal with the two situations where the above-mentioned existing methods are insufficient, and has good robustness in dealing with multiple noise points in the trajectory. Through this method, abnormal points in the trajectory data can be cleaned up, including continuous outliers, so as to obtain a cleaner and more accurate trajectory data set, which is convenient for subsequent data analysis.

[0114] Step S11, analyzing the pre-processed ship trajectory set to extract feature points, clustering and region dividing the feature points to identify and group important feature points to form region boundaries, and establishing the spatial boundaries of the maritime transportation network.

[0115] The feature points include static feature points and dynamic feature points, which are used to represent nodes in the maritime transportation network.

[0116] Static feature points are areas where vessels may be paused for various reasons, including in port, waiting for a berth to become available, conducting customs inspections, or anchoring due to adverse weather conditions.

[0117] During the course of a ship's journey, there will be a place on the planned route where the ship is scheduled to change its course. When the ship reaches this place, it will often slow down and change its course. Such nodes are dynamic feature points. Subsequent ships need to slow down in advance before reaching the dynamic feature points to avoid collisions. Therefore, it is necessary to extract dynamic feature points in the process of establishing a maritime transportation network.

[0118] Please also see Figure 6 Step S11, i.e., analyzing the pre-processed ship trajectory set to extract feature points, clustering and region dividing the feature points to identify and group important feature points to form region boundaries, and establishing the spatial boundaries of the maritime transportation network, specifically includes:

[0119] Step S110, analyzing the preprocessed ship trajectory set, and marking the data point as a static feature point according to the SOG value of the data point being less than a first threshold.

[0120] In this embodiment, the first threshold is 0.2 knots, that is, data points with SOG < 0.2 knots / hour are marked as static feature points.

[0121] Step S111: Randomly sample a subset from the preprocessed set of vessel trajectories, and process each vessel trajectory in the subset according to the maximum deviation threshold. T j By checking the data points forming the vessel trajectory T j to extract key dynamic feature points based on the influence on the shape of the vessel trajectory T j to obtain a set of dynamic feature points. FP j .

[0122] Among them, the vessel trajectory T j includes multiple trajectory segments, and the maximum deviation threshold defines the maximum allowable deviation of the distance between a data point and the corresponding trajectory segment.

[0123] The AIS data points of each vessel form a corresponding trajectory set T j , and the trajectory set T j contains a series of longitude and latitude coordinate pairs representing the positions of the vessel at different time points, as described in formula (5).

[0124] Formula (5)

[0125] Process the trajectory through the RDP algorithm T j , as described in formula (6), to extract key dynamic feature points, which include the starting point, ending point of the trajectory, and key points where the course changes during navigation. The extracted set of dynamic feature points is denoted as FP j , and contains the key points on the trajectory after simplification and feature extraction.

[0126] Formula (6)

[0127] Due to the large volume of AIS data, randomly sample a subset from the full amount of AIS data. In this embodiment, data sampled evenly for 5 days each month is used as the dynamic feature point extraction set.

[0128] The RDP algorithm is a polyline simplification algorithm that can reduce the number of intermediate points while maintaining the basic shape of the polyline. The maximum deviation threshold defines the maximum deviation allowed during the simplification process. If the distance from an intermediate point to the simplified polyline segment does not exceed the first parameter value, that point can be removed.

[0129] In step S10, after the AIS data is preprocessed, the trajectory of each ship has been appropriately segmented so that the start and end of the voyage are both static feature points corresponding to potential docking areas. As described in step S111, the ship trajectory is processed by the RDP algorithm, effectively reducing the number of points within the polyline segment while retaining its basic shape, helping to identify the start point, end point, and turning points of the trajectory, that is, dynamic feature points.

[0130] Step S112, analyze the set of dynamic feature points FP j Construct a clustering hierarchy based on the local density of each feature point in the set, extract the clustering results according to the stability of the clustering hierarchy, and group the static feature points and dynamic feature points according to the clustering results to obtain the feature points of the clustering groups.

[0131] In this embodiment, the HDBSCAN clustering algorithm is applied to effectively identify and group the static feature points and dynamic feature points.

[0132] HDBSCAN is an extension of the DBSCAN algorithm and is a density-based hierarchical clustering method that can identify different density clustering regions in the same dataset. Without pre-specifying the number of clusters, it can automatically identify and extract stable clustering structures, thereby connecting data points according to density to form different clusters. This is particularly important for maritime traffic analysis because the distribution and behavior of ships may have different density characteristics in different regions.

[0133] HDBSCAN forms different clusters by constructing a clustering hierarchy and connecting data points according to density. In this hierarchical structure, the mutual relationships and stabilities between clusters can be observed. HDBSCAN uses a stability metric to evaluate the reliability of the clustering structure. High-stability clusters mean that their representation in the data is significant, which is of great significance for subsequent analysis and decision-making. Taking static feature points (such as ports, anchorages, etc.) and dynamic feature points (such as turning points on the route) as inputs, the HDBSCAN algorithm can identify the clusters to which these feature points belong, so as to identify and distinguish different maritime traffic patterns and behavior patterns, providing strong data support and insights for maritime management, route planning, and safety analysis.

[0134] Please also refer to Figure 7, which is a schematic diagram of clustering static feature points and dynamic feature points in the research area through step S112. As Figure 7 shown, each cluster is represented by a different gray level, indicating its class label.

[0135] Step S113: Divide the feature points of the preprocessed ship trajectory set into different grids according to geographical locations for clustering and regional division, project the static feature points and dynamic feature points of the clustering groups onto the grids, filter the grids, retain the grids containing important feature points, and remove the grids containing noise or insignificant feature points.

[0136] Among them, each grid is assigned a corresponding class according to the majority class label of the feature points inside it.

[0137] Specifically, the research area is spatially partitioned through geohash encoding, and the research area is recursively divided into smaller grids, improving the geographical resolution. In this embodiment, geohash encoding with 6-bit precision is used, that is, the research area is recursively divided into 4 equal parts 6 times, obtaining a geographical resolution of 0.1º latitude and 0.2º longitude.

[0138] Among them, the important feature points are extracted from the preprocessed AIS data, which are representative and significant points, such as static feature points like ports and anchorages, and dynamic feature points like turning points on the shipping route. After being processed by the clustering algorithm (such as HDBSCAN) in step S112, these points are identified and belong to clusters with high stability, indicating their significance in data analysis. The grids containing noise or insignificant feature points appear in the data set due to data errors, transmission problems, or other non-representative factors, and do not represent the main maritime traffic patterns or behaviors. During the clustering process in step S112, these points may not be assigned to any stable clusters or are identified as outliers, indicating their unimportance in the current analysis and regional division.

[0139] Step S114: From the filtered grids, according to the class labels of the feature points inside the grids, compile the feature points belonging to the same class into a feature point set; calculate the convex hull for each class of feature point set to obtain a polygon, providing a boundary for the feature point region of each class and forming the spatial boundary of the maritime traffic network.

[0140] Among them, the feature point region V i is expressed as: V i = ( C i , lat i , lon i, H i ) C i is a category label, lat i is the latitude of the regional centroid, lon i is the longitude of the regional centroid, H i is the convex hull.

[0141] After extracting the nodes, the existing methods for extracting the maritime traffic network often use the method of enclosing a specified-sized area centered on the feature points as the selection of the feature point area. This method cannot adapt to the unique spatial heterogeneity of maritime traffic. As a representative static feature point area, ports can be divided into large ports and small ports, and the dynamic feature point areas also often have different-sized waypoint areas due to the different sizes and numbers of ships passing through them.

[0142] As described in step S114, the feature point area division algorithm obtains the polygon of the corresponding feature area by calculating the convex hull of the feature point set. This method represents the maritime traffic node area more meticulously, adapts to the inherent spatial heterogeneity of such data, and can accommodate feature point areas of different sizes at the same time. Compared with the fixed feature point area, it is more in line with the actual situation and can better adapt to complex maritime scenarios.

[0143] Step S12: Analyze the preprocessed ship trajectory set to extract the feature points of the ship trajectory, identify and record the order in which the ship passes through the feature points to form the edges of the traffic network, and construct the flow direction of the maritime traffic network based on the edges of the traffic network.

[0144] Among them, the edges of the traffic network are given additional attribute information. In this embodiment, the additional attribute information includes:

[0145] Passing frequency: Records the frequency of ships passing through each edge.

[0146] Maximum width: Records the maximum width of the ships passing through this edge.

[0147] Maximum length: Records the maximum length of the ships passing through this edge.

[0148] Maximum draft: Records the maximum draft of the ships passing through this edge.

[0149] The attribute information as described above provides additional context information for each edge, which helps to analyze the ship size distribution, traffic flow, and navigation conditions, and is of great significance for maritime management, route planning, and safety monitoring, etc.

[0150] The description of "analyzing the preprocessed set of ship trajectories to extract the feature points of the ship trajectories, identifying and recording the order in which the ships pass through the feature points to form the edges of the traffic network, and constructing the flow direction of the maritime traffic network based on the edges of the traffic network" specifically includes: analyzing the preprocessed set of ship trajectories, traversing each ship trajectory, and identifying and extracting the feature points of the ship trajectories; determining and recording the order in which each ship passes through the feature points, connecting the traffic nodes in sequence to form the edges in the maritime traffic network diagram, and constructing the flow direction of the maritime traffic network.

[0151] Among them, the traffic nodes represent ports, channel intersections, or other important maritime traffic locations. The edges represent the actual maritime traffic routes.

[0152] While traversing and establishing the edges, collect the additional attribute information on each edge.

[0153] Recording the order in which each ship passes through the traffic nodes helps to understand the navigation paths of the ships and the traffic flow patterns.

[0154] Step S13, construct the maritime traffic network according to the spatial boundary of the maritime traffic network and the flow direction of the maritime traffic network.

[0155] Among them, the nodes and edges together reflect the actual layout and dynamic characteristics of the maritime traffic.

[0156] Please also refer to Figure 8 , which is a schematic diagram of the maritime traffic network in a partial area of Denmark established by a method for constructing a maritime traffic network in an embodiment of the present invention.

[0157] As described above, the method for constructing a maritime traffic network in the embodiments of the present invention has the following technical effects: By specifying the research scope to filter abnormal data, screening ship AIS data, performing trip segmentation processing, and calculating and complementing the SOG value, the quality and accuracy of the data are ensured; By analyzing the SOG value to extract static feature points and extracting dynamic feature points based on the maximum deviation threshold, the key nodes and routes of maritime traffic can be accurately identified, providing important spatial information for network construction; By clustering the feature points using the density-based clustering algorithm, the main category clusters are effectively identified and retained, while noise and outliers are filtered out, and the regional boundary is determined by calculating the convex hull of the feature point set, adapting to feature point regions of different sizes and shapes, and more precisely representing the maritime traffic node regions; Further, by projecting the feature points into different grids and filtering them, the grids containing important feature points are retained, reducing the impact of noise and enhancing the adaptability to spatial heterogeneity; By identifying and recording the order in which ships pass through the feature points, the flow direction of the maritime traffic network is constructed, and combining the spatial boundary and flow direction information, a complete maritime traffic network is constructed; According to the original AIS trajectory generated during the ship's voyage, a maritime traffic network is generated end-to-end, which is helpful for subsequent maritime traffic flow prediction and trajectory prediction, helps to understand the complexity of maritime traffic, and also provides a powerful tool for maritime management, route planning, and safety monitoring.

[0158] Please refer to Figure 9 , corresponding to the above method for constructing a maritime traffic network, an embodiment of the present invention further provides a maritime traffic network construction device 30, including:

[0159] A data preprocessing unit 31, configured to obtain AIS data and preprocess the data within the specified research scope of the AIS data to obtain a set of preprocessed ship trajectories.

[0160] A node establishment unit 32, configured to analyze the set of preprocessed ship trajectories to extract feature points, cluster and partition the feature points to identify and group important feature points to form a regional boundary, and establish the spatial boundary of the maritime traffic network. Among them, the feature points include static feature points and dynamic feature points, which are used to represent the nodes in the maritime traffic network.

[0161] A traffic network construction unit 33, configured to:

[0162] Analyze the set of preprocessed ship trajectories to extract the feature points of the ship trajectories, identify and record the order in which the ships pass through the feature points to form the edges of the traffic network, and construct the flow direction of the maritime traffic network according to the edges of the traffic network; and

[0163] Construct a maritime traffic network according to the spatial boundary of the maritime traffic network and the flow direction of the maritime traffic network.

[0164] Among them, edges in the maritime traffic network are given additional attribute information. In this embodiment, the additional attribute information includes:

[0165] Passing frequency: Records the frequency of ships passing through each edge.

[0166] Maximum width: Records the maximum width of the ships passing through this edge.

[0167] Maximum length: Records the maximum length of the ships passing through this edge.

[0168] Maximum draft: Records the maximum draft of the ships passing through this edge.

[0169] The traffic network construction unit 33 is specifically used for:

[0170] Analyzing the preprocessed set of ship trajectories, traversing each ship trajectory, identifying and extracting the feature points of the ship trajectory; and

[0171] Determining and recording the order in which each ship passes through the feature points, connecting the traffic nodes in sequence to form the edges in the maritime traffic network diagram, and constructing the flow direction of the maritime traffic network.

[0172] Please also refer to Figure 10 , specifically, the data preprocessing unit 31 includes: a data cleaning module 310, a trajectory segmentation module 311, a data completion module 312, and a drift point filtering module 313.

[0173] The data cleaning module 310 is used to obtain AIS data, filter out abnormal data in the AIS data through a specified research scope, and screen ship AIS data from the AIS data through ship identity recognition to obtain ship AIS data after data cleaning. Among them, the specified research scope is a scope divided according to specified longitude and latitude.

[0174] In this embodiment, the data cleaning module 310 is used to screen ship AIS data from the AIS data using the MMSI coding rule to obtain ship AIS data after data cleaning.

[0175] The trajectory segmentation module 311 is used to determine that the AIS data between two consecutive recording points in the ship AIS data after data cleaning is an AIS data segment with missing data if the time interval between the two consecutive recording points exceeds the maximum segmentation time threshold, and perform trip segmentation processing on the AIS data segment with missing data to obtain a segmented data set.

[0176] Among them, each segment contains a series of consecutive AIS data points with no time interval exceeding the maximum segment time threshold. The recording point is the time point in the AIS data that records the ship position information, and the time point includes at least the longitude and latitude coordinates, timestamp, speed over ground (SOG), and course information of the ship.

[0177] In this embodiment, the maximum segment time threshold is 30 minutes.

[0178] The data completion module 312 is configured to calculate the SOG for each continuous AIS data point within each segment in the segmented dataset, and complete the calculated SOG value into the corresponding AIS data point to obtain AIS data points with speed information.

[0179] Further, please refer to Figure 11 at the same time. The data completion module 312 includes:

[0180] The SOG calculation sub-module 3120 is configured to determine the longitude and latitude coordinates between every two AIS data points for each continuous AIS data point within each segment in the segmented dataset, calculate the great circle distance between the two AIS data points and the time difference between the two AIS data points, and divide the great circle distance between the two points by the time difference between the observations of the two points to obtain the SOG. As described in formula (1):

[0181] SOG = great circle distance / time difference Formula (1)

[0182] The information completion sub-module 3121 is configured to analyze the AIS data points within each segment in the segmented dataset to determine the missing attribute information, obtain the corresponding ship information from the ship database through ship identification, and extract the missing attribute information from the ship database and complete it into the corresponding AIS data point to obtain AIS data points with complete attribute information. Among them, the attribute information includes at least the ship type category, width, and length information.

[0183] The drift point filtering module 313 is configured to filter the drift points in the AIS data points with speed information. Among them, the drift points include independent drift points and continuous drift points.

[0184] Please refer to Figure 12 at the same time. The drift point filtering module 313 includes:

[0185] The independent drift point filtering sub-module 3130 is configured to filter independent drift points according to the SOG in the AIS data points with speed information being less than the SOG threshold.

[0186] The continuous drift point filtering sub-module 3131 is configured to adjust the neighborhood radius ε and the minimum number of samplesMinPts The parameter clusters the AIS data points within each segment of the segmented dataset to form clusters, identifies and retains the cluster with the largest number as the preprocessed ship trajectory set, and filters out the data points that do not belong to any preprocessed ship trajectory set as noise or outliers to filter out continuous drift points.

[0187] In this embodiment, the SOG threshold is 40 nautical miles per hour.

[0188] In this embodiment, the drift point filtering module 313 applies the DBSCAN algorithm to filter continuous drift points for the AIS data points (i.e., each segmented ship trajectory) within each segment of the segmented dataset.

[0189] Ship j 's trajectory is defined by an observation value sequence where the data point is defined by formula (2):

[0190] Formula (2)

[0191] In formula (2), i is the index of a total of m points. Denote MMSI j as the j of ship MMSI , , and respectively represent the SOG, COG and navigation status of ship j , and are respectively the latitude and longitude positions of ship j at time .

[0192] The dataset D sub-dataset whose distance from the data point X is not greater than ε is called the neighborhood sub-dataset N ε(xj) , and is calculated by formula (3):

[0193] Formula (3)

[0194] The continuous drift point filtering sub-module 3131 is specifically used for:

[0195] Obtain the segmented dataset D , and set the neighborhood radius ε and the minimum number of samples MinPts ;

[0196] Traverse all data points in the segmented data set D ;

[0197] Select any unlabeled object in the segmented data set D and mark the object p marked as visited as visited; p If the number of data points within the neighborhood radius

[0198] centered on the object p marked as visited is less than the minimum number of samples ε then mark the object p marked as visited as a noise point and re - execute traversing all data points in the segmented data set MinPts ; If the number of data points within the neighborhood radius D centered on the object p marked as visited is not less than the minimum number of samples ε then add the object p marked as visited to a new cluster MinPts and form a neighborhood sub - data set from all points within the neighborhood radius C of the object p marked as visited ε ; N ;

[0199] Traverse all data points in the neighborhood sub - data set N ;

[0200] Select any unlabeled object in the neighborhood sub - data set N and mark the object p marked as visited as visited; q If the number of data points within the neighborhood radius

[0201] centered on the object p marked as visited is less than the minimum number of samples ε then traverse all points in the neighborhood sub - data set MinPts ; If the number of data points within the neighborhood radius N centered on the object p marked as visited is not less than the minimum number of samples ε then add all objects within the neighborhood radius MinPts of the object p marked as visited to the neighborhood sub - data set ε , add the object p marked as visited that has not been assigned to any cluster to the cluster N , and execute traversing all data points in the neighborhood sub - data set C ; N ;

[0202] Traverse all data points in the neighborhood sub - data set to obtain the clustering result ( N 1, C 1, C2, ……, C k ), calculate the number of data points in different clusters, retain the cluster with the largest number as the pre - processed ship trajectory set, and regard the data points that do not belong to any pre - processed ship trajectory set as noise or outliers for filtering to filter continuous drift points.

[0203] Specifically, the continuous drift point filtering sub - module 3131 calculates the number of data points in different clusters according to formula (4) and uses argmax c | C | to identify the cluster with the largest number of data points, retain the identified cluster as the pre - processed ship trajectory set, considering it represents a normal trajectory pattern; regard the data points in other clusters as noise or outliers and filter them from the dataset.

[0204] Formula (4)

[0205] Among them, respectively represent the latitude and longitude coordinates of the j th data point, C i is the cluster obtained after clustering, and there are k different clusters in total; for each cluster C i , calculate the number of data points inside it, that is, | C i |.

[0206] Please also refer to Figure 13 , the node establishment unit 32 includes: a static feature point extraction module 320, a dynamic feature point extraction module 321, a feature point clustering module 322, and a feature point region division module 323.

[0207] The static feature point extraction module 320 is used to analyze the pre - processed ship trajectory set and mark the data points with the SOG value less than the first threshold as static feature points.

[0208] In this embodiment, the first threshold is 0.2 knots.

[0209] The dynamic feature point extraction module 321 is used to randomly sample a subset from the pre - processed ship trajectory set and process each ship trajectory in the subset according to the maximum deviation threshold T j by checking the data points forming the ship trajectory T j for the ship trajectory T jExtract key dynamic feature points based on the shape to obtain a set of dynamic feature points FP j 。

[0210] In this embodiment, the dynamic feature point extraction module 321 uses the RDP algorithm to process each trajectory in the subset T j Extract key dynamic feature points to obtain a set of dynamic feature points FP j 。

[0211] The AIS data points of each ship form a corresponding trajectory set T j ,The trajectory set T j contains a series of longitude and latitude coordinate pairs, representing the positions of the ship at different time points, as described in formula (5).

[0212] Formula (5)

[0213] Process the trajectory through the RDP algorithm T j ,As described in formula (6), extract the key dynamic feature points, which include the starting point, ending point of the trajectory and the key points where the course changes during navigation. The extracted set of dynamic feature points is denoted as FP j ,Contains the key points on the trajectory after simplification and feature extraction.

[0214] Formula (6)

[0215] In this embodiment, 5 days of data are evenly sampled each month as the dynamic feature point extraction set.

[0216] The feature point clustering module 322 is used to analyze the local density of each feature point in the set of dynamic feature points to construct a clustering hierarchy, extract the clustering result according to the stability of the clustering hierarchy, and group the static feature points and dynamic feature points according to the clustering result to obtain the feature points of the clustering grouping. FP j 。

[0217] In this embodiment, the feature point clustering module 322 applies the HDBSCAN clustering algorithm to effectively identify and group the static feature points and dynamic feature points.

[0218] The feature point region division module 323 is used for:[[]]

[0219] The feature points of the preprocessed ship trajectory set are divided into different grids according to geographical locations for clustering and regional division, and the static feature points and dynamic feature points of the clustering groups are projected onto the grids. Then, the grids are filtered to retain the grids containing important feature points and remove the grids containing noise or insignificant feature points. Among them, each grid is assigned a corresponding category according to the majority category label of the feature points inside it;

[0220] From the filtered grids, according to the category labels of the feature points inside the grids, the feature points belonging to the same category are compiled into a feature point set; and

[0221] For each category of feature point sets, the convex hull is calculated to obtain a polygon, which provides a boundary for the feature point region of each category and forms the spatial boundary of the maritime traffic network.

[0222] Among them, the feature point region V i is expressed as: V i =( C i , lat i , lon i , H i ), where C i is the category label, lat i is the latitude of the regional centroid, lon i is the longitude of the regional centroid, H i is the convex hull.

[0223] Specifically, the feature point region division module 323 performs spatial partitioning on the research area through geohash encoding, and recursively divides the research area into smaller grids. In this embodiment, geohash encoding with 6-bit precision is used, that is, the research area is recursively divided into 4 equal parts 6 times, obtaining a geographical resolution of 0.1º latitude and 0.2º longitude.

[0224] The advantages and beneficial effects of a method for constructing a maritime traffic network have been elaborated above and will not be repeated here. Moreover, since a device for constructing a maritime traffic network is applied to a method for constructing a maritime traffic network, it has the same advantages and beneficial effects as the method for constructing a maritime traffic network.

[0225] One embodiment of the present invention further provides a readable storage medium storing a computer program, the computer program including program instructions which, when executed by a processor of an electronic device, cause the processor to execute the steps of the maritime traffic network construction method described in any of the above embodiments.

[0226] One embodiment of the present invention further provides an electronic device, including: a processor and a memory, the memory being used for storing computer program code, the computer program code including computer instructions which, when executed by the processor, cause the electronic device to execute the steps of the maritime traffic network construction method described in any of the above embodiments.

[0227] Please refer to Figure 14 , which is a schematic hardware structure diagram of an electronic device provided by an embodiment of the present invention.

[0228] The electronic device 4 includes a processor 41, a memory 42, an input device 43, and an output device 44. The processor 41, the memory 42, the input device 43, and the output device 44 are coupled through a connector, and the connector includes various interfaces, transmission lines, buses, etc., and the embodiments of the present invention do not limit this. It should be understood that in various embodiments of the present invention, coupling refers to the mutual connection in a specific manner, including directly connected or indirectly connected through other devices, for example, connected through various interfaces, transmission lines, buses, etc.

[0229] The processor 41 described in the embodiments of the present application can be implemented by hardware, firmware, software, or a combination thereof. It can use circuits, a single or multiple application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), central processing units (CPUs), controllers, microcontrollers, microprocessors, or at least one of them. It also includes other physical, biological, or chemical structures that can achieve functions similar to or equivalent to those of the above-listed processors, such as biological neurons, quantum computing units, DNA computing units, etc., so that the processor can execute some steps, all steps, or any combination of the steps mentioned in the computer programs or methods involved in the various embodiments of the present application. In the electronic device 4 in this embodiment, physically, the processor 41 can be a single processor or a combination of two or more processors. For example, a combination of a CPU and a GPU, a combination of a CPU and an FPGA, a combination of a CPU and an ASIC, a combination of a CPU and a coprocessor, and a combination of a CPU and a DSP, etc. In the embodiments, these combinations are also collectively referred to as processors like a single processor.

[0230] The computer program 45 involved in the embodiments can be stored in a computer-readable storage medium of a computer device. The computer-readable storage medium of the computer device includes, but is not limited to, magnetic disks, magnetic tapes, magnetic cards, floppy disks, flash memories, optical discs, optical cards, read-only memories (ROMs), random access memories (RAMs), erasable programmable ROMs (EPROMs), and electrically erasable programmable ROMs (EEPROMs), etc. It also includes other biological, physical, or chemical structures that can achieve functions similar to or equivalent to those of the above-listed storage media, such as units with information storage capabilities like DNA, RNA, proteins, etc. In a specific embodiment, the storage medium involved can be one of the above medium types or a combination of the above medium types. In different embodiments, the computer program involved in the embodiments can be centrally stored in a single medium or distributedly stored in multiple media.

[0231] The memory 42 containing a computer-readable storage medium can be a non-volatile memory or a random access memory. These computer-readable storage media can be built into the device or connected to the device involved in the embodiment as an external device or a part of an external device. In some embodiments, the memory with a computer-readable storage medium is deployed locally; in other embodiments, a scheme of deploying the memory away from the processor can also be adopted, such as a network-attached memory accessed via an RF circuit or an external port and a communication network, where the communication network can be the Internet, one or more intranets, a local area network (LAN), a wide area wireless network (WLAN), a storage area network (SAN), etc., or a suitable combination thereof, as long as the computer device can access the memory. In addition, the computer programs involved in the embodiments can be stored in plaintext / ciphertext form or designed as training data and integrated and reorganized implicitly and saved in the parameter states of a deep neural network or other machine learning models through model training.

[0232] As Figure 14 shown in the embodiment, the computer program 45 is stored in the memory 42 of the electronic device 4. However, the reader should be aware that other schemes capable of enabling the processor to execute the computer program are feasible.

[0233] The input device 43 is used to input data and / or signals, and the output device 44 is used to output data and / or signals. The output device 44 and the input device 43 can be independent devices or an integrated device.

[0234] It can be understood that in the embodiment of the present invention, the memory 42 is not only used to store relevant instructions, and the embodiment of the present invention does not limit the specific data stored in this memory.

[0235] It can be understood that Figure 14 only a simplified design of an electronic device is shown. In practical applications, the electronic device may also separately include other necessary elements, including but not limited to any number of input / output devices, processors, memories, etc., and all methods for constructing a maritime traffic network that can implement the embodiment of the present invention are within the protection scope of the present invention.

[0236] In several embodiments provided by the present invention, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the modules or units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the coupling or direct coupling or communication connection between each other can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be in an electrical or other form.

[0237] The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units, that is, they may be located in one place, or they may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0238] In addition, each functional unit in various embodiments of the present invention can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units.

[0239] If the above-mentioned integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, all or part of the technical solution of the present invention can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a management server, or a network device, etc.) or a processor to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: USB flash drives, mobile hard disks, read-only memories (English: read-only memory, abbreviated: ROM), random access memories (English: Random Access Memory, abbreviated: RAM), magnetic disks, or optical disks, etc., which can store program codes.

[0240] Finally, it should be noted that although the above embodiments have been described in the text of the specification and drawings of this application, the patent protection scope of this application cannot be limited thereby. All technical solutions obtained by equivalent structure or equivalent process substitution or modification based on the essential concept of this application and using the content recorded in the text of the specification and drawings of this application, as well as those directly or indirectly implementing the technical solutions of the above embodiments in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for constructing a maritime traffic network, characterized in that The method includes: Obtaining AIS data and preprocessing the AIS data within a specified research scope to obtain a set of preprocessed ship trajectories; wherein, preprocessing the AIS data within the specified research scope to obtain the set of preprocessed ship trajectories specifically includes: Screening ship AIS data from the AIS data through ship identification, and performing journey segmentation processing according to the time interval between consecutive recording points exceeding the maximum segmentation time threshold to generate a segmented data set; Calculating the SOG value for consecutive AIS data points within each segment in the segmented data set, and complementing the SOG value to the corresponding AIS data points to form an AIS data set with speed information; and Filtering independent drift points based on the SOG threshold, and clustering the AIS data points in the segmented data set by dynamically adjusting the neighborhood radius and the minimum sample parameter, identifying and retaining the cluster with the largest number as the main category cluster, and filtering out consecutive drift points; Analyzing the set of preprocessed ship trajectories, extracting static feature points and dynamic feature points; performing hierarchical clustering on the static feature points and dynamic feature points, constructing a clustering hierarchy based on local density, and screening high-stability clustering groups according to the clustering stability of the clustering hierarchy; projecting the clustered feature points onto an adaptive geographic grid, and filtering according to the class labels of the feature points within the grid, retaining the grids containing important feature points; wherein, the static feature points are determined by the SOG value of the data points being less than the first threshold, representing the ship berthing area; the dynamic feature points are determined by randomly sampling a subset from the set of preprocessed ship trajectories and processing each ship trajectory in the subset based on the maximum deviation threshold to extract the data points that are crucial for the shape of the ship trajectory; Calculating the convex hull for the set of feature points of each category in the filtered grid to generate a polygon boundary, and establishing the spatial boundary of the maritime traffic network; Analyzing the order in which ships pass through the feature points, connecting the traffic nodes in sequence to form the edges and flow directions of the traffic network, and constructing a complete maritime traffic network in combination with the spatial boundary.

2. The method for constructing a maritime traffic network according to claim 1, wherein In filtering independent drift points based on the SOG threshold, and clustering the AIS data points in the segmented data set by dynamically adjusting the neighborhood radius and the minimum sample parameter, identifying and retaining the cluster with the largest number as the main category cluster, and filtering out consecutive drift points, dynamically adjusting the neighborhood radius and the minimum sample parameter includes: Dynamically reducing the neighborhood radius or increasing the minimum sample number according to the distribution density of the data points in the segmented data set to identify the outliers in the sparse areas.

3. The method for constructing a maritime traffic network according to claim 1, characterized in that The division accuracy of the adaptive geographic grid is dynamically adjusted according to the feature point density.

4. The method for constructing a maritime transportation network according to claim 1, characterized in that In performing hierarchical clustering on the static feature points and dynamic feature points, constructing a clustering hierarchy based on local density, and screening high-stability clustering groups according to the clustering stability, the hierarchical clustering is based on the HDBSCAN algorithm, and high-stability clustering groups are screened according to local density and clustering stability.

5. The method for constructing a maritime traffic network according to claim 1, wherein, When the dynamic feature points are obtained by randomly sampling a subset from the preprocessed set of ship trajectories and processing each ship trajectory in the subset based on a maximum deviation threshold to extract the data points that are crucial for influencing the trajectory shape, the maximum deviation threshold is defined as follows: calculate the distance between adjacent trajectory points and the corresponding trajectory line segment, and if the distance exceeds the maximum deviation threshold, mark this point as a dynamic feature point.

6. The method for constructing a maritime traffic network according to claim 1, characterized in that, When analyzing the order in which ships pass through the feature points, connecting the traffic nodes in sequence to form the edges and flow directions of the traffic network, and constructing a complete maritime traffic network in combination with the spatial boundaries, the edges of the traffic network are assigned attribute information, including at least one of the number of passages, maximum ship width, maximum ship length, and maximum draft.

7. A device for constructing a maritime traffic network, characterized in that, The device includes: A data preprocessing unit for obtaining AIS data and preprocessing the AIS data within a specified research scope to obtain a preprocessed set of ship trajectories; the data preprocessing unit includes: A data cleaning module for screening ship AIS data from the AIS data through ship identification. A trajectory segmentation module for performing trip segmentation processing on the screened ship AIS data according to the time interval between consecutive recorded points exceeding the maximum segmentation time threshold to generate a segmented data set. A data completion module for calculating the SOG value for each consecutive AIS data point within each segment in the segmented data set and completing the SOG value into the corresponding AIS data points to form an AIS data set with speed information. A drift point filtering module for filtering independent drift points based on the SOG threshold and clustering the AIS data points in the segmented data set by dynamically adjusting the neighborhood radius and minimum sample parameters, identifying and retaining the cluster with the largest number as the main category cluster, and filtering out consecutive drift points. A node establishment unit for analyzing the preprocessed set of ship trajectories, extracting static feature points and dynamic feature points; calculating the convex hull for each category of feature point set in the filtered grid to generate a polygon boundary, and establishing the spatial boundary of the maritime traffic network; the node establishment unit includes: A static feature point extraction module for determining the static feature points by the SOG value of the data points being less than the first threshold to represent the ship berthing area. A dynamic feature point extraction module for obtaining the dynamic feature points by randomly sampling a subset from the preprocessed set of ship trajectories and processing each ship trajectory in the subset based on a maximum deviation threshold to extract the data points that are crucial for influencing the trajectory shape. A feature point clustering module for performing hierarchical clustering on the static feature points and dynamic feature points, constructing a clustering hierarchy based on local density, and screening high-stability clustering groups according to clustering stability. A feature point area division module for projecting the clustered feature points into an adaptive geographical grid, filtering according to the category labels of the feature points within the grid, and retaining the grids containing important feature points. A traffic network construction unit for analyzing the order in which ships pass through the feature points, connecting the traffic nodes in sequence to form the edges and flow directions of the traffic network, and constructing a complete maritime traffic network in combination with the spatial boundary.

8. The marine traffic network construction device according to claim 7, characterized in that, The drift point filtering module includes: A continuous drift point filtering sub-module, which is used to dynamically reduce the neighborhood radius or increase the minimum number of samples according to the distribution density of data points in the segmented dataset to identify the outliers in the sparse area and filter the continuous drift points.

9. The device for constructing a maritime traffic network according to claim 7, wherein The division accuracy of the adaptive geographic grid is dynamically adjusted according to the feature point density.

10. The marine traffic network construction device according to claim 7, characterized in that, The feature point clustering module is used to perform hierarchical clustering on static feature points and dynamic feature points based on the HDBSCAN algorithm, construct a clustering hierarchy based on local density, and screen high-stability clustering groups according to local density and clustering stability.

11. The device for constructing a maritime transportation network according to claim 7, wherein, The dynamic feature point extraction module is used to calculate the distance between adjacent trajectory points and the corresponding trajectory line segment. If the distance exceeds the maximum deviation threshold, the point is marked as a dynamic feature point.

12. The offshore traffic network construction device according to claim 7, characterized in that, The edges of the traffic network are given attribute information, including at least one of the number of passing times, the maximum ship width, the maximum ship length, and the maximum draft depth.

13. An electronic device, comprising: A processor and a memory, characterized in that the memory is used to store computer program code, the computer program code includes computer instructions, and when the processor executes the computer instructions, the electronic device executes the steps of the maritime traffic network construction method according to any one of claims 1 to 6.

14. A readable storage medium stores a computer program, characterized in that, The computer program includes program instructions, and when the program instructions are executed by the processor of the electronic device, the processor is caused to execute the steps of the maritime traffic network construction method according to any one of claims 1 to 6.

Images