Ocean wave disaster defensive area demarcation optimization method based on ship AIS data

Abstract

The invention provides an ocean wave disaster defensive area demarcation optimization method based on ship AIS data, and belongs to the technical field of natural disaster prevention and control, and the ocean wave disaster defensive area demarcation optimization method based on ship AIS data comprises the steps: dividing a target sea area into regular grids, and building coordinate mapping; cleaning the data, removing abnormal values and segmenting the data; a distributed track segment pre-clustering method is adopted, and clustering is carried out through dimension reduction, feature extraction and an Euclidean distance algorithm; establishing a multi-dimensional feature tensor according to ship length classification; constructing a composite sea wave disaster danger index model, and comprehensively evaluating the risk; carrying out statistics on ship track space-time characteristics in each grid, and constructing a passing frequency tensor; fusing the trajectory frequency and the disaster risk model, and constructing a dynamic risk assessment hypergraph; based on multi-criterion threshold clustering, defensive areas are identified, boundaries are optimized, and key defensive areas are delimited; according to the invention, sea area risk conditions can be provided for marine mobile disaster-bearing bodies such as sailing ships and the like.

Description

Technical Field

[0001] The present invention belongs to the technical field of natural disaster prevention and control, and in particular, relates to a method for optimizing the delineation of sea wave disaster defense zones based on ship AIS data. Background Art

[0002] The ocean is an important area of ​​human activity. In recent years, frequent wave disasters have caused serious damage to shipping, fisheries and coastal facilities, posing a major threat to the safety of ship navigation and the safety of people and property in coastal areas.

[0003] At present, the industry mainly adopts three defense measures, namely, disaster forecasting and early warning systems based on marine environmental monitoring data, wave disaster risk assessment based on historical data, and emergency response based on maritime control measures. However, these measures have problems such as limited coverage, inability to reflect dynamic changes, lack of relevance to actual ship operation conditions, and extensive emergency measures, making it difficult to fully grasp the risk status of the sea area and provide strong support for disaster prevention and mitigation work.

[0004] In addition, there are currently no relevant technical guidelines for the delineation of key defense areas for sea wave disasters. In addition, in the "Technical Guidelines for the Delineation of Key Defense Areas for Storm Surge Disasters" that has been issued, the disaster-prone objects are mainly fixed disaster-prone objects within the assessment area, and no analysis is conducted on mobile disaster-prone objects such as sailing ships at sea. Summary of the invention

[0005] In view of this, the present invention provides a method for optimizing the delineation of wave disaster defense zones based on ship AIS data, which can provide sea area risk conditions for mobile disaster-bearing bodies at sea such as sailing ships.

[0006] The present invention is achieved in that: The present invention provides a method for optimizing the delineation of sea wave disaster defense zones based on ship AIS data, which comprises the following steps: S10, dividing the target sea area into regular grids according to 0.025 degrees longitude and 0.025 degrees latitude, and establishing a grid coordinate mapping matrix; S20, performing data cleaning and segmentation processing on the acquired navigation trajectory data of the ship navigation automatic identification system, specifically including data cleaning, outlier removal, and trajectory segmentation processing based on a speed change rate threshold and a trajectory envelope; S30, using a distributed trajectory segment pre-clustering method based on massive logarithmic approximate cardinality estimation, through information entropy dimension reduction, spatiotemporal feature extraction and Euclidean distance clustering algorithm, to achieve efficient clustering of trajectory segments; S40, dividing the ships into four categories according to their lengths, namely, ship length less than 50 meters, ship length between 50 and 100 meters, ship length between 100 and 150 meters, and ship length greater than or equal to 150 meters, and establishing a multi-dimensional ship feature tensor; S50, construct a composite wave disaster risk index model, including historical wave height data analysis, ship dynamic response feature extraction, sea condition complexity index calculation and comprehensive wave disaster risk assessment based on support vector machine; S60, statistics the spatiotemporal characteristics of the trajectories of different types of ships in each grid in the past five years, and construct a multi-dimensional sparse passage frequency tensor; S70, multi-dimensional feature fusion of the ship trajectory frequency tensor and the wave disaster risk assessment model, and use principal component decomposition and Bayesian probability graph model to construct a dynamic risk assessment hypergraph; S80. Based on multi-criteria dynamic thresholds, probabilistic cluster analysis is performed on the risk assessment hypergraph to identify key defense areas for wave disasters and establish a multi-level defense zone classification standard; S90. An improved meta-heuristic optimization algorithm is used to dynamically iteratively optimize the boundaries of the defense zone, minimize the overall disaster risk and computational complexity in the area, and obtain a robust delineation of key defense areas for wave disasters.

[0007] On the basis of the above technical solution, the wave disaster defense zone delineation optimization method based on ship AIS data of the present invention can also be improved as follows: Wherein, the S10 specifically includes: Step 101: Obtain the geospatial data of the target sea area and determine the longitude and latitude range of the sea area; Step 102, dividing the target sea area into regular grids according to the accuracy of 0.025 degrees longitude and 0.025 degrees latitude; Step 103: Establish a grid coordinate mapping matrix to map each grid cell to a unique two-dimensional coordinate index; Step 104: verifying the spatial topological relationship of the grid coordinate mapping matrix to ensure the integrity and continuity of the grid division; Step 105: Record the boundary coordinates and spatial reference system of the grid coordinate mapping matrix.

[0008] Furthermore, the S20 specifically includes: Step 201: Perform preliminary data cleaning on the ship navigation automatic identification system data to remove records with missing timestamps and incomplete location information; Step 202: Perform outlier detection on the data after preliminary cleaning, including data points of speed exceeding the limit and position jump; Step 203: based on the ship speed change rate threshold, segment the navigation trajectory and divide the trajectory segments with continuous abnormal speed changes into independent research units; Step 204: construct a trajectory envelope, identify and filter trajectory points beyond the envelope range; Step 205: Perform repeatability check and consistency marking on the segmented trajectory data.

[0009] Furthermore, the S30 specifically includes: Step 301, preprocessing the trajectory segmentation data using a massive logarithmic approximate cardinality estimation algorithm; Step 302: extract key spatial features of the trajectory segment by using an information entropy dimension reduction method; Step 303, extracting multi-dimensional feature vectors such as time, space, and speed of the trajectory segment; Step 304: applying the Euclidean distance to calculate the similarity between the trajectory segments; Step 305: Based on the similarity of trajectory segment features, a clustering algorithm is executed to achieve efficient grouping of trajectory segments.

[0010] Furthermore, the S40 specifically includes: Step 401, obtaining ship length measurement data; Step 402, classifying the ships according to four categories: ship length less than 50 meters, ship length 50 to 100 meters, ship length 100 to 150 meters, and ship length greater than or equal to 150 meters; Step 403: extracting the dynamic characteristics, navigation characteristics, and load characteristics of each ship category; Step 404: construct a multi-dimensional ship feature tensor, including dimensions such as length, draft, sailing speed, and load; Step 405: Standardize the multi-dimensional ship characteristic tensor to eliminate dimensional differences.

[0011] Furthermore, the S50 specifically includes: Step 501: Collect historical wave height data and analyze wave height distribution patterns and statistical characteristics; Step 502: extracting dynamic response characteristics of the ship, including roll angle, pitch angle, and heave displacement; Step 503, calculating sea condition complexity indexes, including wave height, period, and wave direction change rate; Step 504: construct a support vector machine training data set, including marine environmental characteristics and historical disaster data; Step 505: Train the support vector machine model to achieve comprehensive assessment of wave disaster risks.

[0012] Furthermore, the S60 specifically includes: Step 601, collecting data on ship navigation tracks in the past five years; Step 602: extracting spatiotemporal features according to ship type classification; Step 603, calculating the passage frequency of different types of ships in each grid; Step 604: construct a multi-dimensional sparse traffic frequency tensor, including time, space, and ship type dimensions; Step 605: normalize the passing frequency tensor.

[0013] Furthermore, the S70 specifically includes: Step 701: align the ship trajectory frequency tensor with the wave disaster risk assessment model; Step 702: Apply principal component decomposition method to reduce the dimension of feature tensor; Step 703: construct a Bayesian probability graph model to establish a probability dependency relationship between features; Step 704: Calculate the node and edge weights of the dynamic risk assessment hypergraph; Step 705: Perform probability inference and risk prediction on the dynamic risk assessment hypergraph.

[0014] Furthermore, the S80 specifically includes: Step 801, defining multi-criteria dynamic thresholds, including dimensions such as disaster risk, traffic frequency, and ship type; Step 902: Performing probabilistic cluster analysis on the risk assessment hypergraph; Step 803: Identify high-risk areas for sea wave disasters; Step 804: Establish a multi-level defense zone classification standard and divide the risk levels; Step 805: Generate a map of key areas for defense against sea wave disasters.

[0015] Furthermore, the S90 specifically includes: Step 901: Select an improved meta-heuristic optimization algorithm, such as an improved particle swarm optimization algorithm or a simulated annealing algorithm; Step 902: setting an optimization objective function, including minimizing the overall disaster risk in the region and the computational complexity; Step 903, initializing a candidate solution set for the defense zone boundary; Step 004, iteratively update the defense zone boundary and evaluate each solution using the fitness function; Step 905: output the delineation result of the key defense area for sea wave disasters with the highest fitness.

[0016] 1. Grid coordinate mapping matrix construction step (S10): Grid coordinate mapping matrix Defined as: ; In the formula, Represents longitude coordinates; Represents latitude coordinates; Indicates the grid cell number; is the number of grids in the longitude direction; is the number of grids in the latitude direction.

[0017] Grid division accuracy calculation formula: ; Grid cell unique identification mapping function: ; 2. Ship track data processing (S20): Trajectory outlier removal criteria: ; In the formula, is a single velocity observation value; is the mean speed; is the speed standard deviation; is the threshold coefficient (usually ).

[0018] Calculation of velocity change rate of trajectory segment: ; when Segment the trajectory.

[0019] 3. Trajectory pre-clustering method (S30): (1) Massive Logarithmic Approximate Cardinality Estimation Algorithm (HyperLogLog): Cardinality estimation formula: ; In the formula, is the correction constant; is the number of barrels (barrels); For the Maximum number of leading zeros in a bucket.

[0020] (2) Information entropy dimension reduction formula: Information entropy calculation: ; Feature weight calculation: ; Feature vector after dimensionality reduction: ; (3) Trajectory segment feature extraction: Time feature vector: ; Spatial eigenvectors: ; Velocity eigenvector: ; Comprehensive feature vector: ; 1. Time feature vector Calculation: ; Where: is the starting timestamp of the trajectory segment, in seconds; is the end timestamp of the trajectory segment, in seconds; is the duration of the trajectory segment, and the calculation formula is: ; is the average time interval of the trajectory segment, and the calculation formula is: ; is the time interval variance of the trajectory segment, and the calculation formula is: .

[0021] 2. Spatial eigenvectors Calculation: ; Where: is the total length of the trajectory segment, and the calculation formula is: , the unit is nautical miles; is the tortuosity of the trajectory segment, and the calculation formula is: ,in is the straight-line distance from the starting point to the end point; is the average heading angle of the trajectory segment, and the calculation formula is: , the unit is degree; is the average curvature of the trajectory segment, and the calculation formula is: ,in For the The tangent angle of a point, is the arc length between adjacent points.

[0022] 3. Velocity eigenvector Calculation: ; Where: is the average speed of the trajectory segment, and the calculation formula is: , the unit is node; is the maximum speed of the trajectory segment, and the calculation formula is: ; is the minimum speed of the trajectory segment, and the calculation formula is: ; is the velocity standard deviation of the trajectory segment, and the calculation formula is: ; is the median velocity of the trajectory segment.

[0023] 4. Comprehensive feature vector Build: ; This vector concatenates the above three feature vectors into a complete feature description with a total dimension of 14 (5+4+5).

[0024] The design principles and significance of these feature vectors are as follows: 1. The temporal feature vector reflects the temporal characteristics of the trajectory: the start and end time mark the specific time when the trajectory occurs; the duration reflects the time scale of the movement; the mean and variance of the time interval reflect the uniformity of the sampling.

[0025] 2. The spatial eigenvector describes the geometric characteristics of the trajectory: the trajectory length reflects the spatial range of movement; the tortuosity characterizes the complexity of the path; the heading angle reflects the main direction of movement; and the curvature describes the degree of curvature of the path.

[0026] 3. The speed eigenvector characterizes the dynamic characteristics of the motion: the mean and median reflect the overall level of speed; the maximum and minimum values ​​reflect the range of speed variation; and the standard deviation reflects the degree of speed fluctuation.

[0027] (4) Feature similarity measurement: Euclidean distance calculation: ; Cosine Similarity: ; (5) Clustering algorithm (DBSCAN): Core object judgment: ; Density reachable definition: If there is a point sequence ,in ,and Depend on The density can reach from Density can reach.

[0028] Cluster definition: ; (6) Clustering quality assessment: Silhouette Coefficient: ; In the formula, is the average distance within the same cluster; is the average distance of the nearest clusters.

[0029] 4. Ship feature tensor construction (S40): Multidimensional ship feature tensor definition: ; In the formula, For the captain; is the draft depth; is the sailing speed; For load capacity.

[0030] Tensor normalization transformation: ; 5. Ocean wave disaster risk index model (S50): Support vector machine risk assessment objective function: ; Constraints: ; ; 6. Ship traffic frequency tensor (S60): Passing frequency calculation: ; In the formula, The number of passages for a particular type of vessel; is the total number of passes.

[0031] 7. Risk assessment hypergraph construction (S70): Principal component decomposition: ; Bayesian Probabilistic Graphical Model Joint Probability: ; 8. Defense zone cluster analysis (S80): Multi-criteria dynamic threshold: ; In the formula, for disaster risk; is the passing frequency; is the vessel type; is the weight coefficient.

[0032] 9. Defense zone boundary optimization (S90): Objective function: ; Indicates risk, Represents complexity; Simulated annealing acceptance probability: ; In the formula, For energy changes; is the Boltzmann constant; For temperature.

[0033] in, ; In the formula, is the wave disaster risk index, with a value range of [0,1]; is the ship vulnerability risk index, with a value range of [0,1]; is the navigation density risk index, with a value range of [0,1]; is the environmental factor risk index, with a value range of [0,1]; is the weight coefficient and satisfies .

[0034] Among them, the wave disaster risk index The calculation formula is: ; In the formula, For the The effective wave height at the moment, in meters; is the historical maximum significant wave height, in meters; For the The wave period at a certain moment, in seconds; is the reference period, which is 10 seconds; is the length of the observation time series.

[0035] Ship Vulnerability Risk Index The calculation formula is: ; In the formula, For the Weight coefficient of the type of ship; For the Number of vessels of this category; is the total number of ships; For the Average length of ships of this category in metres; The maximum ship length is 300 meters.

[0036] Traffic density risk index The calculation formula is: ; In the formula, For the The frequency of traffic in each grid; For the The average speed of the grid, in knots; is the maximum passing frequency; For reference speed, the value is 12 knots; is the total number of grids.

[0037] Environmental Factor Risk Index The calculation formula is: ; In the formula, is the water depth in meters; is the wind speed in meters per second; is the flow velocity in meters per second; They are the maximum water depth, maximum wind speed and maximum current velocity respectively; is the weight coefficient and satisfies .

[0038] The calculation formula of complexity is specifically expressed as follows: ; In the formula, is the mesh complexity; is the computational complexity; is the boundary complexity; is the weight coefficient and satisfies .

[0039] Mesh complexity The calculation formula is: ; In the formula, is the number of active grids; is the total number of grids; is the grid resolution.

[0040] Computational complexity The calculation formula is: ; In the formula, is the data processing time in seconds; is the reference processing time, and the value is 1 second; is the used memory; is the total available memory.

[0041] Boundary Complexity The calculation formula is: ; In the formula, is the total length of the border line in kilometers; is the reference length, which is 100 km; is the number of boundary vertices; The reference number of vertices is 100.

[0042] The construction principle and significance of the above equation are as follows: 1. The Risk equation adopts the form of weighted summation, taking into account multi-dimensional risk factors, including waves, ships, transportation and environment, and realizes the differentiated impact of each factor through weight coefficients; 2. The exponential function relationship is used to reflect the nonlinear effect of wave period on risk, and the wave height is processed by linear normalization; 3. The quantity distribution and size characteristics of different types of ships are taken into consideration, and an inverse relationship is used to reflect the stronger wind and wave resistance of larger ships; 4. The normalized product form is adopted to reflect the coupling effect of passage frequency and ship speed; 5. The linear combination form is adopted to reflect the comprehensive influence of multiple environmental factors; 6. The Complexity equation also uses a weighted summation form to evaluate complexity from three dimensions: grid, computation, and boundary; 7. The logarithmic relationship is adopted to reflect the nonlinear complexity growth caused by the increase in the number of grids; 8. The ratio relationship of resource utilization is adopted to reflect the computing load; 9. The comprehensive influence of boundary length and number of vertices is considered, and the coupling effect of the two is reflected in the form of product.

[0043] The present invention proposes an optimization method for delineating sea wave disaster defense zones based on ship AIS data, which has the following significant advantages over the prior art: 1. Make full use of the spatiotemporal characteristics of AIS data. The AIS system can collect dynamic operation data such as the position, heading, and speed of ships in real time, providing rich basic data for accurately assessing the real-time risk status of the sea area. The method of the present invention fully exploits the spatiotemporal characteristics contained in AIS data, establishes a model to describe the frequency of passage of different types of ships in various sea areas, and provides valuable input for subsequent risk assessment and defense zone optimization.

[0044] 2. Integrate marine environment and ship response characteristics. In addition to using AIS data, the method of the present invention also combines historical wave height data analysis, ship dynamic response feature extraction, and sea condition complexity assessment to construct a more comprehensive wave disaster risk assessment model. This assessment method that integrates multi-source data can more accurately reflect the actual disaster risk status of the sea area compared to traditional models that rely solely on historical accident data.

[0045] 3. Realize refined defense zone delineation. Based on the above risk assessment results, the method of the present invention further uses probabilistic clustering and optimization algorithms to dynamically identify key defense areas for wave disasters, and iteratively optimizes the boundaries of the defense zones, ultimately obtaining a defense zone delineation scheme that is robust and can minimize overall risks. This refined defense zone delineation can better match the differentiated needs of different sea areas and ship types compared to traditional extensive emergency control measures.

[0046] 4. Improve the intelligent level of sea wave disaster defense. The present invention integrates advanced mathematical modeling, machine learning, optimization algorithms and other technologies to achieve automatic assessment of sea wave disaster risks and intelligent optimization of defense areas, greatly improving the informationization and intelligent level of disaster prevention and mitigation. This intelligent defense method can provide more accurate decision-making support for maritime authorities and improve the efficiency and accuracy of emergency response.

[0047] In summary, the wave disaster defense zone demarcation optimization method based on AIS data proposed in the present invention, by making full use of maritime big data and integrating multi-source marine environment and ship characteristics, constructs an intelligent and sophisticated wave disaster defense system. Compared with the existing technology, it significantly improves the accuracy and effectiveness of disaster prevention and mitigation work, and makes important contributions to ensuring maritime traffic safety and coastal facility safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0049] Figure 1This is a flow chart of an optimization method for delineating wave disaster defense zones based on ship AIS data. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0051] like Figure 1 As shown, it is a first embodiment of a method for optimizing the delineation of wave disaster defense zones based on ship AIS data provided by the present invention. In this embodiment, the following steps are included: S10, dividing the target sea area into regular grids according to 0.025 degrees longitude and 0.025 degrees latitude, and establishing a grid coordinate mapping matrix; S20, performing data cleaning and segmentation processing on the acquired navigation trajectory data of the ship navigation automatic identification system, specifically including data cleaning, outlier removal, and trajectory segmentation processing based on a speed change rate threshold and a trajectory envelope; S30, using a distributed trajectory segment pre-clustering method based on massive logarithmic approximate cardinality estimation, through information entropy dimension reduction, spatiotemporal feature extraction and Euclidean distance clustering algorithm, to achieve efficient clustering of trajectory segments; S40, dividing the ships into four categories according to their lengths, namely, ship length less than 50 meters, ship length between 50 and 100 meters, ship length between 100 and 150 meters, and ship length greater than or equal to 150 meters, and establishing a multi-dimensional ship feature tensor; S50, construct a composite wave disaster risk index model, including historical wave height data analysis, ship dynamic response feature extraction, sea condition complexity index calculation and comprehensive wave disaster risk assessment based on support vector machine; S60, statistics the spatiotemporal characteristics of the trajectories of different types of ships in each grid in the past five years, and construct a multi-dimensional sparse passage frequency tensor; S70, multi-dimensional feature fusion of the ship trajectory frequency tensor and the wave disaster risk assessment model, and use principal component decomposition and Bayesian probability graph model to construct a dynamic risk assessment hypergraph; S80. Based on multi-criteria dynamic thresholds, probabilistic cluster analysis is performed on the risk assessment hypergraph to identify key defense areas for wave disasters and establish a multi-level defense zone classification standard; S90. An improved meta-heuristic optimization algorithm is used to dynamically iteratively optimize the boundaries of the defense zone, minimize the overall disaster risk and computational complexity in the area, and obtain a robust delineation of key defense areas for wave disasters.

[0052] Among them, in the above technical scheme, the specific implementation method of step S10 is: first obtain the geographic spatial data of the target sea area and determine the longitude and latitude range of the sea area. Then, the target sea area is divided into regular grids according to the accuracy of 0.025 degrees longitude and 0.025 degrees latitude. Then, a grid coordinate mapping matrix is ​​established to map each grid unit to a unique two-dimensional coordinate index. On this basis, the spatial topological relationship of the grid coordinate mapping matrix is ​​verified to ensure the integrity and continuity of the grid division. Finally, the boundary coordinates and spatial reference system of the grid coordinate mapping matrix are recorded. The purpose of this step is to establish a structured grid space to provide a unified geographic framework for subsequent data analysis and model building.

[0053] Furthermore, in the above technical solution, the specific implementation method of step S20 is: first, the acquired ship navigation automatic identification system data is preliminarily cleaned to remove records with missing timestamps and incomplete location information. Then, the preliminarily cleaned data is subjected to outlier detection, including data points of speed exceeding the limit and position jump. Then, based on the ship speed change rate threshold (usually set to 20%), the navigation trajectory is segmented, and the trajectory segments with continuous abnormal speed changes are divided into independent research units. Then, the trajectory envelope is constructed to identify and filter the trajectory points that exceed the envelope range. Finally, the segmented trajectory data is subjected to repeatability verification and consistency marking. The purpose of this step is to denoise and preprocess the original ship trajectory data to ensure the reliability and consistency of the data.

[0054] Furthermore, in the above technical solution, the specific implementation method of step S30 is: first, a massive logarithmic approximate cardinality estimation algorithm is used to pre-process the trajectory segment data to estimate the cardinality scale of the trajectory segment. Then, the information entropy dimension reduction method is used to extract the key spatial features of the trajectory segment and eliminate redundant information. Then, the multi-dimensional feature vectors of the trajectory segment such as time, space, and speed are extracted. The Euclidean distance is used to calculate the similarity between the trajectory segments. Finally, based on the similarity of the trajectory segment features, a density-based clustering algorithm (DBSCAN) is executed to achieve efficient grouping of the trajectory segments. The purpose of this step is to perform effective feature extraction and clustering analysis on the massive ship trajectory data, laying the foundation for subsequent risk assessment.

[0055] Furthermore, in the above technical solution, the specific implementation method of step S40 is: first, obtain the measurement data of the ship's length. Then classify the ships according to four categories: ship length less than 50 meters, ship length 50 to 100 meters, ship length 100 to 150 meters, and ship length greater than or equal to 150 meters. Then extract the dynamic characteristics, navigation characteristics, and load characteristics of each ship category. Then construct a multidimensional ship feature tensor, including dimensions such as length, draft, navigation speed, and load. Finally, standardize the multidimensional ship feature tensor to eliminate dimensional differences. The purpose of this step is to establish a multidimensional ship feature description model to provide necessary ship feature inputs for subsequent disaster risk assessment.

[0056] Furthermore, in the above technical solution, the specific implementation method of step S50 is: first collect historical wave height data and analyze the distribution law and statistical characteristics of wave height. Then extract the dynamic response characteristics of the ship, including indicators such as roll angle, pitch angle and heave displacement. Then calculate the complexity index of the sea condition, including factors such as wave height, period and wave direction change rate. Then construct a support vector machine training data set, which includes marine environmental characteristics and historical disaster data. Finally, train the support vector machine model to achieve a comprehensive assessment of the risk of wave disasters. The purpose of this step is to establish a wave disaster risk assessment model that can comprehensively consider the characteristics of the marine environment and the dynamic response of the ship.

[0057] Furthermore, in the above technical solution, the specific implementation method of step S60 is: first, the ship navigation trajectory data for the past five years is counted. Then, the spatiotemporal characteristics are extracted according to the ship type classification. Then, the passage frequency of different types of ships in each grid is calculated. Then, a multidimensional sparse passage frequency tensor is constructed, which includes three dimensions of time, space and ship type. Finally, the passage frequency tensor is normalized. The purpose of this step is to establish a tensor model that describes the spatiotemporal passage characteristics of different types of ships in each grid area, and provide data support for subsequent risk assessment and defense zone delineation.

[0058] Furthermore, in the above technical solution, the specific implementation method of step S70 is: first, the ship trajectory passage frequency tensor is feature aligned with the wave disaster risk assessment model. Then the principal component decomposition method is applied to reduce the dimension of the feature tensor. Then a Bayesian probability graph model is constructed to establish a probabilistic dependency relationship between features. Then the node and edge weights of the dynamic risk assessment hypergraph are calculated. Finally, the dynamic risk assessment hypergraph is subjected to probability inference and risk prediction. The purpose of this step is to integrate the ship passage frequency information with the wave disaster risk assessment model, construct a dynamic risk assessment hypergraph, and provide a basis for the subsequent defense zone delineation.

[0059] Furthermore, in the above technical solution, the specific implementation method of step S80 is: first, dynamic thresholds including multiple dimensions such as disaster risk, traffic frequency and ship type are defined. Then, a probability clustering analysis is performed on the risk assessment hypergraph to identify high-risk areas for wave disasters. Then, a multi-level defense zone classification standard is established to divide different risk levels. Finally, a spatial mapping of key defense areas for wave disasters is generated. The purpose of this step is to zoning and demarcate the sea area according to the multi-dimensional risk characteristics, so as to provide a reference for the subsequent optimization algorithm.

[0060] Furthermore, in the above technical solution, the specific implementation method of step S90 is: first, select an improved meta-heuristic optimization algorithm, such as an improved particle swarm algorithm or a simulated annealing algorithm. Then set the optimization objective function, including minimizing the overall disaster risk in the region and the computational complexity. Then initialize the candidate solution set of the defense zone boundary. Then perform iterative updates, and use the fitness function to evaluate the pros and cons of each solution. Finally, output the delineation result of the key defense zone for sea wave disasters with the highest fitness. The purpose of this step is to use the optimization algorithm to dynamically adjust the defense zone boundary that has been preliminarily determined, and finally obtain a defense zone delineation scheme with strong robustness.

[0061] The specific steps of the present invention are as follows: First, step S10 is to establish a structured grid space to provide a unified geographic framework for subsequent data analysis and model building. The specific implementation method is as follows: Grid coordinate mapping matrix Defined as .in, represents the longitude coordinate, represents the latitude coordinate, represents the grid cell number, is the number of grids in the longitude direction, is the number of grids in the latitude direction.

[0062] The formula for calculating the grid division accuracy is: This means that the target sea area is gridded with a latitude and longitude accuracy of 0.025 degrees.

[0063] The grid unit unique identification mapping function is: Through this mapping function, each grid cell can be mapped to a unique two-dimensional index number.

[0064] Next is step S20, the purpose of which is to remove noise and pre-process the original ship trajectory data to ensure the reliability and consistency of the data. The specific implementation is as follows: The criterion for removing trajectory outliers is: .in, is a single velocity observation value, is the mean velocity, is the speed standard deviation, is the threshold coefficient (usually ). This formula is used to identify and remove abnormal data points that are outside the normal speed range.

[0065] The calculation formula of trajectory segment speed change rate is: .when When the trajectory is segmented, the trajectory segments with continuous abnormal speed changes are divided into independent research units. Typically set to 20%.

[0066] The purpose of step S30 is to perform effective feature extraction and cluster analysis on massive amounts of ship trajectory data, laying the foundation for subsequent risk assessment. The specific implementation is as follows: (1) The cardinality estimation formula of the massive logarithmic approximate cardinality estimation algorithm (HyperLogLog) is: .in, is the correction constant, is the number of barrels (barrels), For the The maximum number of leading zeros in a bucket. This algorithm can efficiently estimate the cardinality of massive trajectory segments with extremely low storage space.

[0067] (2) The information entropy calculation formula is: The feature weight calculation formula is: The eigenvector after dimensionality reduction is: Through information entropy dimension reduction, the key spatial features of the trajectory segment can be effectively extracted.

[0068] (3) The trajectory segment feature vector is defined as follows: time feature vector , space feature vector , velocity feature vector , comprehensive feature vector .

[0069] (4) The Euclidean distance calculation formula is: The formula for calculating cosine similarity is: . These two formulas are used to measure the similarity between trajectory segments.

[0070] (5) In the DBSCAN clustering algorithm, the core object judgment formula is: Density reachability is defined as: if there exists a point sequence ,in ,and Depend on The density can reach , then it is called from The density can be reached. The cluster is defined as: .

[0071] (6) The calculation formula of silhouette coefficient is: .in, is the average distance within the same cluster, is the average distance of the nearest clusters. The silhouette coefficient is used to evaluate the clustering quality.

[0072] The purpose of step S40 is to establish a multi-dimensional ship characteristic description model to provide necessary ship characteristic input for subsequent disaster risk assessment. The specific implementation method is as follows: Multidimensional ship feature tensor Defined as .in, For the Captain, is the draft depth, is the sailing speed, For load capacity.

[0073] The tensor normalization transformation formula is: This formula is used to eliminate the dimensional differences between different features so that the eigenvalues ​​are at the same level.

[0074] The purpose of step S50 is to establish a wave disaster risk assessment model that can comprehensively consider the characteristics of the marine environment and the dynamic response of the ship. The specific implementation is as follows: The risk assessment objective function of the support vector machine is: The constraints are: This optimization problem aims to find the optimal classification hyperplane so that the training data can be correctly classified while minimizing the classification error.

[0075] The purpose of step S60 is to establish a tensor model that describes the spatiotemporal characteristics of different types of ships in each grid area, providing data support for subsequent risk assessment and defense zone delineation. The specific implementation method is as follows: The formula for calculating the frequency of traffic is: .in, is the number of passages of a specific type of ship, is the total number of passes.

[0076] The purpose of step S70 is to integrate the ship traffic frequency information with the wave disaster risk assessment model to construct a dynamic risk assessment hypergraph to provide a basis for the subsequent defense zone delineation. The specific implementation method is as follows: The principal component decomposition formula is: . This formula is used to reduce the dimensionality of the feature tensor.

[0077] The joint probability formula of the Bayesian probabilistic graphical model is: . This formula describes the probabilistic dependency between features.

[0078] The purpose of step S80 is to zoning and demarcate the sea area according to the multi-dimensional risk characteristics, so as to provide a reference for the subsequent optimization algorithm. The specific implementation is as follows: The multi-criteria dynamic threshold calculation formula is: .in, For disaster risk, is the passing frequency, For ship type, is the weight coefficient.

[0079] The purpose of step S90 is to dynamically adjust the defense zone boundary initially determined by using an optimization algorithm, and finally obtain a defense zone demarcation scheme with strong robustness. The specific implementation method is as follows: The optimization objective function is defined as: This formula describes the optimization objective of minimizing the overall disaster risk in the region and the computational complexity.

[0080] The acceptance probability calculation formula in the simulated annealing algorithm is: .in, For energy change, is the Boltzmann constant, This formula is used to evaluate the acceptance probability of a new solution and control the degree of exploration of the algorithm.

[0081] Table 1 shows the meaning of the formula symbols in the present invention:

[0082] The present invention provides a second embodiment of a method for optimizing the delineation of wave disaster defense zones based on ship AIS data. In this embodiment, a specific application scenario of the present invention is provided: A maritime administration department is responsible for the jurisdiction of an important sea area, which is a typical shipping and fishery gathering area. In history, there have been many serious ship accidents and coastal facility damage caused by large waves and huge waves. In order to improve the sea wave disaster defense capability of the sea area, the department decided to adopt the sea wave disaster defense zone delineation optimization method based on ship AIS data proposed in this invention.

[0083] First, the maritime department collected the AIS navigation track data of the sea area in the past five years, as well as historical wave height observation data and some ship accident records. According to step S10 of the method of the present invention, they gridded the target sea area with a longitude and latitude accuracy of 0.025 degrees. Specifically, the longitude and latitude range of the sea area is (longitude 110°E-115°E, latitude 22°N-26°N). After gridding, it is divided into 1600 grid cells (40×40). The maritime department established the grid coordinate mapping matrix shown in Table 2 , records the latitude and longitude coordinates of each grid cell and a unique number index.

[0084] Table 2 Grid coordinate mapping matrix

[0085] Next, according to step S20, the maritime department preprocessed and analyzed the original AIS trajectory data. First, they removed records with missing timestamps and incomplete location information, and cleaned up 185,642 valid trajectory data. Then, they segmented the trajectory according to the ship speed change rate threshold (set to 20%), and obtained a total of 293,586 trajectory segments. On this basis, they further identified and filtered out abnormal points that exceeded the trajectory envelope range to ensure the consistency and reliability of the data. Table 3 shows some sample data of preprocessed trajectory segments.

[0086] Table 3 Example of preprocessed AIS trajectory segment data

[0087] According to step S30, the maritime department used the massive logarithmic approximate cardinality estimation (HyperLogLog) algorithm to perform preliminary clustering analysis on the preprocessed trajectory segment data. This algorithm can efficiently estimate the cardinality scale of massive trajectory segments, laying the foundation for subsequent feature extraction and clustering analysis. After that, they extracted the time, space and speed features of each trajectory segment and constructed a comprehensive feature vector F_trajectory. Through information entropy dimensionality reduction, they extracted the most critical 8 features from the original 23-dimensional features, greatly reducing the computational complexity. On this basis, they applied the DBSCAN clustering algorithm to efficiently cluster 293,586 trajectory segments into 9,786 clusters based on the Euclidean distance similarity between trajectory segments.

[0088] Next, according to step S40, the maritime department collected detailed scale data of various types of ships in the sea area and divided the ships into four categories: less than 50 meters, 50-100 meters, 100-150 meters, and greater than or equal to 150 meters. For each type of ship, they extracted features such as length, draft, sailing speed, and load, and constructed a multidimensional ship feature tensor as shown in Table 4 .

[0089] Table 4 Multidimensional ship feature tensor

[0090] Then, according to step S50, the maritime department conducted a statistical analysis of the historical wave height data of the sea area and found that the maximum wave height in the sea area reached 15 meters, with an annual frequency of about 1 / 50. At the same time, they extracted the dynamic response characteristics of typical ships under severe sea conditions, including roll angle ±8°, pitch angle ±5° and heave displacement ±1.2 meters. Based on these environmental and ship characteristics, they constructed a wave disaster risk assessment model based on support vector machines and trained it to obtain a relatively accurate risk prediction capability.

[0091] According to step S60, the maritime department counted the passage frequency of different types of ships in each grid area in the past five years and constructed a multidimensional sparse passage frequency tensor F_vessel as shown in Table 5. It can be seen that small ships have a higher passage frequency in shallow waters near the coast, while large ships are mainly concentrated in the offshore area. This difference in spatiotemporal characteristics provides an important basis for the subsequent refined delineation of defense zones.

[0092] Table 5 Multidimensional sparse frequency tensor F_vessel Grid number ω_(i,j) <50m ship 50-100m ship 100-150m ship >=150m ship

[0093] Further according to step S70, the maritime department integrated the above risk assessment model and the traffic frequency tensor to construct a dynamic risk assessment hypergraph model. Specifically, the principal component analysis method was first used to reduce the dimension of the feature tensor, and 8 principal components with a contribution rate of 95% of the explained variance were extracted. Then, they used the Bayesian probabilistic graph model to describe the probabilistic dependency between these principal components, calculated the node and edge weights of each grid unit, and formed a dynamic risk assessment hypergraph.

[0094] Based on step S80, the maritime department defined a multi-criteria dynamic threshold, including three dimensions: disaster risk, passage frequency, and ship type, and performed a probabilistic cluster analysis on the risk assessment hypergraph. After clustering, they divided the sea area into five defense areas with different risk levels, as shown in Table 6. Among them, the high-risk areas are mainly distributed in near-shore shallow waters and certain waterways, while the low-risk areas are mainly located in the offshore area. This refined delineation of defense areas can provide targeted support for subsequent emergency management and control.

[0095] Table 6 Results of delineation of sea wave disaster protection zones

[0096] Finally, according to step S90, the maritime department used an improved simulated annealing algorithm to further optimize the boundaries of the above defense zones. The optimization objective function they set included minimizing the overall disaster risk in the region and the computational complexity. After multiple rounds of iterations, the algorithm finally output a more robust defense zone delineation scheme. Compared with the preliminary clustering division, the optimized scheme appropriately merged Area 1 and Area 2, reducing the overall defense cost, while also ensuring that the high-risk defense needs of key areas are met.

[0097] The core of the wave disaster defense zone delineation optimization method of the present invention is to fully explore the spatiotemporal characteristics contained in AIS data, combine the marine environment and the dynamic response characteristics of ships, build a multi-dimensional and dynamic risk assessment model, and realize refined defense zone delineation based on this model. The key principles of this method are as follows: First, a structured grid space model is established. The invention method divides the target sea area into grids according to a certain latitude and longitude accuracy (such as 0.025 degrees), and establishes a grid coordinate mapping matrix to associate the actual geographic space with the numerical calculation model. This gridded spatial model lays a solid foundation for subsequent data analysis, feature extraction, and risk assessment.

[0098] Secondly, the original AIS trajectory data was preprocessed and analyzed in depth. The invented method firstly cleaned the AIS data and eliminated outliers to ensure the reliability of the data. Then, through trajectory segmentation, feature extraction and clustering, the spatiotemporal characteristics of various types of ships in different sea areas were effectively extracted from the massive AIS trajectory data, and a multi-dimensional ship traffic frequency tensor model was constructed. Such a model can comprehensively describe the actual traffic conditions of different types of ships in each grid area, providing key input for subsequent risk assessment.

[0099] Thirdly, the invented method integrates the characteristics of the marine environment and the dynamic response of the ship to construct a more comprehensive wave disaster risk assessment model. Specifically, the model combines the analysis of historical wave height data, the extraction of dynamic response features such as ship roll angle / pitch angle / heave displacement, and a comprehensive risk assessment method based on support vector machine. This assessment model that integrates multi-source information can more accurately reflect the actual disaster risks faced by different sea areas and ship types.

[0100] Finally, the inventive method uses probabilistic clustering and optimization algorithms to further analyze and optimize the risk assessment results. First, the sea area is divided into different risk level areas through dynamic threshold division of multiple criteria. Then, the improved meta-heuristic optimization algorithm is applied to dynamically iteratively optimize the initially determined defense zone boundaries, minimize the overall disaster risk and computational complexity, and obtain a robust defense zone delineation scheme. This dynamic adjustment based on the optimization algorithm enables the defense zone delineation to better adapt to changes in the sea environment and ship conditions.

Claims

1. A method for optimizing the delineation of sea wave disaster defense zones based on ship AIS data, characterized in that: The following steps are involved: S10, dividing the target sea area into regular grids according to 0.025 degrees longitude and 0.025 degrees latitude, and establishing a grid coordinate mapping matrix; S20, performing data cleaning and segmentation processing on the acquired navigation trajectory data of the ship navigation automatic identification system, specifically including data cleaning, outlier removal, and trajectory segmentation processing based on a speed change rate threshold and a trajectory envelope; S30, using a distributed trajectory segment pre-clustering method based on massive logarithmic approximate cardinality estimation, through information entropy dimension reduction, spatiotemporal feature extraction and Euclidean distance clustering algorithm, to achieve efficient clustering of trajectory segments; S40, dividing the ships into four categories according to their lengths, namely, ship length less than 50 meters, ship length 50 to 100 meters, ship length 100 to 150 meters, and ship length greater than or equal to 150 meters, and establishing a multi-dimensional ship feature tensor; S50, construct a composite wave disaster risk index model, including historical wave height data analysis, ship dynamic response feature extraction, sea condition complexity index calculation and comprehensive wave disaster risk assessment based on support vector machine; S60, statistics the spatiotemporal characteristics of the trajectories of different types of ships in each grid in the past five years, and construct a multi-dimensional sparse passage frequency tensor; S70, multi-dimensional feature fusion of the ship trajectory frequency tensor and the wave disaster risk assessment model, and use principal component decomposition and Bayesian probability graph model to construct a dynamic risk assessment hypergraph; S80. Based on multi-criteria dynamic thresholds, probabilistic cluster analysis is performed on the risk assessment hypergraph to identify key defense areas for wave disasters and establish a multi-level defense zone classification standard; S90. An improved meta-heuristic optimization algorithm is used to dynamically iteratively optimize the boundaries of the defense zone, minimize the overall disaster risk and computational complexity in the area, and obtain a robust delineation of key defense areas for wave disasters.

2. The method for optimizing the delineation of wave disaster defense zones based on ship AIS data according to claim 1 is characterized in that: The S10 specifically includes: Step 101: Obtain the geospatial data of the target sea area and determine the longitude and latitude range of the sea area; Step 102, dividing the target sea area into regular grids according to the accuracy of 0.025 degrees longitude and 0.025 degrees latitude; Step 103: Establish a grid coordinate mapping matrix to map each grid cell to a unique two-dimensional coordinate index; Step 104: verifying the spatial topological relationship of the grid coordinate mapping matrix to ensure the integrity and continuity of the grid division; Step 105: Record the boundary coordinates and spatial reference system of the grid coordinate mapping matrix.

3. The method for optimizing the delineation of wave disaster defense zones based on ship AIS data according to claim 2 is characterized in that: The S20 specifically includes: Step 201: Perform preliminary data cleaning on the ship navigation automatic identification system data to remove records with missing timestamps and incomplete location information; Step 202: Perform outlier detection on the data after preliminary cleaning, including data points of speed exceeding the limit and position jump; Step 203: based on the ship speed change rate threshold, segment the navigation trajectory and divide the trajectory segments with continuous abnormal speed changes into independent research units; Step 204: construct a trajectory envelope, identify and filter trajectory points beyond the envelope range; Step 205: Perform repeatability check and consistency marking on the segmented trajectory data.

4. The method for optimizing the delineation of wave disaster defense zones based on ship AIS data according to claim 3 is characterized in that: The S30 specifically includes: Step 301, preprocessing the trajectory segmentation data using a massive logarithmic approximate cardinality estimation algorithm; Step 302: extract key spatial features of the trajectory segment by using an information entropy dimension reduction method; Step 303, extracting the time, space and speed multi-dimensional feature vectors of the trajectory segment; Step 304: applying the Euclidean distance to calculate the similarity between the trajectory segments; Step 305: Based on the similarity of trajectory segment features, a clustering algorithm is executed to achieve efficient grouping of trajectory segments.

5. The method for optimizing the delineation of wave disaster defense zones based on ship AIS data according to claim 4 is characterized in that: The S40 specifically includes: Step 401, obtaining ship length measurement data; Step 402, classifying the ships according to four categories: ship length less than 50 meters, ship length 50 to 100 meters, ship length 100 to 150 meters, and ship length greater than or equal to 150 meters; Step 403: extracting the dynamic characteristics, navigation characteristics, and load characteristics of each ship category; Step 404: construct a multi-dimensional ship feature tensor, including ship length, draft, sailing speed, and load dimensions; Step 405: Standardize the multi-dimensional ship characteristic tensor to eliminate dimensional differences.

6. The method for optimizing the delineation of wave disaster defense zones based on ship AIS data according to claim 5 is characterized in that: The S50 specifically includes: Step 501: Collect historical wave height data and analyze wave height distribution patterns and statistical characteristics; Step 502: extracting dynamic response characteristics of the ship, including roll angle, pitch angle, and heave displacement; Step 503, calculating sea condition complexity indexes, including wave height, period, and wave direction change rate; Step 504: construct a support vector machine training data set, including marine environmental characteristics and historical disaster data; Step 505: Train the support vector machine model to achieve comprehensive assessment of wave disaster risks.

7. The method for optimizing the delineation of wave disaster defense zones based on ship AIS data according to claim 6 is characterized in that: The S60 specifically includes: Step 601, collecting data on ship navigation tracks in the past five years; Step 602: extracting spatiotemporal features according to ship type classification; Step 603, calculating the passage frequency of different types of ships in each grid; Step 604: construct a multi-dimensional sparse traffic frequency tensor, including time, space, and ship type dimensions; Step 605: normalize the passing frequency tensor.

8. The method for optimizing the delineation of wave disaster defense zones based on ship AIS data according to claim 7 is characterized in that: The S70 specifically includes: Step 701: align the ship trajectory frequency tensor with the wave disaster risk assessment model; Step 702: Apply principal component decomposition method to reduce the dimension of feature tensor; Step 703: construct a Bayesian probability graph model to establish a probability dependency relationship between features; Step 704: Calculate the node and edge weights of the dynamic risk assessment hypergraph; Step 705: Perform probability inference and risk prediction on the dynamic risk assessment hypergraph.

9. The method for optimizing the delineation of wave disaster defense zones based on ship AIS data according to claim 8, characterized in that: The S80 specifically includes: Step 801, defining multi-criteria dynamic thresholds, including disaster risk, traffic frequency, and ship type dimensions; Step 902: Performing probabilistic cluster analysis on the risk assessment hypergraph; Step 803: Identify high-risk areas for sea wave disasters; Step 804: Establish a multi-level defense zone classification standard and divide the risk levels; Step 805: Generate a map of key areas for defense against sea wave disasters.

10. The method for optimizing the delineation of wave disaster defense zones based on ship AIS data according to claim 9, characterized in that: The S90 specifically includes: Step 901: Select an improved meta-heuristic optimization algorithm, such as an improved particle swarm optimization algorithm or a simulated annealing algorithm; Step 902: setting an optimization objective function, including minimizing the overall disaster risk in the region and the computational complexity; Step 903, initializing a candidate solution set for the defense zone boundary; Step 904, iteratively update the defense zone boundary and evaluate each solution using the fitness function; Step 905: output the delineation result of the key defense area for sea wave disasters with the highest fitness.

Images