AIS data-based sea wave disaster defensive area demarcation method, medium and system

Abstract

The invention provides a sea wave disaster defensive area demarcation method based on AIS data, a medium and a system, and belongs to the technical field of sea wave disaster defensive area demarcation The method comprises the following steps: firstly, establishing an initial matrix through grid division, carrying out statistics on ship traffic conditions of each grid for many years of AIS data, and constructing a multi-dimensional traffic frequency tensor and a first change matrix; meanwhile, collecting sea wave reanalysis data to calculate occurrence frequencies of different levels of wave heights, and establishing a sea wave risk degree matrix; constructing a composite disaster index model based on ship response characteristics, calculating a grid aggregation degree and dispersion, and generating a second change matrix by applying a risk propagation function; fusing ship and sea wave data to construct a risk assessment hypergraph, and performing region marking through a multi-criterion threshold and a graph coloring algorithm to form a third change matrix; and finally, optimizing the boundary of the defensive area by adopting an improved particle swarm algorithm, and outputting a final demarcation result, thereby realizing accurate and efficient demarcation of the ocean wave disaster defensive area.

Description

Technical Field

[0001] The present invention belongs to the technical field of wave disaster defense zone delineation, and in particular relates to a wave disaster defense zone delineation method, medium, and system based on AIS data. Background Art

[0002] Wave disasters are one of the main natural factors affecting maritime navigation safety. Accurately delineating wave disaster defense zones is of great significance to ensuring the safety of ship navigation. Traditional wave disaster defense zone delineation mainly relies on static ocean hydrological data and historical disaster records, and uses statistical models or numerical simulation methods to determine risk areas. This type of method usually establishes a wave disaster risk distribution map based on a fixed geographic grid and divides defense areas according to risk levels; however, traditional delineation methods mainly rely on a single data source and lack a comprehensive consideration of actual ship navigation behavior and wave hazard levels. Existing technologies often treat ship activities and wave hazard levels as independent factors, ignoring the interaction between the two, resulting in a misalignment between the delineated defense zones and the actual ship activity areas. At the same time, static delineation methods are difficult to respond to seasonal changes and extreme weather events, and cannot provide accurate dynamic defense guidance.

[0003] Especially in sea areas with high ship density and complex shipping routes, existing technologies have difficulty in effectively integrating massive amounts of AIS data with multi-dimensional wave information, and are even more unable to establish a scientific risk assessment system and optimize boundary demarcation algorithms based on the fused data, resulting in inefficient allocation of defense resources and limited emergency response capabilities. Therefore, how to achieve the dynamic fusion of ship navigation data and wave hazard levels and improve the accuracy and adaptability of defense zone demarcation has become a technical problem that needs to be solved urgently. In other words, the existing technology has a technical problem in that the demarcation of wave disaster defense zones lacks the effective fusion of ship navigation data and wave hazard levels, resulting in the regional demarcation not being suitable for the complex environment of actual sea areas. Summary of the Invention

[0004] In view of this, the present invention provides a method, medium and system for demarcating wave disaster defense zones based on AIS data, which can solve the technical problem in the existing technology that the demarcation of wave disaster defense zones lacks effective integration of ship navigation data and wave hazard levels, resulting in the area division being unsuitable for the complex environment of actual sea areas.

[0005] The present invention is implemented as follows: In a first aspect, the present invention provides a method for delineating wave disaster defense zones based on AIS data, comprising: dividing the target sea area into regular grids and establishing an initial division matrix; collecting ship AIS data and constructing a ship type division matrix; counting the number of times different types of ships pass through each grid, constructing a multidimensional passage frequency tensor and a first change matrix; collecting historical wave reanalysis data, and constructing a wave hazard degree division matrix; constructing a composite wave disaster hazard index model, calculating the aggregation and dispersion of each grid, and applying a risk propagation function to generate a second change matrix; fusing the ship passage frequency tensor and the wave hazard index, and applying a group of risk comprehensive assessment equations to construct a dynamic risk assessment hypergraph; setting multi-criteria dynamic thresholds to mark risk areas, and constructing a third change matrix; using an improved particle swarm algorithm to dynamically iteratively optimize the defense zone boundaries, and outputting a final wave disaster key defense zone delineation matrix.

[0006] Among them, in the step of dividing the target sea area into a regular grid and establishing an initial division matrix, the initial division matrix refers to a two-dimensional array formed by dividing the target sea area into a regular grid according to an accuracy of 0.25 degrees longitude and 0.25 degrees latitude, and each element corresponds to a grid unit of a geographical location; the step of collecting ship AIS data and constructing a ship type division matrix specifically involves collecting and preprocessing ship AIS data from recent years (optionally 3-10 years, preferably 5 years), classifying ships according to their length, and constructing a ship type division matrix, wherein the ship type division matrix refers to a classification matrix formed by classifying all ships according to ship length characteristics into four categories: less than 50 meters, 50 to 100 meters, 100 to 150 meters, and greater than or equal to 150 meters, and is used to represent the distribution of different ship types in each grid.

[0007] Among them, in the step of counting the number of times different types of ships pass through each grid and constructing the multidimensional passage frequency tensor and the first change matrix, the contribution of ship AIS data to the parameters of each sub-area and the correlation between adjacent sub-areas are calculated, where the contribution of ship AIS data to the parameters of each sub-area refers to the degree of influence of a single ship AIS data point on the calculation of relevant parameters of the grid where it is located, including quantitative indicators such as passage frequency and navigation density.

[0008] Among them, historical wave reanalysis data are collected, specifically the historical wave reanalysis data of the past 40 years. In the step of constructing the wave hazard division matrix, the annual average occurrence number of level I, level II, level III, and level IV wave heights at each grid point is calculated, and a weighted formula is used to construct the wave hazard division matrix. This matrix refers to a hazard assessment matrix constructed by calculating the annual average occurrence number of wave heights of different levels at each grid point, reflecting the degree of wave hazard in each area.

[0009] Among them, in the steps of constructing a composite wave disaster risk index model, calculating the aggregation and dispersion of each grid, and applying the risk propagation function to generate the second change matrix, the aggregation degree refers to the density of ship trajectory points in the grid, and the calculation method is the number of trajectory points per unit area divided by the regional average; the dispersion degree refers to the degree of dispersion of the ship navigation direction in the grid, and the calculation method is the standard deviation of the ship heading in the grid divided by the mean standard deviation of the heading in the entire area.

[0010] Among them, the risk propagation function is used to simulate the propagation effect of wave disaster risks between adjacent areas. The input includes the risk level of the source area, the geographical distance between regions, the wave propagation speed, the complexity of the seabed terrain and the wind field influence factor. The output is the intensity coefficient of the risk propagation from the source area to the target area.

[0011] Among them, the step of integrating the ship passage frequency tensor and the wave hazard index and applying the risk comprehensive assessment equation group to construct a dynamic risk assessment hypergraph is carried out. The risk comprehensive assessment equation group includes risk quantification equation, resource allocation equation, benefit assessment equation and time constraint equation, which are used to evaluate the comprehensive risk index of each grid point, the defense resource allocation ratio, the comprehensive benefit value of the defense plan and the time feasibility index of the plan.

[0012] Among them, a multi-criteria dynamic threshold is set to mark the risk area, and in the step of constructing the third change matrix, a graph coloring problem optimization algorithm is applied to perform differentiated marking.

[0013] A second aspect of the present invention provides a computer-readable storage medium having program instructions stored therein. When the program instructions are executed in a computer, the program instructions are used to execute the above-mentioned method for demarcating wave disaster protection zones based on AIS data.

[0014] The third aspect of the present invention provides a wave disaster protection zone demarcation system based on AIS data, which includes the above-mentioned computer-readable storage medium. The system is any one of a computer, a server, and a single-chip microcomputer. The computer-readable storage medium is arranged in the system, and the system is provided with a microprocessor that executes the program instructions stored in the computer-readable storage medium.

[0015] This paper organically integrates ship navigation characteristics with wave hazard levels by constructing a multidimensional data fusion model and risk assessment system. This method not only considers the differences in wave tolerance among different types of ships but also analyzes the temporal and spatial distribution of ship activity patterns and wave hazards. This creates a dynamic risk assessment hypergraph structure, providing comprehensive data support for defense zone delineation.

[0016] Compared to traditional methods, this invention introduces a ship type classification matrix, a wave hazard classification matrix, and a multi-dimensional change matrix, significantly improving the accuracy of defense zone delineation. By quantifying the convergence and dispersion of ship trajectories and combining them with a risk propagation function to simulate the spatial propagation characteristics of wave disaster risk, the problem of blurred defense zone boundaries is resolved. Furthermore, using a comprehensive risk assessment equation system and a graph coloring optimization algorithm, the method achieves scientific allocation of defense resources and differentiated labeling of regional divisions.

[0017] Through multi-step iterative optimization, the present invention successfully solves the technical problem in the existing technology that the delineation of wave disaster defense zones lacks effective integration of ship navigation data and wave hazard levels, resulting in the regional division being unsuitable for the complex environment of actual sea areas. It significantly improves the accuracy and practicality of the delineation results, provides a more reliable decision-making basis for maritime safety management, and effectively reduces the potential risks of wave disasters to ship navigation. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 is a flow chart of the method of the present invention.

[0019] Figure 2 This is a risk propagation attenuation curve diagram for high-risk grids in Example 2.

[0020] Figure 3 This is the colored result diagram of the figure in Example 2. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0022] like Figure 1 FIG. 1 is a flow chart of a method for delineating a wave disaster defense zone based on AIS data according to the first aspect of the present invention. The method comprises the following steps: S01. Divide the target sea area into a regular grid at 0.25 degrees longitude and 0.25 degrees latitude, establish an initial division matrix and perform coordinate mapping to ensure the integrity and continuity of the grid division; S02. Collect and pre-process AIS data of ships from recent years, clean outliers and missing information, and classify ships into four categories based on their length: less than 50 meters, 50 to 100 meters, 100 to 150 meters, and greater than or equal to 150 meters, to construct the first ship type classification matrix; S03. Count the number of different types of ships passing through each grid in recent years, calculate the contribution of ship AIS data to each sub-area parameter and the correlation between adjacent sub-areas, and construct a multi-dimensional passage frequency tensor and the first change matrix; S04. Collect historical wave reanalysis data, calculate the annual average occurrence of wave heights of levels I, II, III, and IV at each grid point, and construct a second wave hazard classification matrix using a weighted formula; S05. Based on the wave data and ship response characteristics, a composite wave disaster risk index model is constructed, the aggregation and dispersion of each grid are calculated, and the risk propagation function is applied to generate a second change matrix; S06. Integrate the ship traffic frequency tensor and the wave hazard index, apply the risk comprehensive assessment equation group to conduct risk assessment, and construct a dynamic risk assessment hypergraph; S07. Set a multi-criteria dynamic threshold, perform probabilistic cluster analysis on the risk assessment hypergraph, apply a graph coloring problem optimization algorithm to differentially mark adjacent areas that satisfy Pi>mean(PZ) and are at risk level I or II, and construct a third change matrix; S08. Use the improved particle swarm algorithm to dynamically iteratively optimize the defense zone boundary, minimize the overall disaster risk in the area, and output the final wave disaster key defense zone delineation matrix.

[0023] The initial division matrix refers to a two-dimensional array formed by dividing the target sea area into regular grids with an accuracy of 0.25 degrees longitude and 0.25 degrees latitude, and each element corresponds to a grid unit of a geographical location.

[0024] The first ship type classification matrix refers to the classification matrix formed after all ships are divided into four categories according to the ship length characteristics, which is used to represent the distribution of different ship types in each grid.

[0025] Among them, the second wave hazard classification matrix refers to the hazard assessment matrix constructed by calculating the annual average number of occurrences of different levels of wave heights at each grid point, reflecting the degree of wave hazard in each area.

[0026] Among them, the final wave disaster key defense area delineation matrix refers to the wave disaster key defense area range matrix determined after multi-step calculation and optimization, which identifies the sea area grids that need key defense.

[0027] Among them, the first change matrix refers to the matrix formed after adjusting the initial division matrix based on the statistical results of AIS data, which reflects the impact of ship passage frequency on the demarcation of defense zones.

[0028] Among them, the second change matrix refers to the matrix formed by adjusting the grid based on the wave hazard assessment, which reflects the impact of wave hazard factors on the delineation of defense areas.

[0029] Among them, the third change matrix refers to the matrix formed by adjusting the dynamic risk assessment hypergraph after multi-criteria analysis to further refine the boundaries of the defense zone.

[0030] The contribution of ship AIS data to the parameters of each sub-area refers to the degree of influence of a single ship AIS data point on the calculation of relevant parameters of the grid in which it is located, including quantitative indicators such as passage frequency and navigation density.

[0031] The correlation degree of ship AIS data to each adjacent sub-area refers to the potential impact of ship activities in a certain grid on the safety of ships in the surrounding grids, which is calculated by analyzing the continuity of ship trajectories.

[0032] The degree of aggregation refers to the density of ship trajectory points within the grid, reflecting the frequency of ship activities in the area. It is calculated by dividing the number of trajectory points per unit area by the regional average.

[0033] The dispersion refers to the degree of dispersion of the ship navigation directions within the grid, reflecting the complexity of ship navigation in the area. It is calculated by dividing the standard deviation of the ship headings within the grid by the mean standard deviation of the headings in the entire area.

[0034] The risk propagation function simulates the propagation of wave disaster risk between adjacent regions. Inputs include the source region's risk level, interregional geographic distance, wave propagation speed, seabed topography complexity, and wind field influencing factors. The output is the intensity coefficient of risk propagation from the source region to the target region. The source region's risk level is derived from the risk index value in the second wave hazard partitioning matrix; the interregional geographic distance is derived from the geographic location calculation between grids in the initial partitioning matrix; the wave propagation speed is derived from the wave velocity parameters in historical wave reanalysis data; the seabed topography complexity is derived from the calculation of the fluctuation rate of sea bathymetric data; the wind field influencing factor is derived from the coupling analysis of regional wind field reanalysis data and wave direction; and the propagation intensity coefficient is used to adjust the risk assessment results of the target region in the second change matrix to ensure that the risk assessment considers the dynamic correlation between regions.

[0035] Among them, the comprehensive risk assessment equation group includes risk quantification equation, resource allocation equation, benefit assessment equation and time constraint equation.

[0036] Among them, the risk quantification equation is used to calculate the comprehensive risk index of each grid point. The input includes wave hazard index, ship passage frequency, ship type weight, historical disaster records and seasonal factors, and the output is the normalized risk index of the grid point; the wave hazard index comes from the second wave hazard classification matrix; the ship passage frequency comes from the multidimensional passage frequency tensor; the ship type weight comes from the analysis of the wind and wave resistance of different ship types in the first ship type classification matrix; the historical disaster records come from the historical maritime accident database provided by the maritime department; the seasonal factor comes from the statistical analysis of wave characteristics in different seasons; the normalized risk index is used to construct a dynamic risk assessment hypergraph.

[0037] Among them, the resource allocation equation is used to determine the proportion of defense resource allocation required for areas with different risk levels. The inputs include risk index, regional area, total defense resources, regional importance weight and adjacent area resource density, and the output is the resource allocation coefficient of each area; the risk index comes from the output result of the risk quantification equation; the regional area comes from the actual geographical area of the grid cells in the initial division matrix; the total defense resources come from the disaster prevention resource database provided by the maritime administration; the regional importance weight comes from a comprehensive analysis of economic zoning and shipping density; the adjacent area resource density comes from the resource allocation of adjacent areas in the third change matrix; the resource allocation coefficient is used to guide the resource deployment strategy in the final wave disaster key defense area demarcation matrix.

[0038] Among them, the benefit evaluation equation is used to evaluate the comprehensive benefits of the defense zone delineation plan. The inputs include risk coverage, resource utilization efficiency, monitoring point distribution uniformity, response time and defense cost, and the output is the comprehensive benefit value of the defense plan; the risk coverage comes from the demarcation range of the high-risk area in the dynamic risk assessment hypermap; the resource utilization efficiency comes from the matching degree between the resource allocation result of the resource allocation equation and the actual demand; the monitoring point distribution uniformity comes from the spatial distribution of the monitoring stations and the spatial overlap of the high-risk area; the response time comes from the time estimate from the triggering of the early warning to the implementation of the defense measures; the defense cost comes from the cost accounting database of various defense measures; the comprehensive benefit value is used to evaluate the optimization effect of the third change matrix.

[0039] Among them, the time constraint equation is used to evaluate the time feasibility of the implementation of the defense plan. The input includes defense resource allocation time, extreme weather response time, warning propagation time, ship avoidance time and regional navigation density, and the output is the time feasibility index of the plan; the defense resource allocation time is derived from the geographical distribution and scheduling capacity evaluation of defense resources; the extreme weather response time is derived from the response time statistics of historical extreme weather events; the warning propagation time is derived from the coverage and propagation efficiency evaluation of the warning information system; the ship avoidance time is derived from the maneuverability analysis of different types of ships in AIS data; the regional navigation density is derived from the density analysis of the multidimensional traffic frequency tensor; the time feasibility index is used to evaluate the time effectiveness of the defense zone division in the third change matrix.

[0040] The graph coloring problem optimization algorithm refers to a mathematical method for differentially labeling adjacent regions, ensuring that any two adjacent high-risk areas have different defense strategy types, thereby optimizing resource allocation and defense efficiency. The dynamic risk assessment hypergraph refers to a hypergraph data structure constructed by integrating multidimensional risk factors, used to express complex risk correlations between regions. The multidimensional traffic frequency tensor refers to a high-dimensional data structure that includes time, space, and ship type dimensions and is used to represent ship traffic patterns. The multi-criteria dynamic threshold refers to a risk identification standard that is dynamically adjusted based on the characteristics of the risk distribution, including a set of critical values for multiple assessment dimensions.

[0041] Among them, historical wave reanalysis data refers to a long-term wave parameter dataset generated by fusing multiple source data such as numerical simulation, satellite remote sensing, and measured wave height, which contains spatiotemporal distribution information of key indicators such as wave height, wave direction, and wave period. This data combines discrete observation point data with wave numerical models through assimilation technology to generate a continuous gridded historical wave state record covering the target sea area. Currently, many organizations provide historical wave reanalysis data, such as the wave database of the Marine Warning and Monitoring Center of the Ministry of Natural Resources, the ERA5 reanalysis data of ECMWF (European Centre for Medium-Range Weather Forecasts), which provides global wave parameters including significant wave height, wave direction, and wave period, the CFSR and NCEP / NCAR reanalysis datasets of NCEP (National Centers for Environmental Prediction of the United States), the JRA-55 reanalysis data of MA (Japan Meteorological Agency), and so on.

[0042] The specific implementation of the above steps is described in detail below.

[0043] The specific implementation of step S01 involves dividing the target sea area into regular grid cells with an accuracy of 0.25 degrees longitude and 0.25 degrees latitude, achieving initial gridding and coordinate mapping. First, using geographic information system software, the boundary coordinates of the target sea area, including the northernmost latitude, southernmost latitude, easternmost longitude, and westernmost longitude, are obtained, and the longitude and latitude spans of the target sea area are calculated. Then, based on the 0.25-degree accuracy requirement, the number of grid rows and columns is calculated, and a two-dimensional array is constructed as the initial partitioning matrix. Each grid cell is then assigned a unique identification code to ensure the continuity and uniqueness of the grid numbering. A bidirectional mapping relationship between geographic coordinates and grid indexes is then established. Using longitude and latitude, grid locations can be quickly located, and using grid indexes, geographic locations can be quickly retrieved. Finally, the integrity of the grid division is verified to ensure complete coverage of the target sea area and proper boundary processing. This step utilizes the regular grid method within the spatial partitioning algorithm. The grid size of 0.25 degrees × 0.25 degrees is selected based on the commonly used resolution standard for meteorological and oceanographic data, which both meets accuracy requirements and avoids excessive consumption of computing resources.

[0044] The specific implementation of step S02 is to collect and pre-process ship AIS data to achieve data cleaning and classification. First, obtain AIS data of ships in the target sea area for recent years from the maritime administration or commercial data providers, including information such as MMSI code, ship type, length, speed, course, timestamp, and geographic location. Then, perform data cleaning, delete obviously incorrect geographic location records, such as points on land, points outside the target sea area, records with abnormal speed, duplicate records, etc., and use linear interpolation to fill in missing data within small time periods. Then, detect and mark track breaks caused by changes in the AIS device's on-off state. Continuous data loss for more than 30 minutes is considered a new voyage. Then, ships are divided into four categories according to their length: less than 50 meters is marked as Category 1, 50 to 100 meters is marked as Category 2, 100 to 150 meters is marked as Category 3, and 150 meters or more is marked as Category 4, thus constructing a first ship type classification matrix. This step utilizes anomaly detection algorithms and hierarchical classification methods from data preprocessing. Classification by length reflects the differences in the wind and wave resistance of ships of different sizes, providing a basis for subsequent risk assessment. The length classification threshold is based on the International Maritime Organization's commonly used ship size classification standards, while also taking into account the balanced distribution of the number of ships.

[0045] The specific implementation of step S03 involves statistically analyzing vessel traffic and constructing a multidimensional data structure. First, the preprocessed AIS data points are mapped to the grid cells divided in step S01. The grid cells where each AIS data point resides are determined based on geographic location. The number of vessel passages of different types within each grid cell over recent years is then counted, with each vessel's entry and exit from the grid counted as one passage. The contribution of vessel AIS data to each sub-area parameter is then calculated using a decaying weight model, where the closer the AIS point is to the grid center, the greater its contribution to that grid. Weights are calculated using a Gaussian decay function. The coherence of vessel trajectories is then analyzed, and the correlation between adjacent sub-areas is calculated. The higher the frequency of vessel movement between adjacent grid cells, the greater the correlation. The correlation threshold is set to 0.3. Finally, a multidimensional traffic frequency tensor is constructed, containing information on spatial location, vessel type, and time dimensions. Based on the AIS data statistics, the initial partitioning matrix is adjusted to generate a first modified matrix. This step utilizes spatial statistical analysis and tensor construction techniques. The use of a Gaussian decay function to calculate contribution conforms to the objective law that spatial influences decay with distance, while the tensor structure effectively preserves the multidimensional nature of the data.

[0046] The specific implementation method of step S04 is to process the wave data and evaluate the hazard. First, collect the wave reanalysis data of the target sea area in recent years, including parameters such as significant wave height, wave direction, and wave period. Then, classify the waves according to the significant wave height, with level I being greater than or equal to 3.5 meters, level II being 2.5 to 3.5 meters, level III being 1.5 to 2.5 meters, and level IV being less than 1.5 meters. Then count the annual average number of occurrences of each level of wave height at each grid point, and the calculation method is the total number of occurrences of a certain level of wave height divided by the number of statistical years. Then use the weighted formula to construct the second wave hazard classification matrix, and the formula is hazard index = ,in to are the annual average occurrences of wave heights from I to IV, to The corresponding weights are 1.0, 0.7, 0.4, and 0.1, respectively. Finally, the hazard index is normalized and divided into risk levels: a hazard index greater than 0.7 is considered high-risk, 0.4 to 0.7 is medium-risk, and less than 0.4 is considered low-risk. This step uses a multi-level weighted assessment method. The wave height classification thresholds are based on the World Meteorological Organization's sea state classification standards. The weighting reflects the varying degrees of impact of waves of different levels on ship safety. Higher wave heights are assigned higher weights, reflecting the higher risk of extreme sea conditions.

[0047] The specific implementation method of step S05 is to construct a composite wave disaster risk index model. First, analyze the response characteristics of different types of ships to waves, and establish a sensitivity matrix between ship type and wave hazard. Small ships have a larger sensitivity coefficient to high waves, and large ships have a relatively small sensitivity coefficient. Then calculate the aggregation degree of each grid, that is, the number of trajectory points per unit area divided by the regional average. A high aggregation degree indicates frequent ship activities. Then calculate the discreteness of each grid, that is, the standard deviation of the ship heading in the grid divided by the mean standard deviation of the heading in the entire area. A high discreteness indicates a high complexity of ship navigation. Then construct a risk propagation function to simulate the propagation effect of wave disaster risk between adjacent areas. The function form is risk propagation coefficient = ,in is the source area risk level, is the distance between regions, is the characteristic distance parameter, is the complexity of the seabed topography, is the wind field influencing factor, 、 、 are empirical coefficients, with values of 1.2, 0.8, and 0.6, respectively. Finally, the risk propagation function is applied to generate the second change matrix. This step employs a risk propagation model, combining ship response characteristics and regional environmental factors to achieve dynamic and relevant risk assessment. The exponential decay term reflects the natural law of risk decay with distance.

[0048] The specific implementation of step S06 is to fuse multi-source data for comprehensive risk assessment. First, the ship traffic frequency tensor and the wave hazard index are fused, and the weighted summation method is used to calculate the comprehensive risk value. Then, a detailed risk assessment is performed using the risk comprehensive assessment equation group, which includes four sub-equations: risk quantification equation, resource allocation equation, benefit assessment equation, and time constraint equation. The risk quantification equation is then used to calculate the comprehensive risk index of each grid point, where risk index = ,in is the wave hazard index, is the frequency of ship traffic, is the ship type weight, For historical disaster records, is the seasonal factor, to and is the weight coefficient. A dynamic risk assessment hypergraph is then constructed to establish complex risk associations between grids. Each hyperedge in the hypergraph connects multiple grid points with high risk associations. Finally, the risk probability distribution is calculated to determine the risk level classification threshold. A risk index greater than 0.75 is considered Level I risk, 0.5 to 0.75 is considered Level II risk, 0.3 to 0.5 is considered Level III risk, and less than 0.3 is considered Level IV risk. This step applies hypergraph theory and multi-equation coupling evaluation methods. The hypergraph structure can effectively express complex many-to-many relationships. The equation group design takes into account the four dimensions of risk, resources, benefits, and time, achieving a comprehensive assessment.

[0049] The specific implementation of step S07 is to perform cluster analysis and label optimization of risk areas. First, set multi-criteria dynamic thresholds, including risk index threshold, cluster distance threshold, regional area threshold, etc. The threshold values are dynamically adjusted according to the data distribution characteristics. Then, perform probabilistic cluster analysis on the risk assessment hypergraph, use the improved k-means clustering algorithm to identify the risk distribution pattern, and select the regional point with the highest risk index as the initial cluster center. Then calculate the risk probability of each area. and regional average risk probability relationship, filter to meet Areas with a risk level of Level I or II are then identified using a graph coloring optimization algorithm to differentiate adjacent high-risk areas. A greedy algorithm variant is used to ensure that adjacent areas receive different defense strategy types, using at least four different marking colors. Finally, a third change matrix is constructed to further optimize the defense zone boundaries. This step utilizes probabilistic clustering and graph coloring algorithms. Probabilistic clustering identifies the spatial distribution characteristics of risk, while the graph coloring algorithm addresses the optimal allocation of defense resources, ensuring that adjacent areas utilize differentiated defense strategies and improving overall defense efficiency.

[0050] The specific implementation of step S08 is to dynamically iterate and optimize the defense zone boundary. First, define the objective function to minimize the weighted sum of the overall disaster risk and defense resource consumption in the area. Then, use the improved particle swarm algorithm to optimize the boundary. The particle position represents the boundary node coordinates, the speed represents the boundary adjustment direction and amplitude, and the global optimal position represents the optimal boundary configuration. Then set the algorithm parameters, including the number of particles 50 to 100, the maximum number of iterations 300 to 500 times, the inertia weight linearly decreasing from 0.9 to 0.4, and the acceleration constant and All are set to 2.0. Iterative optimization calculations are then performed, with particle positions and velocities updated in each iteration to evaluate the risk coverage and resource utilization efficiency of the new boundary. Finally, when the maximum number of iterations is reached or the improvement is less than 0.001 after 20 consecutive iterations, the optimization is stopped, and the final wave disaster key defense zone delineation matrix is output. This step uses an improved particle swarm optimization algorithm, which is more suitable for handling high-dimensional nonlinear problems than traditional optimization methods. The introduction of shrinkage factors and mutation operations improves the convergence and diversity of the algorithm. The linearly decreasing inertia weight balances the global search and local refinement capabilities. The acceleration constant setting ensures a balance between social learning and cognitive learning.

[0051] A second aspect of the present invention provides a computer-readable storage medium having program instructions stored therein. When the program instructions are executed in a computer, the program instructions are used to execute the above-mentioned method for demarcating wave disaster protection zones based on AIS data.

[0052] The third aspect of the present invention provides a wave disaster protection zone demarcation system based on AIS data, which includes the above-mentioned computer-readable storage medium. The system is any one of a computer, a server, and a single-chip microcomputer. The computer-readable storage medium is arranged in the system, and the system is provided with a microprocessor that executes the program instructions stored in the computer-readable storage medium.

[0053] The mathematical model or calculation process involved in the present invention is described in detail below.

[0054] In step S01, the initial partition matrix can be expressed as: ; Where, is the initial partition matrix; The first Rank The elements of the column represent the unique identification code of the corresponding grid cell; is the number of grids in the latitude direction, and the calculation method is ; is the number of grids in the longitude direction, and the calculation method is ; is the northernmost latitude of the target sea area; is the southernmost latitude of the target sea area; is the easternmost longitude of the target sea area; is the westernmost longitude of the target sea area; Represents the ceiling function.

[0055] The mapping relationship between geographic coordinates and grid index is: ; ; Where, is the latitude coordinate; is the longitude coordinate; is the grid row index; The grid column index.

[0056] In step S02, the first ship type classification matrix can be expressed as: ; Where, Partitioning matrix for first ship type; Indicates in Rank In column grid Number of ships of the type; , corresponding to four types of ships, where 1 represents a ship with a length of less than 50 meters, 2 represents a ship with a length of 50 to 100 meters, 3 represents a ship with a length of 100 to 150 meters, and 4 represents a ship with a length of 150 meters or more.

[0057] In step S03, the calculation formula of the contribution of AIS data to grid parameters is: ; Where, AIS data point Grid Contribution of AIS data point To Grid The distance from the center in nautical miles; is the standard deviation parameter of the Gaussian function. The recommended value is 1 / 4 of the grid side length, or approximately 7.5 nautical miles. The Gaussian decay function is used here to simulate the natural phenomenon that the influence of an AIS data point on the surrounding area gradually decreases with increasing distance, which conforms to the first law of spatial data.

[0058] The calculation formula of the correlation degree of adjacent sub-areas is: ; Where, For Grid With Grid The degree of correlation between From the grid To Grid Frequency of ship movements; For Grid Total number of vessel passages within the area; For Grid Total number of vessel passages within the area; For Grid With Grid distance between centers; is the distance attenuation coefficient, and a recommended value is 25 nautical miles. This formula combines the normalized ship flow frequency and the distance attenuation factor to effectively measure the correlation of ship activities between adjacent grids.

[0059] The multidimensional passing frequency tensor can be expressed as: ; Where, is the multidimensional passing frequency tensor; Indicates in Rank Column grid Type of ship Number of passes within a time unit; The total number of time units, for example, monthly statistics =60 (recent years x 12 months).

[0060] The first change matrix calculation formula is: ; Where, is the first change matrix; represents matrix Hadamard product (element-wise multiplication); For the adjustment coefficient, a value of 0.02 is recommended; For the The weight coefficients of the ships are , , , The Hadamard product is used here instead of conventional matrix multiplication to keep the matrix dimension unchanged and achieve element-by-element risk adjustment, while the weight setting reflects the higher importance of large ships in contributing to the passage frequency.

[0061] In step S04, the calculation formula of the second wave hazard division matrix is: ; ; Where, Matrix for second wave hazard classification; For the Rank The wave hazard index of the column grid has a value range of ; 、 、 、 are the annual average occurrence times of wave heights of level I, II, III, and IV at the grid point, respectively; , , , is the weight coefficient of each level of wave height; is the total number of wave height records at this grid point; The formula adopts the weighted average principle, assigning different weights to different levels of wave height according to their degree of danger, and then normalizing them to obtain the wave danger index.

[0062] In step S05, the sensitivity matrix between ship type and wave hazard is: ; Where, is the sensitivity matrix; Indicates the Class I ships The sensitivity coefficient of the wave height is obtained by analyzing historical maritime accident data and ship stability theory: , , , ; , , , ; , , , ; , , , .

[0063] The calculation formula of grid aggregation degree is: ; Where, For Grid degree of polymerization; For Grid Number of AIS data points within; For Grid The actual geographic area of the region, expressed in square nautical miles; the denominator is the average density for the entire region. The degree of aggregation reflects the density of regional ship activity; larger values indicate more frequent ship activity.

[0064] The formula for calculating grid discreteness is: ; Where, For Grid The discreteness of For Grid The standard deviation of ship headings within the region; the denominator is the average of the standard deviations of headings across the entire region. Dispersion reflects the complexity of ship navigation within the region; larger values indicate more complex navigation patterns and higher risks.

[0065] The risk propagation function expression is: ; Where, For risk from source area Towards the target area The propagation coefficient of is the source area risk level, and its value range is , comes from the second wave hazard classification matrix; is the geographical distance between the source area and the target area, in nautical miles; It is a characteristic distance parameter, and the recommended value is 30 nautical miles; is the complexity of the seabed topography, and its value range is , calculated as the coefficient of variation of seafloor depth within the region; is the wind field influencing factor, and its value range is , calculated as the cosine of the angle between the dominant wind direction and the main wave direction; 、 、 is an empirical coefficient. This function draws on the attenuation law and risk propagation theory in physics. It simulates the characteristic that risk decreases with distance through the exponential decay term, while the denominator takes into account the obstruction of terrain and wind field to propagation.

[0066] The second change matrix calculation formula is: ; Where, For the second change matrix in the target area The value of For the second wave hazard matrix in the target area The value of Target area The set of adjacent regions; For risk from source area Towards the target area The propagation coefficient of For the second wave hazard matrix in the source area The formula adopts a form similar to the diffusion equation, simulating the diffusion and propagation effects of risks in space. Controls the direction and intensity of propagation.

[0067] In step S06, the risk quantification equation is: ; Where, For Grid Comprehensive risk index; is the wave hazard index, with a value range of , comes from the second wave hazard classification matrix; is the frequency of ship passage, and after normalization, its value range is ; For Grid Neidi proportion of ships of this type; For the The weight coefficient of the ship type is recommended to be , , , ; is the historical disaster record index, with a value range of , calculated by dividing the number of historical maritime accidents in the grid by the maximum number of accidents in the entire region; is a seasonal factor, and its value range is , calculated as the risk increment of the grid during the storm season; to is the weight coefficient, the recommended value is , , , ; is the error term, which takes into account the random factors that the model fails to capture and is usually assumed to follow a normal distribution The equation comprehensively considers wave hazard, ship traffic characteristics, historical disaster records and seasonal factors, and the weights of each factor reflect the contribution of different factors to the risk.

[0068] The resource allocation equation is: ; Where, For Grid The resource allocation coefficient; For Grid Comprehensive risk index; is the power parameter of the risk index, with a recommended value of 1.5, which is used to emphasize resource allocation in high-risk areas; For Grid the actual geographical area; For Grid The importance weight of considering economic and shipping factors is in the range of ; is the collection of all grids; is the total defense resource amount; is the difference in resource density between adjacent regions, which is calculated as the standard deviation of the resource density between the current grid and the adjacent grid; is a random adjustment factor, which obeys uniform distribution , used to introduce small perturbations to avoid local optima; 、 is the adjustment coefficient. This equation adheres to the principle of "matching risk with resources" while taking into account factors such as regional size, importance, and the distribution of adjacent resources. The power design ensures that high-risk areas receive a disproportionate amount of resources.

[0069] The benefit evaluation equation is: ; Where, is the comprehensive benefit value of the defense plan; Assemble for high-risk areas; Gather for the designated defense zone; is the indicator function, when the grid If it is classified as a defense zone, it takes 1, otherwise it takes 0; is the distribution uniformity of monitoring points, and its value range is , calculated as the complement of the spatial Gini coefficient of the monitoring site; is the response time, in hours, and the value range after normalization is ; is the defense cost, and its value range after normalization is ; to is the weight coefficient, the recommended value is , , , , The first term of this equation is the risk coverage rate, the second term is the resource utilization efficiency, the third term is the uniformity of monitoring point distribution, the fourth term is the response time penalty, and the fifth term is the cost penalty. This equation comprehensively evaluates the multi-dimensional benefits of the defense solution.

[0070] The time constraint equation is: ; Where, is the time feasibility index of the scheme, and its value range is ,The larger the value, the higher the time feasibility; Defense resource deployment time, in hours; is the extreme weather response time, in hours; is the warning propagation time, in hours; The ship's avoidance time, in hours; to is the threshold value of each time dimension, and the recommended values are 6 hours, 12 hours, 1 hour, and 3 hours respectively; to is the maximum tolerance value of each time dimension, and the recommended values are 24 hours, 36 hours, 6 hours, and 12 hours respectively; is the regional navigation density, and after normalization, its value range is ; is the adjustment coefficient. This equation uses the "barrel principle" to determine overall feasibility based on the shortest link, reflecting the feasibility assessment of defense solutions under time constraints.

[0071] In step S07, the risk probability The calculation formula is: ; Where, For Grid the risk probability; For Grid Risk index; is the risk threshold, and the recommended value is 0.6; is the probability density function of the risk index; is the cumulative distribution function of the risk index. This formula calculates the probability that the grid risk index exceeds the threshold and is used to screen high-risk areas.

[0072] The formula for calculating the regional average risk probability is: ; Where, For the region The average risk probability; For the region The number of grid cells; For Grid probability of risk.

[0073] The mathematical description of the graph coloring problem is: ; ; ; Where, is the desired number of colors; is a set of grid points; is the adjacent relationship set; For Grid The problem is solved using a greedy algorithm variant, which first sorts the grids in descending order of risk index and then assigns each grid the minimum available color that is not the same as any adjacent grid.

[0074] The third change matrix calculation formula is: ; Where, is the third change matrix; is the second change matrix; represents the Hadamard product; Mark the normalized values for the colors; For Grid the risk probability; is the regional average risk probability; For Grid the risk level; is an indicator function, which takes 1 when the condition is met and 0 otherwise; 、 The formula further optimizes the defense zone boundaries through color marking and risk probability, ensuring that adjacent high-risk areas adopt differentiated defense strategies.

[0075] In step S08, the position and velocity update formula of the improved particle swarm algorithm is: ; ; ; Where, For the In the iteration Particle No. Dimensional speed; For the In the iteration Particle No. Dimensional location; For the The best historical position of the particle dimension; The first position of the global optimal dimension; is the position of another randomly selected particle dimension; For the The inertia weight of the iteration; 、 、 is the acceleration constant; 、 、 for A random number between 、 is the maximum and minimum value of the inertia weight; is the maximum number of iterations. Compared to the traditional particle swarm algorithm, this improved version adds a third term, the “social learning term”, which enables particles to learn from other randomly selected particles, enhancing the algorithm’s global search capability and diversity.

[0076] The objective function is: ; Where, is the objective function value; Gather for the designated defense zone; is the collection of all grids; For Grid Risk index; is the comprehensive benefit value of the defense plan; For Grid The resource allocation coefficient; to is the weight coefficient, the recommended value is , , , The first term of the objective function is risk coverage maximization, the second term is the defense zone area penalty, the third term is the comprehensive benefit reward, and the fourth term is the resource consumption penalty, which comprehensively evaluates the advantages and disadvantages of the defense zone demarcation scheme.

[0077] The final wave disaster key defense area delineation matrix can be expressed as: ; Where, Delineate a matrix for the final wave hazard key defense areas; is a binary variable, when Rank The value of the column grid is 1 if it is in the defense zone, and 0 otherwise.

[0078] Specifically, the present invention is based on the theory of multi-source data fusion and risk propagation models. By constructing a hierarchical data processing framework, it achieves a systematic integration of ship activity characteristics and wave hazard levels. First, an initial matrix is established using a regular grid, providing a unified spatial reference framework for subsequent analysis and ensuring data consistency. By cleaning and classifying recent years of ship AIS data, a ship classification matrix is established, categorizing ships by length into four categories to reflect the differences in their wave resistance.

[0079] During the data fusion phase, this invention relies on two core assumptions: first, the risk level in areas affected by ship activity is closely related to traffic frequency, ship type, and navigation complexity; second, the risk of wave hazards exhibits spatial propagation characteristics and is influenced by the marine environment. Based on these assumptions, a multidimensional traffic frequency tensor and risk propagation function are constructed to dynamically map the spatial distribution of risk. The calculation of aggregation and dispersion quantifies the density and complexity of ship activity, providing key parameters for risk assessment.

[0080] The theoretical core of this invention is a comprehensive risk assessment equation system, which establishes a comprehensive assessment system through four equations: risk quantification, resource allocation, benefit assessment, and time constraints. This system not only considers static risk factors but also incorporates dynamic factors such as resource allocation efficiency and response time, making defense zone delineation practical. During the boundary optimization phase, an iterative optimization algorithm based on a graph coloring problem and an improved particle swarm optimization algorithm are employed to ensure regional continuity and rational resource allocation in the delineation results.

[0081] This invention, through multi-level matrix modifications and dynamic threshold adjustments, makes defense zone delineation adaptive and can dynamically adjust to actual shipping conditions and changes in the marine environment. This data-driven adaptive delineation method breaks through the limitations of traditional static delineation, effectively solving the technical problem of insufficient integration of ship and wave data, and providing more scientific technical support for wave disaster prevention.

[0082] A specific embodiment 1 of the present invention is provided below. The specific implementation of each step in this embodiment 1 is described in detail as follows.

[0083] The specific implementation of step S01 is to divide the target sea area into regular grid cells according to the accuracy of 0.25 degrees longitude and 0.25 degrees latitude to achieve initial gridding processing and coordinate mapping. First, obtain the boundary coordinates of the target sea area, including the northernmost latitude, southernmost latitude, easternmost longitude, and westernmost longitude, and calculate the longitude span and latitude span of the target sea area. Then, according to the accuracy requirement of 0.25 degrees, calculate the number of grid rows and columns, and construct a two-dimensional array as the initial partition matrix. The matrix can be expressed as: ; Where, is the initial partition matrix; The first Rank The elements of the column represent the unique identification code of the corresponding grid cell; is the number of grids in the latitude direction, and the calculation method is ; is the number of grids in the longitude direction, and the calculation method is ; is the northernmost latitude of the target sea area; is the southernmost latitude of the target sea area; The easternmost longitude of the target sea area; is the westernmost longitude of the target sea area; Represents the ceiling function.

[0084] Then, a unique identification code is assigned to each grid cell to ensure the continuity and uniqueness of the grid number. Then, a bidirectional mapping relationship between geographic coordinates and grid index is established. The mapping relationship is: ; ; Where, is the latitude coordinate; is the longitude coordinate; is the grid row index; The grid column index.

[0085] Finally, the integrity of the grid division was verified to ensure complete coverage of the target sea area and proper boundary treatment. This step employed the regular grid method within the spatial partitioning algorithm. The grid size of 0.25 degrees by 0.25 degrees was chosen based on the commonly used resolution standard for meteorological and oceanographic data, which met accuracy requirements while avoiding excessive consumption of computing resources.

[0086] The specific implementation of step S02 is to collect and pre-process ship AIS data to achieve data cleaning and classification. First, obtain the AIS data of ships in the target sea area in recent years from the maritime administration or commercial data providers, including information such as MMSI code, ship type, length, speed, heading, timestamp and geographic location. Then, perform data cleaning to delete obviously erroneous geographic location records, such as points on land, points outside the target sea area, records of abnormal speed (over 50 knots or negative values), duplicate records, etc. Missing data is filled using linear interpolation, but only for short-term missing data within 30 minutes. Then, detect and mark the track breaks caused by changes in the switch state of the AIS device, and consider the continuous data missing time of more than 30 minutes as a new voyage. Then, divide the ships into four categories according to their length: less than 50 meters is marked as category 1, 50 to 100 meters is marked as category 2, 100 to 150 meters is marked as category 3, and greater than or equal to 150 meters is marked as category 4. Construct a first ship type classification matrix, which can be expressed as: ; Where, Partitioning the matrix for the first ship type; Indicates in Rank In column grid Number of ships of the type; , corresponding to four types of ships, where 1 represents a ship with a length of less than 50 meters, 2 represents a ship with a length of 50 to 100 meters, 3 represents a ship with a length of 100 to 150 meters, and 4 represents a ship with a length of 150 meters or more.

[0087] This step utilizes anomaly detection algorithms and hierarchical classification methods from data preprocessing. Classification by length reflects the differences in the wind and wave resistance of ships of different sizes, providing a basis for subsequent risk assessment. The length classification threshold is based on the International Maritime Organization's commonly used ship size classification standards, while also taking into account the balanced distribution of the number of ships.

[0088] The specific implementation of step S03 is to count ship traffic and construct a multidimensional data structure. First, the preprocessed AIS data points are mapped to the grid cells divided in step S01. The grid where each AIS data point is located is determined based on its geographic location. Then, the number of passages of different types of ships within each grid over recent years is counted. A single passage is counted from the time a ship enters and leaves the grid. Next, the contribution of the ship's AIS data to each sub-area parameter is calculated using a decaying weight model. The specific calculation formula is: ; Where, AIS data point Grid Contribution of AIS data point To Grid The distance from the center in nautical miles; is the standard deviation parameter of the Gaussian function. The recommended value is 1 / 4 of the grid side length, which is about 7.5 nautical miles.

[0089] Then the continuity of ship trajectories is analyzed and the correlation between adjacent sub-areas is calculated. The calculation formula is: ; Where, For Grid With Grid The degree of correlation between From the grid To Grid Frequency of ship movements; For Grid Total number of vessel passages within the area; For Grid Total number of vessel passages within the area; For Grid With Grid distance between centers; is the distance attenuation coefficient, and a recommended value is 25 nautical miles. The higher the frequency of ship movement between adjacent grids, the greater the correlation. The correlation threshold is set to 0.3.

[0090] Finally, a multidimensional frequency tensor is constructed, which can be expressed as: ; Where, is the multidimensional passing frequency tensor; Indicates in Rank Column grid Type of ship Number of passes within a time unit; The total number of time units, for example, monthly statistics =60 (recent years x 12 months).

[0091] The initial partition matrix is adjusted based on the AIS data statistical results to generate the first change matrix. The calculation formula is: ; Where, is the first change matrix; represents matrix Hadamard product (element-wise multiplication); For the adjustment coefficient, a value of 0.02 is recommended; For the The weight coefficients of the ships are , , , .

[0092] This step applies spatial statistical analysis and tensor construction technology. The contribution is calculated using the Gaussian attenuation function, which conforms to the objective law that spatial influence decays with distance. The tensor structure effectively retains the multidimensional characteristics of the data.

[0093] The specific implementation method of step S04 is to process the wave data and assess the hazard. First, collect the wave reanalysis data of the target sea area in recent years, including parameters such as significant wave height, wave direction, and wave period. Then, the waves are classified according to the significant wave height, with level I being greater than or equal to 3.5 meters, level II being 2.5 to 3.5 meters, level III being 1.5 to 2.5 meters, and level IV being less than 1.5 meters. Then, count the annual average number of occurrences of each level of wave height at each grid point, and the calculation method is to divide the total number of occurrences of a certain level of wave height by the number of statistical years. Then, use the weighted formula to construct the second wave hazard classification matrix, which can be expressed as: ; The calculation formula for each element is: ; Where, Matrix for second wave hazard classification; For the Rank The wave hazard index of the column grid has a value range of ; 、 、 、 are the annual average occurrence times of wave heights of level I, II, III, and IV at the grid point, respectively; , , , is the weight coefficient of each level of wave height; is the total number of wave height records at this grid point; is the maximum weight.

[0094] Finally, the hazard index is normalized and classified into risk levels: a hazard index greater than 0.7 is considered high risk, 0.4 to 0.7 is medium risk, and less than 0.4 is considered low risk. This step utilizes a multi-level weighted assessment method, with wave height classification thresholds based on the World Meteorological Organization's sea state classification standards. The weighting reflects the varying impacts of different wave levels on ship safety, with higher wave heights being assigned a higher weight, reflecting the increased risk associated with extreme sea conditions.

[0095] The specific implementation of step S05 is to construct a composite wave disaster risk index model. First, the response characteristics of different types of ships to waves are analyzed, and a sensitivity matrix between ship type and wave risk is established, which can be expressed as: ; Where, is the sensitivity matrix; Indicates the Class I ships The sensitivity coefficient of the wave height is obtained by analyzing historical maritime accident data and ship stability theory: , , , ; , , , ; , , , ; , , , Small ships are more sensitive to high waves, while large ships are relatively less sensitive.

[0096] Then calculate the aggregation degree of each grid, the calculation formula is: ; Where, For Grid degree of polymerization; For Grid Number of AIS data points within; For Grid The actual geographical area of the region, expressed in square nautical miles; the denominator is the average density for the entire region. A high degree of aggregation indicates frequent ship activity.

[0097] Then calculate the discreteness of each grid, the calculation formula is: ; Where, For Grid The discreteness of For Grid The standard deviation of ship headings within the region; the denominator is the average of the standard deviations of headings across the entire region. A high degree of dispersion indicates greater complexity and risk in ship navigation.

[0098] Then, a risk propagation function is constructed to simulate the propagation effect of wave disaster risk between adjacent areas. The function form is: ; Where, For risk from source area Towards the target area The propagation coefficient of is the source area risk level, and its value range is , comes from the second wave hazard classification matrix; is the geographical distance between the source area and the target area, in nautical miles; It is a characteristic distance parameter, and the recommended value is 30 nautical miles; is the complexity of the seabed topography, and its value range is , calculated as the coefficient of variation of seafloor depth within the region; is the wind field influencing factor, and its value range is , calculated as the cosine of the angle between the dominant wind direction and the main wave direction; 、 、 is the empirical coefficient.

[0099] Finally, the risk propagation function is applied to generate the second change matrix, and the calculation formula is: ; Where, For the second change matrix in the target area The value of For the second wave hazard matrix in the target area The value of Target area The set of adjacent regions; For risk from source area Towards the target area The propagation coefficient of For the second wave hazard classification matrix in the source area The value of .

[0100] This step adopts a risk propagation model, combined with ship response characteristics and regional environmental factors, to achieve dynamic and correlated risk assessment. The exponential decay term reflects the natural law of risk decay with distance.

[0101] The specific implementation of step S06 is to fuse multi-source data for comprehensive risk assessment. First, the ship traffic frequency tensor and the wave hazard index are fused, and the comprehensive risk value is calculated using the weighted summation method. Then, a detailed risk assessment is performed using the risk comprehensive assessment equation group, which includes four sub-equations: risk quantification equation, resource allocation equation, benefit assessment equation, and time constraint equation. Among them, the risk quantification equation is used to calculate the comprehensive risk index of each grid point, and the calculation formula is: ; Where, For Grid Comprehensive risk index; is the wave hazard index, with a value range of , comes from the second wave hazard classification matrix; is the frequency of ship passage, and after normalization, its value range is ; For Grid Neidi proportion of ships of this type; For the The weight coefficient of the ship type is recommended to be , , , ; is the historical disaster record index, with a value range of , calculated by dividing the number of historical maritime accidents in the grid by the maximum number of accidents in the entire region; is a seasonal factor, and its value range is , calculated as the risk increment of the grid during the storm season; to is the weight coefficient, the recommended value is , , , ; is the error term, which takes into account the random factors that the model fails to capture and is usually assumed to follow a normal distribution .

[0102] The resource allocation equation is used to determine the proportion of defense resources required for areas with different risk levels. The calculation formula is:

[0103] Where, For Grid The resource allocation coefficient; For Grid Comprehensive risk index; is the power parameter of the risk index, with a recommended value of 1.5, which is used to emphasize resource allocation in high-risk areas; For Grid the actual geographical area; For Grid The importance weight of considering economic and shipping factors is in the range of ; is the collection of all grids; is the total defense resource amount; is the difference in resource density between adjacent regions, which is calculated as the standard deviation of the resource density between the current grid and the adjacent grid; is a random adjustment factor, which obeys uniform distribution , used to introduce small perturbations to avoid local optima; 、 is the adjustment factor.

[0104] The benefit evaluation equation is used to evaluate the comprehensive benefits of the defense zone demarcation plan. The calculation formula is: ; Where, is the comprehensive benefit value of the defense plan; Assemble for high-risk areas; Gather for the designated defense zone; is the indicator function, when the grid If it is classified as a defense zone, it takes 1, otherwise it takes 0; is the distribution uniformity of monitoring points, and its value range is , calculated as the complement of the spatial Gini coefficient of the monitoring site; is the response time, in hours, and the value range after normalization is ; is the defense cost, and its value range after normalization is ; to is the weight coefficient, the recommended value is , , , , .

[0105] The time constraint equation is used to evaluate the time feasibility of implementing the defense plan. The calculation formula is: ; Where, is the time feasibility index of the scheme, and its value range is ,The larger the value, the higher the time feasibility; Defense resource deployment time, in hours; is the extreme weather response time, in hours; is the warning propagation time, in hours; The ship's avoidance time, in hours; to is the threshold value of each time dimension, and the recommended values are 6 hours, 12 hours, 1 hour, and 3 hours respectively; to is the maximum tolerance value of each time dimension, and the recommended values are 24 hours, 36 hours, 6 hours, and 12 hours respectively; is the regional navigation density, and after normalization, its value range is ; is the adjustment factor.

[0106] Finally, a dynamic risk assessment hypergraph is constructed to establish complex risk relationships between grids. Each hyperedge in the hypergraph connects multiple grid points with high risk correlation. The risk probability distribution is calculated, and the risk classification thresholds are determined. A risk index greater than 0.75 is designated as Level I, 0.5 to 0.75 as Level II, 0.3 to 0.5 as Level III, and less than 0.3 as Level IV. This step applies hypergraph theory and multi-equation coupled assessment methods. The hypergraph structure can effectively express complex many-to-many relationships. The equation system design considers the four dimensions of risk, resources, benefits, and time, achieving a comprehensive assessment.

[0107] The specific implementation of step S07 is to perform cluster analysis and label optimization of risk areas. First, multi-criteria dynamic thresholds are set, including risk index thresholds, cluster distance thresholds, and regional area thresholds. The threshold values are dynamically adjusted based on the data distribution characteristics. Then, a probabilistic cluster analysis is performed on the risk assessment hypergraph, using an improved k-means clustering algorithm to identify the risk distribution pattern. The initial cluster center is selected as the regional point with the highest risk index. The risk probability of each region is then calculated using the following formula: ; Where, For Grid the risk probability; For Grid Risk index; is the risk threshold, and the recommended value is 0.6; is the probability density function of the risk index; is the cumulative distribution function of the risk index.

[0108] The formula for calculating the regional average risk probability is: ; Where, For the region The average risk probability; For the region The number of grid cells; For Grid probability of risk.

[0109] Filter to meet And the hazard level is Class I or Class II.

[0110] Optionally, a graph coloring problem optimization algorithm is then applied to differentially mark adjacent high-risk areas. This problem is solved using a greedy algorithm variant. The grids are first sorted in descending order of risk index, and each grid is assigned the smallest available color mark that is not the same as any adjacent grid. A minimum of four different marking colors are used, and four are used by default.

[0111] Finally, the third change matrix is constructed, and the calculation formula is: ; Where, is the third change matrix; is the second change matrix; represents the Hadamard product; Mark the normalized values for the colors; For Grid the risk probability; is the regional average risk probability; For Grid the risk level; is an indicator function, which takes 1 when the condition is met and 0 otherwise; 、 is the adjustment factor.

[0112] This step uses probabilistic clustering and graph coloring algorithms. Probabilistic clustering can identify the spatial distribution characteristics of risks, while the graph coloring algorithm solves the problem of optimal allocation of defense resources, ensuring that adjacent areas adopt differentiated defense strategies and improving overall defense efficiency. Graph coloring is an optional method, and multiple colors can also be used directly.

[0113] The specific implementation of step S08 is to dynamically iterate and optimize the defense zone boundary. First, define the objective function to minimize the weighted sum of the overall disaster risk and defense resource consumption in the area. The calculation formula is: ; Where, is the objective function value; Gather for the designated defense zone; is the collection of all grids; For Grid Risk index; is the comprehensive benefit value of the defense plan; For Grid The resource allocation coefficient; to is the weight coefficient, the recommended value is , , , .

[0114] Then, an improved particle swarm algorithm is used to optimize the boundary. The particle position represents the boundary node coordinates, the velocity represents the boundary adjustment direction and amplitude, and the global optimal position represents the optimal boundary configuration. The position and velocity update formulas are: ; ; ; Where, For the In the iteration Particle No. Dimensional speed; For the In the iteration Particle No. Dimensional location; For the The best historical position of the particle dimension; The first position of the global optimal dimension; is the position of another randomly selected particle dimension; For the The inertia weight of the iteration; 、 、 is the acceleration constant; 、 、 for A random number between 、 is the maximum and minimum value of the inertia weight; is the maximum number of iterations.

[0115] Then set the algorithm parameters, including the number of particles 50 to 100, the maximum number of iterations 300 to 500 times, the inertia weight linearly decreasing from 0.9 to 0.4, and the acceleration constant and Both are set to 2.0. Then, an iterative optimization calculation is performed, updating the particle position and velocity in each iteration, and evaluating the risk coverage and resource utilization efficiency of the new boundary. Finally, when the maximum number of iterations is reached or the improvement is less than 0.001 after 20 consecutive iterations, the optimization is stopped and the final wave disaster key defense zone delineation matrix is output, which can be expressed as: ; Where, Delineate a matrix for the final wave hazard key defense areas; is a binary variable, when Rank The value of the column grid is 1 if it is in the defense zone, and 0 otherwise.

[0116] This step uses an improved particle swarm optimization algorithm, which is more suitable for dealing with high-dimensional nonlinear problems than traditional optimization methods. The introduction of shrinkage factors and mutation operations improves the convergence and diversity of the algorithm. The linearly decreasing inertia weight balances the global search and local refinement capabilities. The acceleration constant setting ensures a balance between social learning and cognitive learning.

[0117] In this embodiment, a graph coloring algorithm is selected. The graph coloring algorithm is a key optimization technology, which is mainly used to differentiate the high-risk areas in step S07 to achieve efficient configuration of defense resources and optimization of risk management strategies. The following is a detailed explanation of the application of the graph coloring algorithm in this embodiment 1: The graph coloring algorithm is derived from graph theory. Its basic idea is to assign a color to each vertex of the graph so that any adjacent vertices have different colors, while using the least number of colors possible. In the present invention, each grid unit is regarded as a vertex in the graph, and the relationship between adjacent grids constitutes an edge; in this embodiment, the key to the graph coloring algorithm is to regard different defense strategies as different colors, and to avoid duplicate resource configuration and mutual interference in defense effectiveness by assigning different defense strategies to adjacent high-risk areas.

[0118] The specific steps include: Step 1: Identify risk areas: First, select the areas that meet the risk probability And the high-risk areas with hazard level I or II constitute the vertex set that needs to be colored in the graph.

[0119] Step 2: Adjacent relationship construction: For any two high-risk grids and , if they are geographically adjacent (share a border), then add an edge to the graph , forming an adjacent relationship set .

[0120] Step 3: Risk prioritization: sort all high-risk grids by risk index Sort in descending order, ensuring that areas with the highest risk are assigned defense strategies first.

[0121] Step 4, Greedy Coloring Algorithm: Use a greedy algorithm variant for coloring. The specific steps are as follows: traverse each grid in descending order of risk index; for the current grid, check the colors assigned to all its adjacent grids; select the smallest color that is not used by adjacent grids for assignment; if all used colors are occupied by adjacent grids, introduce a new color.

[0122] Step 5, color optimization: further optimize the color assignment through local search and color swap operations to minimize the number of required colors. Usually at least 4 different marker colors are required to meet the planar graph coloring requirements.

[0123] Step 6: Defense strategy mapping: Map different colors to different defense strategy types, for example: Color 1: Strengthen the deployment of early warning monitoring systems; Color 2: Increase the patrol frequency of rescue vessels; Color 3: Set up temporary sheltered anchorages; Color 4: Strengthen the communication support system.

[0124] The graph coloring algorithm produced significant results in this implementation: Differentiated defense resource allocation: By ensuring that adjacent high-risk areas adopt different defense strategies, duplicate resource investment and redundant allocation are avoided, improving resource utilization efficiency. Maximized synergy: Different defense strategies can complement and synergize with each other. For example, one region can focus on early warning system deployment, while adjacent regions strengthen rescue forces, forming a complete defense system. Flexible response to diverse risks: Different defense strategies are optimized for different risk characteristics through the graph coloring algorithm, enabling the defense system to more flexibly respond to complex and changing marine disaster situations.

[0125] In addition, the boundary optimization basis: the result of graph coloring directly affects the construction of the third change matrix, through the formula: Integrating color marking information into the risk assessment system provides a mathematical basis for subsequent boundary optimization.

[0126] Improved computational efficiency: Compared to traditional exact algorithms such as backtracking, the greedy algorithm variant has higher computational efficiency and can quickly obtain approximate optimal solutions in complex large-scale grid partitioning. In this embodiment, the actual computation time is reduced by up to 25%.

[0127] In general, the graph coloring algorithm in this invention realizes the intelligent differentiated configuration of defense strategies in high-risk areas, provides important theoretical support and technical means for the scientific delineation of wave disaster defense areas, and significantly improves the overall effectiveness and resource utilization efficiency of the defense system.

[0128] Through the above eight steps, this embodiment implements a method for delineating wave disaster defense zones based on AIS data. This method comprehensively considers multi-dimensional information such as ship traffic characteristics, wave hazard levels, historical disaster records, and seasonal factors. Through multi-level matrix transformation and multi-equation coupled evaluation, it constructs a complete risk assessment and defense zone delineation framework, which can effectively guide wave disaster prevention efforts and improve maritime safety. The core innovation of this method lies in the organic combination of ship AIS data and wave hazard levels, the spatial correlation of risk assessment achieved through a risk propagation model, and the efficient allocation of defense resources achieved through a graph coloring algorithm and improved particle swarm optimization. Compared with traditional methods, this method is more targeted and practical.

[0129] To better understand and implement the present invention, Example 2 of a specific application scenario of the present invention is provided below: Researchers conducted a study on the demarcation of wave disaster defense zones in a certain sea area (19°N-23°N, 110°E-117°E). This sea area is an important shipping channel with a large number of ships passing through it every year. At the same time, typhoons are frequent in this area, and the risk of wave disasters is high. First, the target sea area is divided into 16×28=448 grid cells with an accuracy of 0.25 degrees longitude and 0.25 degrees latitude, and the initial division matrix is constructed. , and establish a mapping relationship between geographic coordinates and grid index.

[0130] We collected ship AIS data from January 2018 to December 2022, totaling approximately 127 million records. After data cleaning and removal of outliers and missing information, we obtained approximately 114 million valid records. Vessels were categorized into four categories based on their length: 18,645 vessels were less than 50 meters (Category 1), 12,367 vessels were between 50 and 100 meters (Category 2), 5,826 vessels were between 100 and 150 meters (Category 3), and 2,134 vessels were 150 meters or longer (Category 4). The distribution of different types of vessels is shown in Table 1: Table 1 Statistics of the distribution of different types of ships in the main grid

[0131] The number of passages of different types of ships in each grid was counted, and the contribution of AIS data to the parameters of each sub-area was calculated. The value is 7.5 nautical miles. In the calculation of the correlation between adjacent sub-areas, the distance attenuation coefficient The value is set to 25 nautical miles, and the correlation threshold is set to 0.3. A multidimensional frequency tensor is constructed. ,This tensor contains information in three dimensions: spatial location, ship type, and time (by month).

[0132] We collected wave reanalysis data from 1982 to 2022 for the measurement area and calculated the annual average occurrence of wave heights of Class I (>= 3.5 meters), Class II (2.5-3.5 meters), Class III (1.5-2.5 meters), and Class IV (<1.5 meters) at each grid point. The wave classification statistics for some high-risk areas are shown in Table 2: Table 2 Statistics of the average annual occurrence of wave levels in some high-risk areas

[0133] Constructing the second wave hazard classification matrix using weighted formula , the weight coefficients are 、 、 、 The wave hazard index of each grid was calculated, and 32 grids were in high-risk areas with a hazard index greater than 0.7, accounting for 7.1% of the total number of grids. These areas are mainly distributed in areas with frequent typhoon activity.

[0134] According to the ship type and wave response characteristics, a sensitivity matrix was constructed Small ships have a greater sensitivity to high waves. For example, the sensitivity coefficient of a Class 1 ship to Level I wave height is 1.0, while the sensitivity coefficient of a Class 4 ship to Level I wave height is only 0.4.

[0135] The aggregation and dispersion of each grid were calculated. Areas with high aggregation are mainly concentrated at the intersection of channels. For example, the aggregation degree of the main channel area in the northern part of the sea area is as high as 2.85, while areas with high dispersion are mainly located in areas with complex maritime traffic, with the highest dispersion reaching 1.94.

[0136] The risk propagation function is used to simulate the propagation effect of wave disaster risk between adjacent areas, and the characteristic distance parameter The value is 30 nautical miles, and the empirical coefficients are 、 、 . Figure 2 The risk propagation attenuation law of three high-risk source grids (N17E47, N16E47 and N16E46) is shown. The horizontal axis represents the distance from the source grid (nautical miles), and the vertical axis represents the risk propagation coefficient. The three curves of different colors in the figure represent the changing trend of the risk attenuation of different source grids with distance. The vertical dotted lines in the figure mark the propagation coefficient values of three special distance points of 15 nautical miles, 30 nautical miles and 45 nautical miles. It can be seen that the risk shows an exponential decay trend with increasing distance, and is affected by terrain complexity and wind field factors. This figure effectively demonstrates the spatial propagation effect of wave disaster risk and provides a scientific basis for determining the scope of the defense zone. The risk propagation effect is shown in Table 3: Table 3 Risk transmission coefficient of high-risk grid to surrounding grids

[0137] Generate the second change matrix , reflecting the impact of wave danger factors on the demarcation of defense areas.

[0138] The ship traffic frequency tensor and the wave hazard index are integrated and the risk comprehensive assessment equation group is used to conduct risk assessment. The weight coefficient in the risk quantification equation is 、 、 、 The ship type weight is 、 、 、 Error term Normal distribution .

[0139] In the resource allocation equation, the power parameter of the risk index is The value is 1.5, the adjustment coefficient 、 The weight coefficient of the benefit evaluation equation is 、 、 、 、 In the time constraint equation, the thresholds of each time dimension are 6 hours, 12 hours, 1 hour, and 3 hours respectively, and the maximum tolerance values are 24 hours, 36 hours, 6 hours, and 12 hours respectively. The adjustment coefficient .

[0140] Build a dynamic risk assessment hypergraph and set risk thresholds , filter out risk probability Greater than the regional average risk probability There are 27 high-risk grids, accounting for 6.0% of the total number of grids.

[0141] Apply the graph coloring problem optimization algorithm to differentiate the high-risk areas. Construct 27 high-risk grids into a graph , where the vertex set Contains all high-risk grids, edge sets Represents a neighboring relationship. A greedy algorithm variant was used for coloring. First, high-risk grids were sorted in descending order of risk index, such as N17E47 (risk index 0.873), N16E47 (risk index 0.846), N16E46 (risk index 0.835), and so on. Each grid was then assigned a color, ensuring that adjacent grids had different colors. Through algorithm optimization, four colors were ultimately used to label all high-risk grids. These four colors correspond to four different defense strategies: Color 1 (red) corresponds to strengthening the deployment of monitoring and early warning systems, Color 2 (blue) corresponds to increasing the frequency of rescue vessel patrols, Color 3 (green) corresponds to setting up temporary shelter anchorages, and Color 4 (yellow) corresponds to strengthening the communication support system. The graph coloring results are shown in Table 4: Table 4 High-risk area map coloring results

[0142] The coloring result is as follows Figure 3 As shown, construct the third change matrix , adjustment coefficient 、 .

[0143] The improved particle swarm algorithm is used to dynamically iterate and optimize the defense zone boundary. The number of particles is set to 80, the maximum number of iterations is 400, the inertia weight is linearly reduced from 0.9 to 0.4, and the acceleration constant is 、 、 The objective function weight coefficient is 、 、 、 After iterative optimization, the algorithm converged after 356 iterations and obtained the final wave disaster key defense area demarcation matrix The finalized defense zones covered 27 high-risk grids and 12 medium-risk grids, with a total area of approximately 9,750 square kilometers, accounting for 8.6% of the entire study area.

[0144] The graph coloring algorithm plays a key role in this implementation. Traditional defense zone delineation methods often adopt a homogenized strategy, applying the same defense measures to all high-risk areas, resulting in wasted resources and inefficient defense. This implementation, however, uses a graph coloring algorithm to treat defense zones as vertices in a graph, with the relationships between adjacent zones forming edges. Each vertex is assigned a different "color" (defense strategy) to ensure that adjacent zones employ distinct strategies. The implementation first constructs a neighbor relationship graph for high-risk zones, consisting of 27 vertices and 42 edges. Vertices are then sorted in descending order by risk index, prioritizing the highest-risk zones. A variant of the greedy algorithm is then used to assign the smallest available color to each vertex, ensuring that adjacent vertices have distinct colors. Finally, a local search is performed to optimize the color assignments, ultimately labeling all zones using four colors. This differentiated defense strategy configuration significantly improves resource utilization, avoids duplication of resources in adjacent zones, and ensures complementary and synergistic defense measures. For example, one zone can prioritize monitoring and early warning, while an adjacent zone can focus on deploying rescue forces, forming a complete disaster prevention chain. Experimental results show that compared with traditional methods, the defense strategy optimized by the graph coloring algorithm can improve resource utilization efficiency by 27.8%, increase defense coverage by 15.3%, and shorten response time by 21.6%.

[0145] The traditional method of delineating wave disaster defense zones mainly relies on single-factor assessment and empirical judgment, lacks comprehensive consideration of ship activity characteristics and wave hazard levels, and results in inefficient allocation of defense resources. The method adopted in this embodiment integrates AIS data and wave reanalysis data to construct a multidimensional risk assessment system, introduces a risk propagation model to describe the risk correlation between regions, and innovatively applies a graph coloring algorithm to achieve differentiated optimization of defense resources. Compared with traditional methods, the delineation results of this embodiment are more targeted and scientific. The area of the delineated defense zone is reduced by approximately 22.5%, but the risk coverage rate is increased by 18.4%, and the resource utilization efficiency is improved by 27.8%, which greatly reduces the defense cost and improves the effectiveness of wave disaster defense.

[0146] It should be noted that the variables involved in the present invention are explained in detail as shown in Tables 5, 6 and 7 below.

[0147] Table 5 Variable Explanation Table (Part I)

[0148] Table 6 Variable Explanation Table (Part II)

[0149] Table 7 Variable Explanation Table (Part 3)

[0150] The above description is only a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed by the present invention, which should be covered by the scope of protection of the present invention.

Claims

1. A method for delineating sea wave disaster defense zones based on AIS data, characterized in that: include: Divide the target sea area into regular grids and establish an initial division matrix; Collect ship AIS data and construct a ship type classification matrix; Count the number of different types of ships passing through each grid, and construct a multi-dimensional passage frequency tensor and the first change matrix; Collect historical wave reanalysis data and construct a wave hazard classification matrix; A composite wave disaster risk index model was constructed, the aggregation and dispersion of each grid were calculated, and the risk propagation function was applied to generate a second change matrix. The ship traffic frequency tensor and the wave hazard index were integrated, and a dynamic risk assessment hypergraph was constructed using a comprehensive risk assessment equation group. Multi-criteria dynamic thresholds are set to mark risk areas and construct the third change matrix. The improved particle swarm algorithm is used to dynamically iteratively optimize the defense zone boundaries and output the final wave disaster key defense zone demarcation matrix.

2. The method according to claim 1, characterized in that In the step of dividing the target sea area into a regular grid and establishing an initial division matrix, the initial division matrix refers to a two-dimensional array formed by dividing the target sea area into a regular grid according to an accuracy of 0.25 degrees longitude and 0.25 degrees latitude, and each element corresponds to a grid cell at a geographical location; the step of collecting ship AIS data and constructing a ship type division matrix specifically involves collecting and preprocessing ship AIS data from recent years, classifying ships according to their length, and constructing a ship type division matrix. The ship type division matrix refers to a classification matrix formed by classifying all ships according to their length characteristics into four categories: less than 50 meters, 50 to 100 meters, 100 to 150 meters, and greater than or equal to 150 meters, and is used to represent the distribution of different ship types in each grid.

3. The method according to claim 2, characterized in that In the step of counting the number of times different types of ships pass through each grid and constructing a multidimensional passage frequency tensor and a first change matrix, the contribution of ship AIS data to the parameters of each sub-area and the correlation between adjacent sub-areas are calculated. The contribution of ship AIS data to the parameters of each sub-area refers to the degree of influence of a single ship AIS data point on the calculation of relevant parameters of the grid in which it is located, including quantitative indicators of passage frequency and navigation density.

4. The method according to claim 3, characterized in that In the steps of collecting historical wave reanalysis data and constructing a wave hazard classification matrix, the average annual occurrence number of wave heights of levels I, II, III, and IV at each grid point is calculated, and a weighted formula is used to construct the wave hazard classification matrix. This matrix refers to a hazard assessment matrix constructed by calculating the average annual occurrence number of wave heights of different levels at each grid point, reflecting the degree of wave hazard in each area.

5. The method according to claim 4, characterized in that In the steps of constructing a composite wave disaster risk index model, calculating the aggregation and dispersion of each grid, and applying the risk propagation function to generate the second change matrix, the aggregation refers to the density of ship trajectory points in the grid, which is calculated by dividing the number of trajectory points per unit area by the regional average; the dispersion refers to the dispersion of ship navigation directions in the grid, which is calculated by dividing the standard deviation of ship headings in the grid by the mean standard deviation of headings in the entire region.

6. The method according to claim 5, characterized in that The risk propagation function is used to simulate the propagation effect of wave disaster risks between adjacent areas. The input includes the risk level of the source area, the geographical distance between regions, the wave propagation speed, the complexity of the seabed terrain and the wind field influence factor. The output is the intensity coefficient of the risk propagation from the source area to the target area.

7. The method according to claim 6, characterized in that In the step of constructing a dynamic risk assessment hypergraph by integrating the ship passage frequency tensor and the wave hazard index and applying the risk comprehensive assessment equation group, the risk comprehensive assessment equation group includes risk quantification equation, resource allocation equation, benefit assessment equation and time constraint equation, which are used to evaluate the comprehensive risk index of each grid point, the defense resource allocation ratio, the comprehensive benefit value of the defense plan and the time feasibility index of the plan.

8. The method according to claim 7, characterized in that A multi-criteria dynamic threshold is set to mark the risk area. In the step of constructing the third change matrix, a graph coloring problem optimization algorithm is applied for differential marking.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores program instructions, and when the program instructions are executed in a computer, they are used to execute the method for delineating wave disaster protection zones based on AIS data according to any one of claims 1 to 8.

10. The wave disaster protection zone demarcation system based on AIS data is characterized by: The computer-readable storage medium according to claim 9 is included, the system is any one of a computer, a server, and a single-chip microcomputer, the computer-readable storage medium is arranged in the system, and the system is provided with a microprocessor for executing program instructions stored in the computer-readable storage medium.

Images