Ship channel management method and system based on data analysis

Abstract

The invention discloses a ship channel management method and system based on data analysis, and relates to the technical field of data processing.The method includes the steps that a global hexagonal discrete grid structure is constructed, historical operation data are processed through a large arc gridding algorithm to calculate grid density and time potential energy, and accordingly the grid heat level is judged; the method comprises the following steps: acquiring real-time data of a target ship, calculating a comprehensive risk score and a red region intention probability, constructing a multi-dimensional state vector containing a grid level, establishing a hierarchical control strategy, sequentially judging state vector indexes according to priorities, automatically matching emergency takeover, smooth transition, red region strict management or green cruise modes, and generating a corresponding instruction; according to the method, complex space search is converted into efficient integer operation through the discrete grid technology, risk pre-judgment is achieved through the intention probability model, and the problems that in channel management, calculation power consumption is large, pre-warning lags, instruction conflicts exist and the like are effectively solved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and system for ship and waterway management based on data analysis. Background Technology

[0002] With the rapid development of the global shipping industry, the density of waterway traffic continues to increase, and the navigation environment for ships is becoming increasingly complex. The traditional waterway management model, which relies on human experience and VHF voice dispatch, can no longer meet the requirements of modern maritime supervision for real-time performance and safety. In order to improve management efficiency, existing technologies are mostly committed to enhancing the ability to perceive the ship's status through the fusion of multi-source sensor data. Chinese invention patent application 202510763207.2 discloses a waterway intelligent management method and system based on multi-source data fusion. This technology constructs a joint ship identification model that includes AIS, radar, vision, and infrared, dynamically adjusts the sensor weights and data acquisition frequency according to environmental parameters, and corrects the collision risk score in combination with hydrological and meteorological conditions. To a certain extent, it solves the problems of target loss in harsh environments and insufficient accuracy of single sensors, and improves the robustness of ship dynamic monitoring.

[0003] The aforementioned existing technologies still have significant technical limitations when facing high-concurrency monitoring scenarios involving massive numbers of ships. In terms of spatial data processing, existing technologies typically rely on continuous latitude and longitude coordinates for distance calculation and neighborhood search. This approach involves a large number of floating-point operations and complex geometric relationship judgments. As the number of ships within the monitoring range increases, the computing power consumption rises exponentially, making it difficult to achieve millisecond-level rapid response to large-scale waterway situations. Although existing risk assessment models consider environmental correction factors, they mainly focus on calculating the probability of physical collisions based on the distance and relative speed at the current moment, lacking in-depth analysis of historical waterway traffic patterns. Furthermore, the judgment of a ship's intention to enter a high-risk area is often delayed, usually triggering an early warning only when the ship has already crossed the boundary or is extremely close. Existing technologies often use single-dimensional threshold judgments to screen key ships for monitoring and implement interventions. In complex scenarios, this can easily lead to conflicting instructions generated by different early warning logics, or cause sudden deceleration of ships and traffic flow oscillations due to the lack of gradient in control instructions. These technologies cannot meet the actual needs of refined and automated waterway management. Summary of the Invention

[0004] The technical problem addressed by this invention is that existing technologies typically rely on continuous latitude and longitude coordinates for distance calculation and neighborhood search in spatial data processing. This approach involves a large number of floating-point operations and complex geometric relationship judgments. As the number of ships within the monitoring range increases, the computational power consumption rises exponentially, making it difficult to achieve millisecond-level rapid response to large-scale waterway situations. Although existing risk assessment models consider environmental correction factors, they mainly focus on calculating the probability of physical collisions based on the distance and relative speed at the current moment, lacking in-depth analysis of historical waterway traffic patterns. Furthermore, the judgment of a ship's intention to enter a high-risk area is often delayed, usually triggering an early warning only when the ship has already crossed the boundary or is extremely close. Existing technologies often use single-dimensional threshold judgments to screen key ships and implement interventions. In complex scenarios, this can easily lead to conflicts between instructions generated by different early warning logics, or cause sudden deceleration of ships and traffic flow oscillations due to the lack of gradient in control instructions. This fails to meet the actual needs of refined and automated waterway management.

[0005] To address the aforementioned technical problems, this invention provides the following technical solution: a data analysis-based method for ship channel management, comprising the following steps:

[0006] Step S1: Construct a global hexagonal discrete grid structure. Based on historical operation data, obtain the grid sequence of each historical operation data through the large circular arc gridding algorithm, calculate the heat coefficient of the grid, and determine the grid heat level based on the grid heat coefficient.

[0007] Step S2: Obtain the grid where the target ship is located and the heat level of the grid, calculate the comprehensive risk score and red zone intention probability of the target ship, and construct the state vector of the target ship;

[0008] Step S3: Construct a hierarchical waterway management strategy. According to the preset priority order, judge each indicator in the state vector in turn and execute the corresponding management strategy.

[0009] Preferably, in step S1, the historical operation data includes ship ID, geographic coordinates, heading, speed, and timestamp;

[0010] The historical operation data is denoised, outliers are removed, and interpolation is performed.

[0011] Preferably, in step S2, the specific process of constructing the global hexagonal discrete grid structure includes:

[0012] The icosahedron is used as the basic polyhedron. The mapping relationship between the sphere and the polyhedron is established by using the Snyder equal area projection. A global hexagonal discrete mesh structure is constructed based on the recursive meshing method of type II aperture 4. A unique code is assigned to each hexagonal mesh.

[0013] The specific process of obtaining the grid sequence includes:

[0014] The geographic coordinates in each historical data are arranged in chronological order to obtain the trajectory sequence. For each pair of adjacent sampling points in each trajectory sequence, the large circle arc gridding algorithm is used to calculate all the hexagonal grids traversed by the large circle arc path between the two points, generating the corresponding grid subsequence.

[0015] All grid subsequences are spliced ​​together end-to-end according to the time sequence of the sampling points. During the splicing process, duplicate grids at the connection points of adjacent subsequences are removed to generate the trajectory grid sequence corresponding to each historical running data.

[0016] Preferably, the large circular arc meshing algorithm specifically includes the following steps:

[0017] Extract two adjacent sampling points from each trajectory sequence and denote them as the first trajectory point and the second trajectory point;

[0018] Based on the geographic coordinates of the first trajectory point, obtain the hexagonal grid where the first trajectory point is located and record it as the first starting grid. Use the hexagonal grid where the second trajectory point is located as the ending grid.

[0019] If the first starting grid and the ending grid are the same grid, then the first trajectory point and the second trajectory point are determined to be located in the same hexagonal grid.

[0020] If the first starting grid and the ending grid are not the same grid, then it is determined that the first trajectory point and the second trajectory point are not located in the same hexagonal grid, and the outgoing edge calculation is performed. The calculation of the outgoing edge specifically includes:

[0021] Based on the geographic coordinates of the first trajectory point and the second trajectory point, and combined with the spherical trigonometry formula, the angle from the first trajectory point to the second trajectory point is obtained, and the angle is taken as the first major azimuth angle.

[0022] Based on the geographic coordinates of the first trajectory point and the six vertices of the first starting grid, calculate the geodetic azimuth from the first trajectory point to each vertex.

[0023] The first geodetic azimuth angle and the geodetic azimuth angle from the first trajectory point to each vertex are numerically compared to obtain two adjacent vertices within the numerical range of the geodetic azimuth angle, and the edge formed by the two adjacent vertices is taken as the outgoing edge.

[0024] Based on the topological relationship of the global hexagonal grid, find the unique neighboring grid that shares the outgoing edge with the first starting grid, and use the neighboring grid as the second starting grid. Based on the first major azimuth angle and the major azimuth angle from the first trajectory point to each vertex of the second starting grid, repeat the steps of calculating the outgoing edge and finding the neighboring grid until the starting grid and the ending grid are the same grid.

[0025] Preferably, the specific process of obtaining the heat level of the grid in step S2 includes:

[0026] Traverse all the grid coding sequences generated by historical trajectories, accumulate the total number of times each grid code appears in all sequences, define the total number of times as the density potential energy of the grid, and use the frequency as the density potential energy;

[0027] Based on the total time taken for each historical trajectory and the number of grids traversed, the average dwell time of the ship in each grid is calculated, and the time potential energy of the grid is accumulated.

[0028] The heat coefficient of each hexagonal grid is calculated using a weighted formula. If the heat coefficient is greater than or equal to the preset high potential energy threshold, the heat level of the hexagonal grid is determined to be red.

[0029] If the heat coefficient is less than the preset high potential energy threshold, the heat level of the hexagonal grid will be determined as green.

[0030] Preferably, in step S2, the specific process of constructing the state vector of the target ship includes:

[0031] Obtain the target ship's real-time speed, heading, and heat level of the hexagonal grid in which the target ship is located;

[0032] Calculate the comprehensive risk score of the target vessel. The mathematical expression for the comprehensive risk score is as follows:

[0033] ;

[0034] in, For comprehensive risk scoring, The real-time speed of the target vessel. The preset speed limit threshold for the grid where the target vessel is currently located. The Euclidean distance between the target vessel and the nearest vessel. This is the distance sensitivity attenuation coefficient. This is the grid heat level coefficient. These are the normalized weighting coefficients;

[0035] Calculate the red zone intention probability of the target vessel, the mathematical expression for which is:

[0036] ;

[0037] in, The probability of a target vessel's intention to enter the red-level grid. The target vessel's current heading angle. The azimuth angle of the target vessel pointing towards the center point of the nearest red-level hexagonal grid. This represents the absolute value of the heading deviation. The real-time spatial distance from the target ship to the boundary of the red-level grid. The preset search radius for the intended purpose;

[0038] The heat level of the grid where the target vessel is located, the comprehensive risk score, and the red zone intent probability constitute the state vector of the target vessel.

[0039] Preferably, in step S3, constructing the hierarchical waterway management strategy specifically includes:

[0040] According to the preset priority order, each element in the state vector is judged in turn, and the corresponding management strategy is executed.

[0041] The first priority is the comprehensive risk score, the second priority is the probability of red zone intent, and the third priority is the heat level of the hexagonal grid where the target ship is located;

[0042] Extract the comprehensive risk score from the state vector of the target vessel. If the comprehensive risk score is greater than or equal to the preset emergency threshold, mark the vessel's navigation mode as emergency takeover mode.

[0043] If the overall risk score is less than the preset emergency threshold, the probability of red zone intent is determined. If the probability of red zone intent is greater than or equal to the preset intent threshold, the ship's navigation mode is marked as transition smooth mode.

[0044] If the overall risk score is less than the preset emergency threshold and the red zone intent probability is less than the preset intent threshold, then the heat level of the hexagonal grid where the target ship is located will be determined.

[0045] If the heat level of the hexagonal grid is red, then the ship's navigation mode will be marked as red zone strict control mode.

[0046] If the heat level of the hexagonal grid is green, then the ship's navigation mode will be marked as green cruise mode.

[0047] Preferably, in step S3, executing the corresponding management strategy specifically includes:

[0048] If the ship's navigation mode is emergency takeover mode, a forced maneuver command is generated; if the ship's navigation mode is transition smooth mode, a gradient deceleration command is generated; if the ship's navigation mode is red zone strict control mode, a violation correction command is generated; if the ship's navigation mode is green cruise mode, a gradient acceleration command is generated.

[0049] Preferably, the forced maneuver command includes an emergency rudder angle value, the gradient deceleration command includes a target speed value sequence, the violation type in the violation correction command includes insufficient spacing and speeding, and the gradient acceleration command includes a suggested speed.

[0050] The data-driven ship navigation management system includes a grid construction and heat map module, a state vector construction module, and a hierarchical control strategy execution module.

[0051] The grid construction and heat labeling module is used to construct a global hexagonal discrete grid structure. Based on historical operation data, it obtains the grid sequence of each historical operation data through a large circular arc gridding algorithm, calculates the heat coefficient of the grid, and determines the heat level of the grid based on the heat coefficient of the grid.

[0052] The state vector construction module is used to obtain the grid where the target ship is located and the heat level of the grid, calculate the comprehensive risk score and red zone intent probability of the target ship, and construct the state vector of the target ship.

[0053] The hierarchical control strategy execution module is used to establish hierarchical waterway management strategies, judge each indicator in the state vector in a preset priority order, and execute the corresponding management strategies.

[0054] The beneficial effects of this invention are as follows: By constructing a global hexagonal discrete grid structure, the computational efficiency and dynamic adaptability of waterway environment modeling are significantly improved. This invention utilizes a large circular arc gridding algorithm and integer coordinate offset logic to transform complex spatial geometric search into constant-time complexity operations, enabling rapid mapping and heat map calibration of massive historical trajectory data. Through the weighted aggregation of density potential energy and time potential energy, high-risk red zones (such as port congestion points) and low-risk green zones in waterways can be adaptively identified, overcoming the deficiency of static electronic charts in reflecting real-time traffic flow pressure. By constructing a comprehensive risk score, immediate physical collision risks are captured, and the risks are weighted and amplified by combining grid environment attributes. The introduction of the red zone intent probability model utilizes the mathematical relationship between heading deviation and distance ratio to identify the intention of a vessel to enter the high-risk area before it physically enters, thereby triggering smooth transition control and avoiding the lag of passive intervention only after a vessel crosses the boundary in traditional schemes. Attached Figure Description

[0055] Figure 1 A flowchart illustrating the steps of a data analysis-based ship channel management method according to an embodiment of the present invention;

[0056] Figure 2 This is a partial example of a pre-defined face adjacency topology lookup table. Detailed Implementation

[0057] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0058] Example 1, referring to Figure 1 It provides a data analysis-based method for ship channel management, including the following steps:

[0059] Step S1: Construct a global hexagonal discrete grid structure. Based on historical operation data, obtain the grid sequence of each historical operation data through the large circular arc gridding algorithm, calculate the heat coefficient of the grid, and determine the grid heat level based on the grid heat coefficient.

[0060] Step S2: Obtain the grid where the target ship is located and the heat level of the grid, calculate the comprehensive risk score and red zone intention probability of the target ship, and construct the state vector of the target ship;

[0061] Step S3: Construct a hierarchical waterway management strategy. According to the preset priority order, judge each indicator in the state vector in turn and execute the corresponding management strategy.

[0062] This invention discretizes continuous marine geographic space into grid cells with thermal levels. By combining real-time calculated multidimensional state vectors with hierarchical waterway management strategies, it overcomes the shortcomings of traditional waterway management, such as over-reliance on human experience, delayed early warnings, and conflicting instructions. It can quickly complete the entire process from mapping massive amounts of data to generating precise control instructions, significantly improving the automation level and safe navigation efficiency of waterway management.

[0063] In step S1, the historical operation data includes the vessel ID, geographic coordinates, heading, speed, and timestamp.

[0064] Denoising the historical operating data, removing outliers, and interpolating to fill in the gaps.

[0065] In one specific embodiment of the present invention, historical operational data of different vessels passing through the target waterway in the past year are collected.

[0066] For each operational trajectory, the entry point is defined as entering the target waters and the exit point as leaving the target waters. The operational data of the vessel during the time from the entry point to the exit point is collected, including the vessel ID, the timestamp, geographical coordinates (latitude and longitude), speed, and heading at each sampling moment.

[0067] In step S2, the specific process of constructing the global hexagonal discrete grid structure includes:

[0068] The icosahedron is used as the basic polyhedron. The mapping relationship between the sphere and the polyhedron is established by using the Snyder equal area projection. A global hexagonal discrete mesh structure is constructed based on the recursive meshing method of type II aperture 4. A unique code is assigned to each hexagonal mesh.

[0069] The specific process of obtaining the grid sequence includes:

[0070] The geographic coordinates in each historical data are arranged in chronological order to obtain the trajectory sequence. For each pair of adjacent sampling points in each trajectory sequence, the large circle arc gridding algorithm is used to calculate all the hexagonal grids traversed by the large circle arc path between the two points, generating the corresponding grid subsequence.

[0071] All grid subsequences are spliced ​​together end-to-end according to the time sequence of the sampling points. During the splicing process, duplicate grids at the connection points of adjacent subsequences are removed (i.e., when the tail grid of the previous subsequence is the same as the head grid of the next subsequence, only one is retained), generating the trajectory grid sequence corresponding to each historical running data.

[0072] The spatial geometry of a regular icosahedron includes the coordinates of the center points of 12 vertices and 20 equilateral triangular faces, given the geographic coordinates of any input ship. Based on the characteristic dimensions of the vertices of a regular icosahedron ( The global spatial logic is divided into the Arctic Crown Region, the Equatorial Ring Region, and the Antarctic Crown Region;

[0073] like ,determination The point belongs to one of the five faces of the Arctic crown region, if ,determination The point belongs to one of the five facets of the Antarctic crown region. ,determination The point belongs to one of the 10 faces of the equatorial circumpolar region;

[0074] Within a defined latitude zone, based on the rotational symmetry of the icosahedron, the specific plane to which it belongs is determined using the longitude subdivision method. Longitude range according to The sector is divided by intervals, and the formula for calculating the longitude sector number is as follows:

[0075] ;

[0076] in, For the longitude sector number, This is the preset starting longitude offset (180° in this embodiment). for The geographical longitude of the point;

[0077] For a coordinate point located at the intersection of two surfaces (i.e., near the boundary of the great circle arc), calculate... The point normal distance relative to the plane of the great circle arc will Assign it to the nearest face to determine ID of the surface to which it belongs (denoted as) (Value range: 0-19)

[0078] By applying the Snyder equal-area inverse projection formula, the latitude and longitude coordinates of the sphere are transformed into local two-dimensional Cartesian coordinates on the surface. The local Cartesian coordinates are then converted into a tilted coordinate system (simplex coordinate system) for hexagonal meshes. The conversion formula utilizes the basis vector transformation of hexagonal geometry.

[0079] According to the preset grid level (This embodiment uses level 12, where the average side length of the grid is approximately 1.2 kilometers.) The coordinates in the continuous inclined coordinate system are discretized and rounded. Specifically, the coordinate point is calculated at the 12th level using the recursive subdivision logic of aperture 4. The row and column indices of the specific grid cells belonging to the hierarchy;

[0080] Finally, bitwise operations are used to... hierarchical The row index and column index are combined and encapsulated into a 64-bit long integer, which is: The hierarchical grid coding of the grid to which a point belongs, through this indexing algorithm, enables time complexity mapping from continuous geographic space to discrete grid space, and can quickly locate the hexagonal grid where a ship is located without traversing massive amounts of data;

[0081] In other words, as long as the geographical coordinates of the ship are available, the code of the hexagonal grid where the ship is located can be obtained. By querying the code, it can be determined whether two geographical coordinates are located in the same grid and subsequent statistical operations can be performed.

[0082] This invention employs an icosahedral Sneider equal-area projection combined with a Class II aperture 4 recursive subdivision method, along with unique bitwise operation encoding logic, to solve the problem of low efficiency in traditional latitude and longitude calculations when handling massive spatial queries. By introducing algorithms such as longitude sector division, initial longitude offset, and local Cartesian coordinate transformation, this scheme transforms complex spherical geometric search into efficient integer bitwise operations and constant-level algebraic transformations, allowing direct location of the grid cell where the ship is located and its neighborhood relationships through encoding. This significant reduction in time complexity is the key technology for achieving large-scale real-time monitoring and rapid response of waterways.

[0083] The large circular arc meshing algorithm specifically includes the following steps:

[0084] Extract two adjacent sampling points from each trajectory sequence and denote them as the first trajectory point and the second trajectory point;

[0085] Based on the geographic coordinates of the first trajectory point, obtain the hexagonal grid where the first trajectory point is located and record it as the first starting grid. Use the hexagonal grid where the second trajectory point is located as the ending grid.

[0086] If the first starting grid and the ending grid are the same grid, then the first trajectory point and the second trajectory point are determined to be located in the same hexagonal grid.

[0087] If the first starting grid and the ending grid are not the same grid, then it is determined that the first trajectory point and the second trajectory point are not located in the same hexagonal grid, and the outgoing edge calculation is performed. The calculation of the outgoing edge specifically includes:

[0088] Based on the geographic coordinates of the first and second trajectory points, and combined with the spherical trigonometry formula, the angle from the first trajectory point to the second trajectory point is obtained, and the angle is taken as the first major azimuth.

[0089] Based on the geographic coordinates of the first trajectory point and the six vertices of the first starting grid, calculate the geodetic azimuth from the first trajectory point to each vertex.

[0090] Compare the first geodetic azimuth angle with the geodetic azimuth angle from the first trajectory point to each vertex, obtain the two adjacent vertices of the geodetic azimuth angle within its numerical range, and take the edge formed by the two adjacent vertices as the outgoing edge.

[0091] Based on the topological relationship of the global hexagonal grid, find the unique neighboring grid that shares the outgoing edge with the first starting grid, and use the neighboring grid as the second starting grid. Based on the first major azimuth angle and the major azimuth angle from the first trajectory point to each vertex of the second starting grid, repeat the steps of calculating the outgoing edge and finding the neighboring grid until the starting grid and the ending grid are the same grid.

[0092] The large circular arc meshing algorithm proposed in this invention achieves continuous and seamless mapping of trajectory segments in discrete grid space by calculating the numerical comparison logic between the outgoing edge and the azimuth angle. This solves the problem that traditional sampling point mapping methods are prone to missing intermediate grids under low-frequency sampling, ensuring that every physical grid passed by the ship can be accurately captured and included in the statistical sequence. This not only guarantees the topological continuity of the grid trajectory, but also provides a rigorous geometric basis for subsequent accurate calculation of grid density potential energy.

[0093] In step S2, the specific process of obtaining the heat level of the grid includes:

[0094] Traverse all the grid coding sequences generated by historical trajectories, accumulate the total number of times each grid code appears in all sequences, define the total number of times as the density potential energy of the grid, and use the frequency as the density potential energy;

[0095] Based on the total time taken for each historical trajectory and the number of grids traversed, the average dwell time of the ship in each grid is calculated, and the time potential energy of the grid is accumulated.

[0096] The heat coefficient of each hexagonal grid is calculated using a weighted formula. If the heat coefficient is greater than or equal to the preset high potential energy threshold, the heat level of the hexagonal grid is determined to be red.

[0097] If the heat coefficient is less than the preset high potential energy threshold, the heat level of the hexagonal grid will be determined as green.

[0098] In one specific embodiment of the present invention, for each ship's operating trajectory, a trajectory sequence is obtained. This trajectory sequence includes several geographical coordinates arranged by time; in this embodiment, there are three coordinates, namely... , , ;

[0099] according to Geographic coordinates (latitude and longitude) to obtain The encoding of the hexagonal grid in which it is located is used to designate the hexagonal grid as the first starting grid;

[0100] examine Get the coordinates The code of the hexagonal grid in which it is located is used to designate the hexagonal grid as the termination grid. If the codes of the first starting grid and the termination grid are the same, then... and Within the same grid, arrive The segment mapping ends, and the record starts from... arrive Encoding of the passed hexagonal grid;

[0101] If the encodings of the first starting grid and the ending grid are inconsistent, then and Not located in the same grid, from arrive Its trajectory also passed through other grids, in order to continue searching from... arrive The path passes through the next grid cell, and the exit direction is calculated. The calculation process for the exit direction specifically includes:

[0102] Using spherical geometry formulas, calculate from point to large azimuth Obtain the 6 vertices of the starting mesh. The coordinates are calculated from the following. point to The azimuth of the land Encoding the vertices of the grid in clockwise order;

[0103] The calculation of geodetic azimuth utilizes the difference in latitude and longitude between two points. The formula for calculating geodetic azimuth is:

[0104] ;

[0105] in, The largest local azimuth angle The geographic coordinates of the first trajectory point. The geographic coordinates of the second trajectory point;

[0106] Similarly, based on the formula for calculating the first major azimuth angle, the major azimuth angles from the first trajectory point to the six vertices of the first starting grid can be obtained, denoted as... , The value ranges from 1 to 6;

[0107] The geographic coordinates of the vertices of the hexagonal grid are obtained as follows:

[0108] The indexing algorithm of the discrete mesh model is invoked to parse the encoding of the hexagonal mesh into local triangular coordinates on the face of a regular icosahedron. According to the partitioning geometry rules of Class II aperture 4, the coordinates of the six vertices of the hexagonal mesh element are calculated in the plane coordinate system.

[0109] Using the inverse transformation formula of Snyder equal area projection, the coordinates of the six plane vertices are mapped to the sphere, and the corresponding geographical latitude and longitude coordinates are calculated.

[0110] like lie in and between( The value of is 1~5), and and The triangular region (composed of the grid center and two vertices) covers this direction, and the algorithm determines that the great circle path will pass through the connection. and The side that is emitted;

[0111] Calculate the first major azimuth angle from the first trajectory point to the second trajectory point. Simultaneously, obtain the standard azimuth angle intervals of the six vertices of the current grid (first starting grid) relative to the grid center (e.g., the azimuth angle corresponding to edge 0). Edge 1 corresponds to (And so on), by comparison, determine which side the great circle path originates from, and record the number of that exiting side as... (Values ​​range from 0 to 5);

[0112] A set of coordinate offset constants is preset. The coordinate offset constants define the incremental changes of the row index and column index when the hexagonal grid moves in six adjacent directions in a two-dimensional tilted coordinate system.

[0113] Let the two-dimensional coordinates after the current grid encoding parsing be... The coordinate offset constant array is defined as follows (taking clockwise direction as an example):

[0114] Launching edge 0 (Northeast): ;

[0115] Launching edge 1 (due east): ;

[0116] Shooting side 2 (southeast): ;

[0117] Shooting side 3 (southwest): ;

[0118] Launching edge 4 (due west): ;

[0119] Launching from side 5 (northwest) ;

[0120] Read the calculated exit edge number directly Obtain the corresponding coordinate increment from the offset array mentioned above. ;

[0121] Calculate the coordinates of adjacent grid cells using the integer addition formula:

[0122] ;

[0123] ;

[0124] in, This is the row number of the current starting grid. The row number of the adjacent target grid. The column number of the current starting grid. The column number of the adjacent grid (i.e., the second starting grid);

[0125] Furthermore, the above method of calculating the coordinates of adjacent grid cells by adding and subtracting integers ( This only applies to cases where the candidate coordinates do not cross the boundary of a regular icosahedron. When the value exceeds the current plane's domain, cross-plane processing must be performed.

[0126] In this embodiment, a face adjacency topology lookup table and a cross-face coordinate transformation matrix are preset when constructing a global hexagonal discrete mesh;

[0127] The current grid level is Then the effective range of coordinates for each grid on each face is ,when , , or When an out-of-bounds error occurs, the out-of-bounds direction of the ejection edge is determined based on the specific numerical characteristics of the out-of-bounds error.

[0128] like and If the boundary is crossed from the due north direction (let's say Edge 0), then... and The boundary is determined to be crossed from the due east direction, and so on, covering the boundary crossing conditions in six directions;

[0129] The pre-defined face adjacency topology lookup table is a table with dimension 1. A two-dimensional array, with rows representing the current face numbers and columns representing the calculated outgoing edge numbers. (0-5) After crossing the exit edge along the boundary direction from the current face, enter the next face, which is the target face. Enter the target face (target face number (0-19)). The number of times the coordinate system needs to rotate counterclockwise is the number of coordinate system rotations. This indicates that when crossing from the current plane to the target plane, the coordinate axes need to be rotated counterclockwise. The number of times;

[0130] The pre-defined face adjacency topology lookup table is pre-calculated based on the standard unfolded diagram of a regular icosahedron (using Fuller unfolding), and the number of rotations is obtained. Then, the out-of-bounds coordinates are transformed using a preset coordinate transformation algorithm. Mapped to new coordinates within the target plane ;

[0131] Translate the out-of-bounds coordinates back to the origin of the local coordinate system, rotating counterclockwise each time in the hexagonal tilted coordinate system. The coordinate transformation formula is:

[0132] ;

[0133] in, and These represent the original row and column indices before the coordinate transformation. and These represent the target row index and target column index after coordinate system rotation transformation, respectively. Representing the The maximum boundary index value of the local coordinate system under the hierarchy. This value serves as the translation reference constant for coordinate rotation transformation, ensuring that the rotated coordinates still fall within the valid positive integer range of the target surface.

[0134] The calculated and ( and The cells are recombined to obtain the unique code of the adjacent cells (i.e. the second starting cell). This step replaces the complex geometric search with constant-level integer addition and subtraction operations, achieving millisecond-level localization of neighboring cells.

[0135] After finding neighboring grids that share an outgoing edge with the starting grid, obtaining the codes of these neighboring grids, and using them as the new starting grids (second starting grids), the calculation is performed in these new starting grids. The geodetic azimuth angles of the six vertices of the new starting grid are used to calculate the outgoing edges, obtain adjacent grids, and repeat this process several times until the endpoint is reached. Located inside the grid, at this time, arrive The segment mapping ends; record from arrive Encoding of all grids traversed;

[0136] right arrive Repeat the above process to obtain another encoded sequence. Finally, concatenate the two sequences end to end (removing duplicate intermediate grid points) to obtain the complete grid encoded sequence of the ship within that time period.

[0137] For a given hexagonal grid, query the number of all trajectories passing through the target grid from the grid coding sequences corresponding to all historical running data. That is, query the number of times the hexagonal grid code appears in all grid coding sequences, and use the number as the density potential energy of the hexagonal grid.

[0138] Obtain the time difference between any two adjacent sampling points, and obtain the total number of grids traversed between adjacent sampling points (the number of elements in the sub-grid sequence). Use the ratio of the time difference to the total number of grids as the average dwell time within each grid traversed between adjacent sampling points. Accumulate the average dwell times within the same grid to obtain the time potential energy.

[0139] The density potential energy and total time potential energy of the grid are normalized and then weighted and aggregated to obtain the heat coefficient of the grid.

[0140] The formula for calculating the heat coefficient is:

[0141] ;

[0142] in, For the first The heat coefficient of each grid, For the first Density potential energy of each grid (normalized). This represents the time potential energy of the grid (after normalization). and These are the maximum density potential energy and the maximum time potential energy for all grid cells, respectively. and In this embodiment, the weighting coefficient is used. and Both are 0.5, indicating that the density potential energy and time potential energy of the grid are equally important;

[0143] If the heat coefficient is greater than or equal to the preset high potential energy threshold (set to 0.4 in this embodiment), the heat level of the hexagon is determined to be red (such as the port entrance and exit of a waterway).

[0144] If the heat coefficient is less than the preset high potential energy threshold, the heat level of the hexagonal grid will be determined as green.

[0145] This invention not only statistically analyzes the frequency (density) of ships appearing within the grid, but also innovatively introduces a dwell time (time) calculation based on a trajectory segment equalization strategy. This allows for the accurate differentiation between high-frequency passing areas and truly congested areas with long-term lingering. This heat coefficient calculation method based on weighted aggregation provides a quantitative standard that conforms to objective physical laws for classifying waterways into red and green levels, so that the definition of red zones no longer relies on subjective experience, but truly reflects traffic flow pressure.

[0146] In step S2, the specific process of constructing the state vector of the target ship includes:

[0147] Obtain the target ship's real-time speed, heading, and heat level of the hexagonal grid in which the target ship is located;

[0148] The comprehensive risk score of the target vessel is calculated, and the mathematical expression for the comprehensive risk score is:

[0149] ;

[0150] in, For comprehensive risk scoring, The real-time speed of the target vessel. The preset speed limit threshold is the grid where the target ship is currently located (in this embodiment, the speed limit threshold is 8 knots in the red grid and 18 knots in the green grid). The Euclidean distance (in meters) between the target vessel and the nearest vessel. This is the distance sensitivity attenuation coefficient (used to adjust the sensitivity of distance to risk; the closer the distance, the more exponentially this value increases). The unit is The reciprocal of the unit is 1 / meter, which makes Dimensionless, in this embodiment (Value is 0.005) This is the grid heat level coefficient (in this embodiment, the value is 1.0 if the grid is red and 0.5 if it is green). For normalized weighting coefficients (in this embodiment, It is 0.3. It is 0.5. The normalized weighting coefficient is 0.2, and can be calibrated based on historical accident data and navigation environment of specific waterways.

[0151] The probability of a target vessel's red zone intention is calculated, and the mathematical expression for the red zone intention probability is:

[0152] ;

[0153] in, The probability of a target vessel's intention to enter the red-level grid. The target vessel's current heading angle. The azimuth angle of the target vessel pointing towards the center point of the nearest red-level hexagonal grid. This represents the absolute value of the heading deviation (the smaller the deviation, the closer the cosine value is to 1). The real-time spatial distance from the target ship to the boundary of the red-level grid. The preset intent search radius (when the distance exceeds this radius, the intent probability returns to zero; in this embodiment, the intent search radius is 2-3 nautical miles).

[0154] The heat level of the grid where the target vessel is located, the comprehensive risk score, and the red zone intent probability constitute the state vector of the target vessel.

[0155] In one specific embodiment of the present invention, the geographic coordinates of the target ship are collected in real time by an edge device, mapped to a global hexagonal grid structure, and the hexagonal grid and grid code of the target ship are obtained. The heat level of the grid is found according to the grid code. The heat level of the hexagonal grid is represented by a code, with red level being 1 and green level being 0.

[0156] Based on the hexagonal grid where the target ship is currently located, expand outwards using the grid topology. Layer Neighborhood Grid (in this embodiment) The search involves searching two surrounding grids (a total of 18 grids) to establish a local search domain. Within this domain, traffic situation data for the target vessel is acquired, including the real-time geographic coordinates, speed, and heading of nearby vessels. The real-time geographic coordinates of other vessels are then incorporated into a dynamic candidate set. The electronic chart database is queried to retrieve geographic features marked as non-navigable within the local search domain (such as shoreline boundary points, the edges of island and reef polygons, and the locations of buoys). The geographic coordinates of these features are then incorporated into a static candidate set.

[0157] Calculate the distances between the target ship's geographic coordinates and all elements in both the static and dynamic candidate sets.

[0158] For all elements in the dynamic candidate set, calculate the point-to-point Euclidean distance; for all elements in the candidate set, calculate the shortest perpendicular distance from a point to a line segment (shoreline) or from a point to the edge of a polygon.

[0159] Store all calculated distance values ​​in a list, sort them, and select the minimum value as the final value. ;

[0160] Regarding the method for calculating the real-time spatial distance from the target vessel to the red zone boundary, this embodiment uses the boundary grid indexing method to quickly calculate the shortest spatial distance. The specific steps are as follows:

[0161] Traverse all grid cells marked as red, check the 6 topological neighbors of each grid, and if any neighbor of a red zone grid is not red, define the grid as a red zone boundary grid; obtain the center set composed of the center point coordinates of all boundary grids.

[0162] Obtain the real-time geographic coordinates of the target vessel and calculate its Euclidean distance to the center point of each boundary grid in the central set; select the minimum value among the above distance calculation results as the real-time spatial distance.

[0163] If the target vessel is already inside the red-level grid, then define At this point, the probability of the red zone intention is no longer calculated.

[0164] This invention constructs a multidimensional state vector that includes a comprehensive risk score and a red zone intention probability, enabling the system to have a dual perception capability of the ship's current risk and future intention. The introduction of a risk score model can enhance the ship's nonlinear perception characteristics of near-field dangers and ensure that the risk value increases during a near-collision. The intention probability model based on the ratio of heading deviation to distance provides predictive ability, enabling the identification of the ship's tendency to enter the red zone before the ship physically enters it.

[0165] Step S3, specifically, involves constructing a tiered waterway management strategy, including:

[0166] According to the preset priority order, each element in the state vector is judged in turn, and the corresponding management strategy is executed.

[0167] The first priority is the comprehensive risk score, the second priority is the probability of red zone intent, and the third priority is the heat level of the hexagonal grid where the target ship is located;

[0168] Extract the comprehensive risk score from the state vector of the target vessel. If the comprehensive risk score is greater than or equal to the preset emergency threshold (set to 0.8 in this embodiment), then mark the vessel's navigation mode as emergency takeover mode.

[0169] If the overall risk score is less than the preset emergency threshold, the probability of red zone intent is determined. If the probability of red zone intent is greater than or equal to the preset intent threshold, the ship's navigation mode is marked as transition smooth mode.

[0170] If the overall risk score is less than the preset emergency threshold (set to 0.6 in this embodiment), and the red zone intent probability is less than the preset intent threshold, then the heat level of the hexagonal grid where the target ship is located is determined.

[0171] If the heat level of the hexagonal grid is red, then the ship's navigation mode will be marked as red zone strict control mode.

[0172] If the heat level of the hexagonal grid is green, then the ship's navigation mode will be marked as green cruise mode.

[0173] This invention ensures that, by setting a priority order, the emergency avoidance logic (first priority) involving life safety is always higher than the conventional navigation efficiency logic at any time. This deterministic state machine design eliminates the possibility of command oscillation, enabling the system to always output a unique, safe and optimal control mode when facing complex and ever-changing waterway situations.

[0174] In step S3, executing the corresponding management strategy specifically includes:

[0175] If the ship's navigation mode is emergency takeover mode, a forced maneuver command is generated; if the ship's navigation mode is transition smooth mode, a gradient deceleration command is generated; if the ship's navigation mode is red zone strict control mode, a violation correction command is generated; if the ship's navigation mode is green cruise mode, a gradient acceleration command is generated.

[0176] This invention transforms abstract control models into four types of specific standardized execution instructions, enabling the executable implementation of management decisions. The mandatory maneuver, gradient deceleration, violation correction, and gradient acceleration instructions generated for different models correspond to four typical business scenarios: emergency avoidance, zone transition, strict zone management, and zone efficiency improvement. This instruction-based output method allows the cloud-based control center to directly issue specific operational tasks to the vessel or VTS system.

[0177] Forced maneuver commands include emergency rudder angle values; gradient deceleration commands include a sequence of target speed values; violation correction commands include insufficient spacing and speeding; and gradient acceleration commands include suggested speeds.

[0178] In a specific embodiment of the present invention, the forced maneuver command is a control message containing an emergency rudder angle value, wherein the emergency rudder angle value is the minimum deflection angle (e.g., 20 degrees to the right) calculated by the system through a geometric collision avoidance algorithm, which makes the closest encounter distance between the target vessel and the obstacle target greater than a safety threshold (200 meters in this embodiment).

[0179] Gradient deceleration commands contain a sequence of timestamped target speed values, with the target speed value at each moment based on... The mathematical expression for linear interpolation is as follows:

[0180] ;

[0181] in, The target speed value, Speed ​​limits are preset for red-level grids (Section 8 of this embodiment). The ship's current speed. To determine the search radius, the gradient deceleration command requires the ship to reduce its speed for every kilometer it approaches the boundary of the red-level grid. This ensures that the ship's real-time speed converges precisely to the specified value when it reaches the boundary of the red-level grid. ;

[0182] Violation correction instructions are correction parameters generated when a vessel is in the red zone and violates navigation rules; if the violation type is speeding (i.e. The instruction was to immediately reduce the speed to [a certain value]. ,in The safety margin factor is set at 0.9 in this embodiment, which means that the speed must be reduced to 90% of the speed limit to prevent secondary violations caused by critical fluctuations.

[0183] If the violation type is insufficient spacing (i.e.) The instruction was to adjust the course. ,in To restore the predicted distance between the target vessel and the nearest neighbor obstacle to (In this example, the required lateral yaw angle is 500 meters (e.g., yaw 5 degrees to the side without obstacles).

[0184] The gradient acceleration command is a recommended message containing a suggested cruise speed, which is the maximum speed limit allowed by the green level (18 knots in this embodiment).

[0185] This invention provides refined numerical definitions for various control commands, significantly improving their engineering usability and safety. In particular, the distance-based linear interpolation formula introduced in the gradient deceleration command guides the ship to complete smooth deceleration over long distances, effectively avoiding the risk of rear-end collisions and sudden changes in main engine load caused by sudden braking at the boundary. At the same time, the safety margin coefficient (such as 0.9 times the speed limit) set in the violation correction command provides a reasonable fault tolerance buffer for ship operation, preventing repeated violation alarms caused by critical fluctuations, demonstrating the robustness and humanization of the system design.

[0186] The data-driven ship navigation management system includes a grid construction and heat map module, a state vector construction module, and a hierarchical control strategy execution module.

[0187] The grid construction and heat indexing module is used to construct a global hexagonal discrete grid structure. Based on historical operation data, it obtains the grid sequence of each historical operation data through a large circular arc gridding algorithm, calculates the heat index of the grid, and determines the heat index level of the grid based on the heat index.

[0188] The state vector construction module is used to obtain the grid where the target ship is located and the heat level of the grid, calculate the comprehensive risk score and red zone intention probability of the target ship, and construct the state vector of the target ship.

[0189] The hierarchical control strategy execution module is used to establish hierarchical waterway management strategies. According to the preset priority order, it judges each indicator in the state vector in turn and executes the corresponding management strategy.

[0190] This invention significantly improves the computational efficiency and dynamic adaptability of waterway environment modeling by constructing a global hexagonal discrete grid structure. Unlike traditional geographic information systems based on continuous latitude and longitude, this invention utilizes a large circular arc gridding algorithm and integer coordinate offset logic to transform complex spatial geometric searches into constant-time computations, enabling rapid mapping and heat map calibration of massive historical trajectory data. Through weighted aggregation of density potential energy and temporal potential energy, it can adaptively identify high-risk red zones (such as port congestion points) and low-risk green zones in waterways. This data-driven dynamic environment base construction method overcomes the deficiency of static electronic charts in reflecting real-time traffic flow pressure. Furthermore, by constructing a comprehensive risk score that includes distance sensitivity and grid heat map level, it can not only capture immediate physical collision risks but also amplify risks by combining grid environment attributes. The introduction of a red zone intent probability model, utilizing the mathematical relationship between heading deviation and distance ratio, can... This invention can identify a vessel's intention to enter a high-risk area before it physically does, thus triggering smooth transition control and avoiding the lag in traditional solutions where passive intervention only occurs after a vessel crosses the boundary. The priority-based full-process hierarchical control system established by this invention achieves closed-loop control from emergency avoidance to efficient cruising through standardized quantitative commands. Based on a preset priority arbitration logic, it automatically switches between four modes: emergency takeover, smooth transition, strict red zone control, and green zone cruising, ensuring the absolute priority of safety commands and completely eliminating decision oscillations in multi-objective optimization scenarios. In particular, for the gradient deceleration command generated in the smooth transition mode, a linear interpolation algorithm is used to accurately calculate the target speed at each distance node, guiding the vessel to complete smooth deceleration before reaching the red zone boundary, eliminating the risk of rear-end collisions caused by sudden braking at the boundary. At the same time, the gradient acceleration strategy in the green zone mode fully releases the navigation potential of idle waterways, achieving the global optimal allocation of waterway resource safety and efficiency.

[0191] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0192] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A data analysis-based method for ship and waterway management, characterized in that, Includes the following steps: Step S1: Construct a global hexagonal discrete grid structure. Based on historical operation data, obtain the grid sequence of each historical operation data through the large circular arc gridding algorithm, calculate the heat coefficient of the grid, and determine the grid heat level based on the grid heat coefficient. Step S2: Obtain the grid where the target ship is located and the heat level of the grid, calculate the comprehensive risk score and red zone intention probability of the target ship, and construct the state vector of the target ship; Step S3: Construct a hierarchical waterway management strategy. According to the preset priority order, judge each indicator in the state vector in turn and execute the corresponding management strategy.

2. The data analysis-based ship navigation management method as described in claim 1, characterized in that, In step S1, the historical operation data includes ship ID, geographic coordinates, heading, speed and timestamp; The historical operation data is denoised, outliers are removed, and interpolation is performed.

3. The data analysis-based ship channel management method as described in claim 2, characterized in that, In step S2, the specific process of constructing the global hexagonal discrete grid structure includes: The icosahedron is used as the basic polyhedron. The mapping relationship between the sphere and the polyhedron is established by using the Snyder equal area projection. A global hexagonal discrete mesh structure is constructed based on the recursive meshing method of type II aperture 4. A unique code is assigned to each hexagonal mesh. The specific process of obtaining the grid sequence includes: The geographic coordinates in each historical data are arranged in chronological order to obtain the trajectory sequence. For each pair of adjacent sampling points in each trajectory sequence, the large circle arc gridding algorithm is used to calculate all the hexagonal grids traversed by the large circle arc path between the two points, generating the corresponding grid subsequence. All grid subsequences are spliced ​​together end-to-end according to the time sequence of the sampling points. During the splicing process, duplicate grids at the connection points of adjacent subsequences are removed to generate the trajectory grid sequence corresponding to each historical running data.

4. The data analysis-based ship navigation management method as described in claim 2, characterized in that, The large circular arc meshing algorithm specifically includes the following steps: Extract two adjacent sampling points from each trajectory sequence and denote them as the first trajectory point and the second trajectory point; Based on the geographic coordinates of the first trajectory point, obtain the hexagonal grid where the first trajectory point is located and record it as the first starting grid. Use the hexagonal grid where the second trajectory point is located as the ending grid. If the first starting grid and the ending grid are the same grid, then the first trajectory point and the second trajectory point are determined to be located in the same hexagonal grid. If the first starting grid and the ending grid are not the same grid, then it is determined that the first trajectory point and the second trajectory point are not located in the same hexagonal grid, and the outgoing edge calculation is performed. The calculation of the outgoing edge specifically includes: Based on the geographic coordinates of the first trajectory point and the second trajectory point, and combined with the spherical trigonometry formula, the angle from the first trajectory point to the second trajectory point is obtained, and the angle is taken as the first major azimuth angle. Based on the geographic coordinates of the first trajectory point and the six vertices of the first starting grid, calculate the geodetic azimuth from the first trajectory point to each vertex. The first geodetic azimuth angle and the geodetic azimuth angle from the first trajectory point to each vertex are numerically compared to obtain two adjacent vertices within the numerical range of the geodetic azimuth angle, and the edge formed by the two adjacent vertices is taken as the outgoing edge. Based on the topological relationship of the global hexagonal grid, find the unique neighboring grid that shares the outgoing edge with the first starting grid, and use the neighboring grid as the second starting grid. Based on the first major azimuth angle and the major azimuth angle from the first trajectory point to each vertex of the second starting grid, repeat the steps of calculating the outgoing edge and finding the neighboring grid until the starting grid and the ending grid are the same grid.

5. The data analysis-based ship navigation management method as described in claim 2, characterized in that, In step S2, the specific process of obtaining the heat level of the grid includes: Traverse all the grid coding sequences generated by historical trajectories, accumulate the total number of times each grid code appears in all sequences, define the total number of times as the density potential energy of the grid, and use the frequency as the density potential energy; Based on the total time taken for each historical trajectory and the number of grids traversed, the average dwell time of the ship in each grid is calculated, and the time potential energy of the grid is accumulated. The heat coefficient of each hexagonal grid is calculated using a weighted formula. If the heat coefficient is greater than or equal to the preset high potential energy threshold, the heat level of the hexagonal grid is determined to be red. If the heat coefficient is less than the preset high potential energy threshold, the heat level of the hexagonal grid will be determined as green.

6. The data analysis-based ship navigation management method as described in claim 5, characterized in that, In step S2, the specific process of constructing the state vector of the target ship includes: Obtain the target ship's real-time speed, heading, and heat level of the hexagonal grid in which the target ship is located; Calculate the comprehensive risk score of the target vessel. The mathematical expression for the comprehensive risk score is as follows: ； in, For comprehensive risk scoring, The real-time speed of the target vessel. The preset speed limit threshold for the grid where the target vessel is currently located. The Euclidean distance between the target vessel and the nearest vessel. This is the distance sensitivity attenuation coefficient. This is the grid heat level coefficient. These are the normalized weighting coefficients; Calculate the red zone intention probability of the target vessel, the mathematical expression for which is: ； in, The probability of a target vessel's intention to enter the red-level grid. The target vessel's current heading angle. The azimuth angle of the target vessel pointing towards the center point of the nearest red-level hexagonal grid. This represents the absolute value of the heading deviation. The real-time spatial distance from the target ship to the boundary of the red-level grid. The preset search radius for the intended purpose; The heat level of the grid where the target vessel is located, the comprehensive risk score, and the red zone intent probability constitute the state vector of the target vessel.

7. The data analysis-based ship channel management method as described in claim 6, characterized in that, In step S3, constructing the hierarchical waterway management strategy specifically includes: According to the preset priority order, each element in the state vector is judged in turn, and the corresponding management strategy is executed. The first priority is the comprehensive risk score, the second priority is the probability of red zone intent, and the third priority is the heat level of the hexagonal grid where the target ship is located; Extract the comprehensive risk score from the state vector of the target vessel. If the comprehensive risk score is greater than or equal to the preset emergency threshold, mark the vessel's navigation mode as emergency takeover mode. If the overall risk score is less than the preset emergency threshold, the probability of red zone intent is determined. If the probability of red zone intent is greater than or equal to the preset intent threshold, the ship's navigation mode is marked as transition smooth mode. If the overall risk score is less than the preset emergency threshold and the red zone intent probability is less than the preset intent threshold, then the heat level of the hexagonal grid where the target ship is located will be determined. If the heat level of the hexagonal grid is red, then the ship's navigation mode will be marked as red zone strict control mode. If the heat level of the hexagonal grid is green, then the ship's navigation mode will be marked as green cruise mode.

8. The data analysis-based ship channel management method as described in claim 7, characterized in that, In step S3, executing the corresponding management strategy specifically includes: If the ship's navigation mode is emergency takeover mode, a forced maneuver command is generated; if the ship's navigation mode is transition smooth mode, a gradient deceleration command is generated; if the ship's navigation mode is red zone strict control mode, a violation correction command is generated; if the ship's navigation mode is green cruise mode, a gradient acceleration command is generated.

9. The data analysis-based ship channel management method as described in claim 8, characterized in that, The forced maneuver command includes an emergency rudder angle value, the gradient deceleration command includes a target speed value sequence, the violation correction command includes insufficient spacing and speeding as violation types, and the gradient acceleration command includes a suggested speed.

10. A data analysis-based ship navigation management system, applied in the data analysis-based ship navigation management method as described in any one of claims 1-9, characterized in that, It includes a grid construction and heat labeling module, a state vector construction module, and a hierarchical control strategy execution module; The grid construction and heat labeling module is used to construct a global hexagonal discrete grid structure. Based on historical operation data, it obtains the grid sequence of each historical operation data through a large circular arc gridding algorithm, calculates the heat coefficient of the grid, and determines the heat level of the grid based on the heat coefficient of the grid. The state vector construction module is used to obtain the grid where the target ship is located and the heat level of the grid, calculate the comprehensive risk score and red zone intent probability of the target ship, and construct the state vector of the target ship. The hierarchical control strategy execution module is used to establish hierarchical waterway management strategies, judge each indicator in the state vector in a preset priority order, and execute the corresponding management strategies.

Images