A meteorological risk level interval intelligent division method, device, equipment and medium

Abstract

This application discloses a method, apparatus, equipment, and medium for intelligently dividing meteorological risk level intervals, relating to the field of maritime navigation safety. The method includes: determining a meteorological risk factor index system based on meteorological regulations and maritime accident reports, using principles of ship dynamics and marine meteorology, combined with manual review and expert evaluation; classifying and quantifying the scores of each accident case; dividing meteorological intervals based on the risk thresholds corresponding to each meteorological risk factor and a multi-round iterative solicitation mechanism; clustering the meteorological intervals using a clustering algorithm, and calculating the average accident level of the meteorological cluster interval based on the quantified scores of the accident cases within that interval; optimizing the meteorological cluster intervals using a genetic algorithm based on the average accident level; and generating meteorological risk level intervals based on the optimized meteorological cluster intervals. This application improves the accuracy of intelligent division of meteorological risk level intervals.

Description

Technical Field

[0001] This application relates to the field of maritime navigation safety, and in particular to a method, device, equipment and medium for intelligently dividing meteorological risk level intervals. Background Technology

[0002] Marine weather conditions are one of the core factors affecting navigation safety. Severe weather conditions can easily lead to various maritime accidents such as ship collisions, groundings, and capsizing, causing significant casualties and property losses. The existing meteorological risk level classification technology has significant shortcomings, specifically: (1) The extraction dimension of risk factors is too single, resulting in incomplete factor coverage and easy omission of hidden risk factors; (2) It relies solely on expert experience or standard threshold judgments, resulting in strong subjectivity in interval division and easy to produce deviations; (3) It relies solely on expert experience or standard threshold judgments, resulting in a lack of data-driven interval optimization, which leads to insufficient accuracy of risk warnings and cannot meet the actual needs of navigation safety assessment.

[0003] Therefore, how to improve the accuracy of intelligent classification of meteorological risk level intervals and provide reliable technical support for maritime navigation safety early warning has become an urgent problem to be solved. Summary of the Invention

[0004] The purpose of this application is to provide a method, device, equipment, and medium for intelligently dividing meteorological risk level intervals, which can improve the accuracy of intelligent division of meteorological risk level intervals.

[0005] To achieve the above objectives, this application provides the following solution.

[0006] Firstly, this application provides a method for intelligently dividing meteorological risk level intervals, including the following steps.

[0007] Obtain meteorological regulations and maritime accident reports.

[0008] Based on the meteorological standard documents, meteorological risk factors are extracted to obtain standard factors. Based on the principles of ship dynamics and marine meteorology, meteorological risk factors that induce maritime accidents are explored to obtain theoretical supplementary factors.

[0009] Accident cases related to the normative factors and the theoretical supplementary factors are selected from the maritime accident reports. Based on the meteorological data corresponding to the accident cases, the normative factors and the theoretical supplementary factors are verified and analyzed through manual review and expert evaluation to determine the meteorological risk factor index system.

[0010] Each of the aforementioned accident cases is classified and quantified to determine its accident level and corresponding quantitative score.

[0011] Based on the risk thresholds corresponding to each meteorological risk factor and the multi-round iterative solicitation mechanism, the values ​​of each meteorological risk factor in the meteorological risk factor index system are divided into intervals to obtain the meteorological intervals corresponding to each meteorological risk factor.

[0012] For any meteorological risk factor, a clustering algorithm is used to cluster the meteorological intervals of that risk factor to obtain meteorological clustering intervals.

[0013] For any meteorological risk factor corresponding to any meteorological cluster interval, the average accident level of the meteorological cluster interval is calculated based on the level quantification score of the accident cases within that meteorological cluster interval.

[0014] For any meteorological risk factor, based on the average accident level, a genetic algorithm is used to optimize the meteorological clustering interval of the meteorological risk factor to obtain the optimized meteorological clustering interval.

[0015] The meteorological risk level range of each meteorological risk factor in the meteorological risk factor index system is generated based on the meteorological clustering optimization range; the meteorological risk level range is used for maritime navigation early warning and risk assessment.

[0016] In one embodiment, the values ​​of each meteorological risk factor in the meteorological risk factor index system are divided into intervals based on the risk thresholds corresponding to each meteorological risk factor and a multi-round iterative solicitation mechanism to obtain the meteorological intervals corresponding to each meteorological risk factor. Specifically, this includes: dividing the values ​​of each meteorological risk factor in the meteorological risk factor index system into intervals based on the risk thresholds corresponding to each meteorological risk factor to obtain the basic intervals corresponding to each meteorological risk factor; and adjusting the basic intervals based on the multi-round iterative solicitation mechanism to obtain the meteorological intervals corresponding to each meteorological risk factor in the meteorological risk factor index system.

[0017] In one embodiment, the basic intervals are adjusted based on a multi-round iterative consultation mechanism to obtain the meteorological intervals corresponding to each meteorological risk factor in the meteorological risk factor index system. Specifically, this includes: forming an expert team, conducting multiple rounds of anonymous consultation with each expert in the expert team, and obtaining the suggested values ​​given by each expert based on the basic intervals, the meteorological standard documents, the accident cases, and reference values; the suggested values ​​include: the number of basic intervals and the interval boundary thresholds; the reference values ​​are the suggested values ​​of other experts in the expert team besides the suggested expert; and obtaining the meteorological intervals corresponding to each meteorological risk factor in the meteorological risk factor index system based on the mean and dispersion of the suggested values.

[0018] In one embodiment, a clustering algorithm is used to cluster the meteorological intervals of the meteorological risk factor to obtain meteorological clustering intervals. Specifically, this includes: using the K-means clustering method to cluster the meteorological intervals of the meteorological risk factor to obtain meteorological clustering intervals; the number of clusters in the K-means clustering method is set according to the number of meteorological risk levels; the number of clusters is 4; the meteorological clustering intervals are used to determine the initial division boundaries of the 5 meteorological risk intervals.

[0019] In one embodiment, for any meteorological risk factor, based on the average accident level, a genetic algorithm is used to optimize the meteorological clustering interval of the meteorological risk factor to obtain the meteorological clustering optimization interval, which specifically includes the following steps.

[0020] The population is initialized by treating each segmentation point of the meteorological clustering interval as an individual; the population size is determined based on the total number of accident cases in the maritime accident report; the population size is 50.

[0021] Based on the initial population, multiple genetic operations and elite retention are performed with the goal of maximizing the fitness function value, and the optimal population is obtained after reaching the set maximum number of iterations; wherein, the maximum number of iterations is set according to the convergence stability requirements of the meteorological clustering interval; the maximum number of iterations is 100 generations.

[0022] The optimal interval is generated based on the split point corresponding to each individual in the optimal population, and the effective interval is selected as the meteorological clustering optimization interval based on the optimal interval; the effective interval includes: the interval in the optimal interval where the proportion of accident cases is greater than a set proportion, the set proportion is set according to the sample statistical significance requirements; the set proportion is 1%.

[0023] The fitness function value is calculated based on the absolute value of the Pearson correlation coefficient between the median of the meteorological cluster interval and the corresponding average accident level at each iteration, the sample balance performance score of the meteorological cluster interval, and the set compliance score; the sample balance performance score of the meteorological cluster interval is the normalized value of the inverse of the variance of the proportion of accident cases in the meteorological cluster interval.

[0024] In one embodiment, the fitness function value is calculated using the following formula.

[0025] F=0.9×r_score+0.05×α+0.05×β.

[0026] Where F represents the fitness function value; r_score represents the absolute value of the Pearson correlation coefficient between the median of the meteorological cluster interval and the corresponding average accident level; α represents the sample balance performance score of the meteorological cluster interval; and β represents the setting compliance score.

[0027] Among them, r_score is required to pass the significance test, and the significance level of r_score needs to be less than 0.01; otherwise, r_score is counted as 0. α is obtained by taking the reciprocal of the variance of the proportion of accident cases and then performing Min-Max normalization.

[0028] In one embodiment, the meteorological risk level intervals for each meteorological risk factor in the meteorological risk factor index system are generated based on the meteorological clustering optimization intervals. Specifically, this includes: for any meteorological risk factor, each meteorological clustering optimization interval is taken as a meteorological risk level interval; the meteorological risk level intervals include five levels: slight risk, low risk, moderate risk, high risk, and high risk, and the meteorological parameter threshold range corresponding to each level is determined by the boundary values ​​of the meteorological clustering optimization intervals.

[0029] Secondly, this application provides an intelligent meteorological risk level interval division device, which includes the following modules.

[0030] The document acquisition module is used to acquire meteorological standard documents and maritime accident reports.

[0031] The risk factor extraction module is used to extract meteorological risk factors based on the meteorological standard documents to obtain standard factors, and to mine meteorological risk factors that induce maritime accidents based on the principles of ship dynamics and marine meteorology to obtain theoretical supplementary factors.

[0032] The risk factor indicator system construction module is used to screen accident cases related to the normative factors and the theoretical supplementary factors from the maritime accident reports, and to verify and analyze the normative factors and the theoretical supplementary factors based on the meteorological data corresponding to the accident cases, combined with manual review and expert evaluation, to determine the meteorological risk factor indicator system.

[0033] The accident level quantification module is used to classify and quantify the accident cases, and determine the accident level and corresponding quantification score of each accident case.

[0034] The meteorological interval division module is used to divide the value of each meteorological risk factor in the meteorological risk factor index system into intervals based on the risk threshold corresponding to each meteorological risk factor and the multi-round iterative solicitation mechanism, so as to obtain the meteorological interval corresponding to each meteorological risk factor.

[0035] The meteorological interval clustering module is used to cluster meteorological intervals for any meteorological risk factor using a clustering algorithm, thereby obtaining meteorological cluster intervals.

[0036] The interval risk value calculation module is used to calculate the average accident level of any meteorological cluster interval corresponding to any meteorological risk factor, based on the level quantification score of accident cases within the meteorological cluster interval.

[0037] The interval optimization module is used to optimize the meteorological clustering interval of any meteorological risk factor based on the average accident level using a genetic algorithm, thereby obtaining the meteorological clustering optimization interval.

[0038] The meteorological risk level interval division module is used to generate the meteorological risk level interval of each meteorological risk factor in the meteorological risk factor index system based on the meteorological clustering optimization interval; the meteorological risk level interval is used for maritime navigation early warning and risk assessment.

[0039] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the computer program to implement the intelligent division method for meteorological risk level intervals as described above.

[0040] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the intelligent division method for meteorological risk level intervals described above.

[0041] According to the specific embodiments provided in this application, this application has the following technical effects: This application provides a method, device, equipment, and medium for intelligently dividing meteorological risk level intervals. It extracts meteorological risk-causing factors through a triple fusion of standards (meteorological standard documents), theory (principles of ship dynamics and marine meteorology), and empirical evidence (manual verification and expert review), covering both explicit and implicit risk points. This avoids the limitations of single-dimensional extraction and ensures the integrity and scientific nature of the factor system. Meteorological intervals are divided based on the risk thresholds corresponding to each meteorological risk-causing factor and the Delphi method, achieving dual verification through expert experience and data support, reducing subjective judgment errors. Intelligent optimization of meteorological clustering intervals using a genetic algorithm further enhances the correlation between intervals and accident risks, ensuring the accuracy of risk level division. Therefore, this application improves the accuracy of intelligent division of meteorological risk level intervals, providing a reliable basis for maritime navigation safety early warning. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 This is a flowchart illustrating a method for intelligently dividing meteorological risk level intervals, as provided in an embodiment of this application.

[0044] Figure 2 This is a schematic diagram of the functional modules of a meteorological risk level interval intelligent division device provided in an embodiment of this application.

[0045] Figure 3 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0046] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0047] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0048] This application addresses the shortcomings of current meteorological risk level classification technologies by providing a method, device, equipment, and medium for intelligent classification of meteorological risk level intervals. It achieves intelligent classification of meteorological risk level intervals based on the fusion of maritime accident statistics and meteorological data. Through multi-source fusion, data-driven approaches, and intelligent optimization, it achieves accurate matching between meteorological factors and accident risks, providing reliable technical support for maritime navigation safety early warning.

[0049] In one exemplary embodiment, such as Figure 1 As shown, a method for intelligently dividing meteorological risk level intervals is provided, including: Step 101: Construct a meteorological risk factor indicator system.

[0050] This step specifically includes: obtaining meteorological standard documents and maritime accident reports; extracting meteorological risk factors based on the meteorological standard documents to obtain standard factors; mining meteorological risk factors that induce maritime accidents based on the principles of ship dynamics and marine meteorology to obtain theoretical supplementary factors; selecting accident cases related to the standard factors and theoretical supplementary factors from the maritime accident reports; and verifying and analyzing the standard factors and theoretical supplementary factors based on the meteorological data corresponding to the accident cases, combined with manual review and expert evaluation, to determine the meteorological risk factor indicator system.

[0051] Step 102: Classify and quantify the accident cases, and determine the accident level and corresponding quantitative score of each accident case.

[0052] Step 103: Based on the risk thresholds corresponding to each meteorological risk factor and the multi-round iterative solicitation mechanism, divide the value of each meteorological risk factor in the meteorological risk factor index system into intervals to obtain the meteorological intervals corresponding to each meteorological risk factor.

[0053] Step 104: For any meteorological risk factor, a clustering algorithm is used to cluster the meteorological intervals of the meteorological risk factor to obtain the meteorological clustering intervals.

[0054] Step 105: For any meteorological risk factor corresponding to any meteorological cluster interval, calculate the average accident level of the meteorological cluster interval based on the level quantification score of the accident cases within the meteorological cluster interval.

[0055] Step 106: For any meteorological risk factor, based on the average accident level, a genetic algorithm is used to optimize the meteorological clustering interval of the meteorological risk factor to obtain the optimized meteorological clustering interval.

[0056] Step 107: Generate the meteorological risk level intervals for each meteorological risk factor in the meteorological risk factor index system based on the meteorological clustering optimization intervals; the meteorological risk level intervals are used for maritime navigation early warning and risk assessment.

[0057] In another exemplary embodiment of this application, step 103 specifically includes: dividing the value of each meteorological risk factor in the meteorological risk factor index system into intervals according to the risk threshold corresponding to each meteorological risk factor, to obtain the basic interval corresponding to each meteorological risk factor; adjusting the basic interval based on a multi-round iterative solicitation mechanism to obtain the meteorological interval corresponding to each meteorological risk factor in the meteorological risk factor index system.

[0058] Specifically, the adjustment of the basic intervals based on a multi-round iterative consultation mechanism to obtain the meteorological intervals corresponding to each meteorological risk factor in the meteorological risk factor index system includes: forming an expert team, conducting multiple rounds of anonymous consultations with each expert in the expert team, and obtaining the suggested values ​​given by each expert based on the basic intervals, the meteorological standard documents, the accident cases, and reference values; the suggested values ​​include: the number of basic intervals and the interval boundary thresholds; the reference values ​​are the suggested values ​​of other experts in the expert team besides the suggested expert; and obtaining the meteorological intervals corresponding to each meteorological risk factor in the meteorological risk factor index system based on the mean and dispersion of the suggested values.

[0059] In another exemplary embodiment of this application, step 104 specifically includes: using the K-means clustering method to cluster the meteorological intervals of the meteorological risk factor to obtain meteorological clustering intervals; the number of clusters in the K-means clustering method is set according to the number of meteorological risk levels; the number of clusters is 4; the meteorological clustering intervals are used to determine the initial division boundaries of the 5 meteorological risk intervals.

[0060] In another exemplary embodiment of this application, step 106 specifically includes the following steps.

[0061] The population is initialized by treating each segmentation point of the meteorological clustering interval as an individual; the population size is determined based on the total number of accident cases in the maritime accident report; the population size is 50.

[0062] Based on the initial population, multiple genetic operations and elite retention are performed with the goal of maximizing the fitness function value, and the optimal population is obtained after reaching the set maximum number of iterations; wherein, the maximum number of iterations is set according to the convergence stability requirements of the meteorological clustering interval; the maximum number of iterations is 100 generations.

[0063] The optimal interval is generated based on the split point corresponding to each individual in the optimal population, and the effective interval is selected as the meteorological clustering optimization interval based on the optimal interval; the effective interval includes: the interval in the optimal interval where the proportion of accident cases is greater than a set proportion, the set proportion is set according to the sample statistical significance requirements; the set proportion is 1%.

[0064] The fitness function value is calculated based on the absolute value of the Pearson correlation coefficient between the median of the meteorological cluster interval and the corresponding average accident level at each iteration, the sample balance performance score of the meteorological cluster interval, and the set compliance score; the sample balance performance score of the meteorological cluster interval is the normalized value of the inverse of the variance of the proportion of accident cases in the meteorological cluster interval.

[0065] The formula for calculating the fitness function value is as follows.

[0066] F=0.9×r_score+0.05×α+0.05×β.

[0067] Where F represents the fitness function value; r_score represents the absolute value of the Pearson correlation coefficient between the median of the meteorological cluster interval and the corresponding average accident level (after significance testing). r_score must pass the significance test, and the significance level p of r_score must be less than 0.01 (1% significance level) to be included in the score; otherwise, r_score is counted as 0; α represents the sample balance performance score of the meteorological cluster interval. α is obtained by taking the reciprocal of the variance of the accident case proportion and then performing Min-Max normalization. The smaller the variance (the more balanced the distribution), the higher the score; β represents the compliance score, which can be fixed at 1.0, with a reserved interface.

[0068] In another exemplary embodiment of this application, step 107 specifically includes: for any meteorological risk factor, treating each of the meteorological clustering optimization intervals as a meteorological risk level interval. The meteorological risk level intervals include five levels: slight risk, low risk, moderate risk, high risk, and high risk, and the meteorological parameter threshold range corresponding to each level is determined by the boundary values ​​of the meteorological clustering optimization intervals.

[0069] This application addresses the problem of a single dimension in risk factor extraction in existing technologies by constructing a triple fusion mechanism of "standard extraction - theoretical supplementation - empirical verification". Building upon this, the application further establishes a progressive optimization chain of "multi-round iterative solicitation - clustering - genetic optimization", with each step forming a synergistic effect: First, the multi-round iterative solicitation mechanism transforms expert experience into a quantifiable statistical convergence process, providing initial interval boundaries with domain knowledge support for subsequent clustering, avoiding physically meaningless interval divisions that may occur with purely data-driven methods; second, the clustering algorithm automatically classifies the intervals refined by experts, preserving the rationality of expert knowledge while providing a high-quality initial population for the genetic algorithm, solving the problems of high randomness and slow convergence in the initial population of genetic algorithms in existing technologies; finally, the genetic algorithm optimizes by maximizing the Pearson correlation coefficient between the average accident level and the median of the meteorological interval, while introducing sample balance constraints, achieving precise matching between meteorological interval division and accident risk level, solving the problem of insufficient risk warning accuracy caused by relying solely on standard thresholds or expert experience in existing technologies. Through a closed-loop process of "expert knowledge guiding data clustering, data clustering optimizing the initial solution of the algorithm, and algorithm optimization feeding back into the interval division," the technical problems of strong subjectivity and weak correlation with accident risk in marine meteorological risk level classification are jointly solved, resulting in a synergistic and effective technical outcome. Compared with existing technologies that rely solely on standardized thresholds or expert experience, the embodiments of this application significantly improve the accuracy and objectivity of meteorological risk level interval division through data-driven intelligent optimization.

[0070] The following section provides a detailed description of a specific implementation process of the intelligent meteorological risk level interval division method described in the above embodiments in a practical application.

[0071] Step 1: Extraction of meteorological risk factors based on multi-source information fusion.

[0072] This step constructs a complete indicator system of meteorological risk factors through three dimensions: standardized analysis, theoretical supplementation, and empirical verification, as detailed below.

[0073] (1) Extraction of normative factors: The system analyzes relevant conventions and guidelines of the International Maritime Organization (IMO), including the International Regulations for Preventing Collisions at Sea (COLREGs), Chapter V and Chapter II-1 of the International Convention for the Safety of Life at Sea (SOLAS), and MSC.1 / Circ.1228 Guidelines for Severe Weather; at the same time, it analyzes relevant normative standards, such as the General Design Code for Seaports (JTS165-2013) and the Guidelines for Navigation Safety in Severe Weather. The text analysis method is used to extract the explicit or implicit meteorological hazards and risk thresholds in the above normative documents. Meteorological hazards include, but are not limited to, wind speed, visibility, wave height, swell, ocean currents, sea ice, and tropical cyclones.

[0074] (2) Theoretical supplementary factors: Based on the core principles of ship dynamics and marine meteorology, following the causal chain of “environmental load-ship response-accident consequences”, we will deeply explore the intrinsic mechanism of meteorological conditions inducing maritime accidents, further refine the classification system of meteorological risk factors, and clarify the risk threshold standards corresponding to various factors.

[0075] (3) Empirical and expert verification: In order to further construct a scientific and reliable index system of meteorological risk factors, this embodiment adopts a combination of structured data extraction and manual verification to carry out empirical analysis. The specific process is as follows.

[0076] 1) Preliminary screening and case focus. A preliminary screening was conducted on maritime accident reports collected over the past 20 years from around the world and coastal areas of China, identifying accident cases clearly related to meteorological factors. Screening criteria included: accident reports directly mentioning the impact of meteorological conditions, recording specific meteorological parameters, or determining the cause of the accident as being related to meteorological elements such as wind, waves, fog, and heavy rain. This step eliminated irrelevant cases, ensuring the focus of subsequent analysis.

[0077] 2) Identify the core extraction fields. Based on the filtered cases, determine the three types of key information to be extracted: ① Basic ship information: such as ship type, tonnage, navigation status, company, etc.; ② Meteorological information: such as meteorological parameter type, specific value or level, data source, etc.; ③ Core accident information: such as the time of the accident, precise geographical location, accident type and consequences, etc.

[0078] 3) Implement structured data extraction and cleaning. Using the Python programming language and the Pandas library, the accident cases selected in step 1) were processed and cleaned. Tools such as BeautifulSoup were used to parse unstructured reports in PDF and webpage formats, enabling batch extraction of core fields. Simultaneously, manual verification of the three types of key extracted information was conducted, focusing on checking the accuracy of meteorological parameters and the completeness of field matching, correcting extraction deviations caused by semantic ambiguity or format inconsistencies. Based on this, an accident-meteorological correlation database was constructed, integrating basic ship information, core accident information, and their corresponding meteorological information (i.e., meteorological data) to form a structured dataset.

[0079] 4) Conduct expert review and system improvement. Based on the meteorological risk factors extracted in the early stage and the structured dataset in step 3), a preliminary meteorological risk factor set is constructed and submitted to the expert group for review. The set is verified from both the aspects of completeness and scientificity, and potential omission factors are identified and supplemented. Finally, a complete and reliable meteorological risk factor indicator system is formed.

[0080] Step 2: Accident level classification and quantitative scoring.

[0081] According to China's "Statistical Measures for Maritime Traffic Accidents", maritime accidents are classified into four levels according to their severity, and each level is assigned a corresponding quantitative score. The specific classification standards are as follows: (1) Particularly serious accidents: causing more than 30 deaths or missing persons, or more than 100 serious injuries, or direct economic losses ≥ RMB100 million, are assigned 4 points; (2) Major accidents: causing 10–30 deaths or missing persons, or 50–100 serious injuries, or direct economic losses of RMB50 million–100 million, are assigned 3 points; (3) Relatively serious accidents: causing 3–10 deaths or missing persons, or 10–50 serious injuries, or direct economic losses of RMB10 million–50 million, are assigned 2 points; (4) General accidents: causing 1–3 deaths or missing persons, or 1–10 serious injuries, or direct economic losses < RMB10 million, are assigned 1 point.

[0082] When the same accident meets multiple classification criteria, its final accident level and corresponding quantitative score shall be determined according to the principle of "choosing the higher level".

[0083] Step 3: Basic interval division of meteorological factors.

[0084] Based on the threshold requirements for various meteorological parameters in the "General Design Code for Seaports" (JTS165-2013) and the IMO MSC.1 / Circ.1228 "Guidelines for Severe Weather", the basic intervals for each meteorological risk factor are determined by adopting the "strict principle" (i.e., selecting the stricter threshold as the basis for interval division), thus forming a structured table of basic intervals for meteorological factors.

[0085] For example, taking average wind speed as an example, its basic range is divided as follows: (1) 0-5.4 m / s (corresponding to wind force 0-3): determined according to the wind force limit for routine operations in the "General Design Code for Seaports". (2) 5.5-10.7 m / s (corresponding to wind force 4-5): determined according to the wind tilt moment calculation benchmark in IMO MSC.267 document. (3) 10.8-17.1 m / s (corresponding to wind force 6-7): determined according to the "surf-swing" risk threshold in IMO MSC.1 / Circ.1228 "Guide to Severe Weather". (4) 17.2-24.4 m / s (corresponding to wind force 8-9): determined according to the breakwater overtopping design wind speed in the "General Design Code for Seaports". (5) ≥24.5 m / s (corresponding to wind force 10 and above): determined according to the emergency departure wind speed requirements for dangerous goods vessels.

[0086] Step 4: Refinement of meteorological intervals by experts based on a multi-round iterative consultation mechanism.

[0087] An expert consensus mechanism is built through a multi-round iterative consultation mechanism, and the basic interval obtained in step 3 is refined and adjusted. The specific process is as follows.

[0088] (1) Expert team formation: Form a cross-disciplinary expert team of no less than 20 people. The team members include senior captains, ship structure engineers, maritime safety researchers, port scheduling experts, etc. All members have more than 5 years of experience in maritime navigation safety or maritime management.

[0089] (2) Multiple rounds of anonymous solicitation and feedback: A multi-round iterative mechanism of "anonymous solicitation - result statistics - feedback correction" is adopted to submit the basic interval table of meteorological factors and related supporting materials (including the original text of the standard, accident statistics, etc.) to the expert team. The experts make detailed adjustment suggestions on the number of intervals and boundary thresholds of each meteorological factor.

[0090] (3) Consensus formation: After each round of consultation, the expert suggestions are statistically analyzed, and the mean and dispersion of the suggested values ​​are calculated (the dispersion is measured by the coefficient of variation). The statistical results are fed back to the experts for their reference and adjustment. The process is repeated until the coefficient of variation of the expert suggestions is ≤10%, thus forming an initial refined meteorological interval system under the expert consensus.

[0091] Step 5: Accident-meteorological data matching and interval risk value calculation.

[0092] The historical accident samples in the accident-meteorological correlation database established in step 1 are classified into the initial refined meteorological intervals obtained in step 4 according to the actual meteorological observation values ​​at the time of their occurrence. For each meteorological interval, the average accident level score is calculated as the initial risk value of that interval. The calculation uses the following formula.

[0093] .

[0094] in, For the first The average accident level (i.e., initial risk value) for each meteorological interval; For the first The graded quantitative score of the kth accident case (sample) within a meteorological interval; For the first The number of accident cases included within a weather interval.

[0095] Step 6: Correlation analysis between meteorological interval clustering and accident level.

[0096] To establish a reasonable risk range for meteorological parameters, multiple meteorological ranges were first divided based on expert experience. Then, a clustering algorithm was used to automatically classify and optimize the range boundaries to obtain meteorological clustering ranges.

[0097] The specific process is as follows: Based on the initial interval boundaries set by experts, the K-means clustering method (number of clusters n_clusters=4, random seed fixed) is used to generate 4 cluster centers, which serve as the initial boundaries for the division of 5 meteorological risk intervals.

[0098] To further evaluate the rationality of the clustering, a correlation assessment index is introduced. Specifically, the Pearson correlation coefficient between the median of each interval and the mean of the corresponding accident level is calculated as a correlation assessment index to preliminarily determine the statistical correlation strength between the interval division and the risk level.

[0099] The median value of each meteorological interval The independent variable (representing the quantitative level of meteorological parameters) is the average accident level of the interval calculated after clustering. As the dependent variable (characterizing the degree of accident risk under this meteorological level), the Pearson correlation coefficient is used to calculate the degree of linear correlation between the two. The formula for calculating the Pearson correlation coefficient r is as follows.

[0100] .

[0101] Where n is the total number of meteorological intervals; if the calculated Pearson correlation coefficient r does not reach the preset significance level (e.g., significance level p<0.01), it indicates that the current meteorological interval division is not sufficiently correlated with accident risk.

[0102] Step 7: Intelligent optimization of meteorological intervals based on genetic algorithms.

[0103] With maximizing the Pearson correlation coefficient as the core optimization objective, and combined with constraints on the rationality of the number of intervals and the balance of sample distribution, a genetic algorithm is used to intelligently optimize the initial refined meteorological intervals to obtain the meteorological clustering optimization intervals. The specific steps are as follows.

[0104] (1) Individual coding and population initialization: Fixed-length real number coding is adopted, with each individual corresponding to 4 fixed split points, representing the division of 5 meteorological intervals. The population size is set to 50. The normal distribution perturbation (standard deviation 0.8) is added to the 4 split points obtained by K-means clustering as the center for initialization to ensure that all split points are within a reasonable wind speed range of [0.0, 50.0].

[0105] (2) Fitness function design: Construct a weighted fitness function F=0.9×r_score+0.05×α+0.05×β to evaluate the quality of individuals.

[0106] (3) Valid interval screening: Before calculating fitness, valid intervals with a sample proportion of ≥5% are screened out, and the correlation coefficient and balance score are calculated only based on the valid intervals. If there are less than 2 valid intervals, the fitness is set to 0.

[0107] (4) Genetic manipulation design.

[0108] Selection operation: The selection strategy combines elite retention and roulette wheel selection. The fitness function values ​​are compared, and the top 10% (5) of the best individuals are retained and directly enter the next generation. For the remaining individuals, the algorithm calculates their selection probability based on their fitness function values ​​and then selects them through random sampling.

[0109] Crossover operation: A single-point crossover strategy is adopted, and the split point sequences of two parent individuals are crossed at random positions. After the crossover, the sequences are sorted and deduplicated, ensuring that the number of split points is 4.

[0110] Mutation operation: Randomly select a split point in an individual with a 5% mutation probability, and regenerate within the range of [0.0, 50.0]. The split point remains ordered after mutation.

[0111] (5) Elite retention and iteration termination: The five elite individuals with the highest fitness in each generation are retained and directly enter the next generation to ensure that superior genes are not lost. The iteration termination condition is reaching the set maximum number of iterations (such as the preset 100 generations).

[0112] (6) Final interval generation: After the algorithm converges, five intervals are generated from the four split points of the best individual, and the effective intervals with a sample ratio of ≥5% are selected as the final partitioning scheme. If there are fewer than five effective intervals, some low-ratio intervals are added to meet the output requirements of five intervals.

[0113] Step 8: Optimal meteorological risk level range is generated.

[0114] The intervals obtained from the genetic algorithm optimization results are directly mapped to five risk levels (e.g., interval 1 corresponds to risk level 1, interval 2 corresponds to risk level 2, ..., interval 5 corresponds to risk level 5), forming a risk classification table for maritime navigation safety early warning. A higher Pearson correlation coefficient indicates a closer alignment between the risk level and the actual situation. Data shows that the optimized risk level intervals are more accurate than the unoptimized ones.

[0115] Based on this scheme, a meteorological risk classification table is constructed, which uniformly divides all meteorological risk factors into five levels: "minor risk, low risk, general risk, high risk, and high risk". The threshold range of meteorological parameters corresponding to each level is clearly defined, which is used for maritime navigation safety early warning and dynamic risk assessment.

[0116] This embodiment achieves intelligent division of meteorological risk level intervals based on the fusion of maritime accident statistics and meteorological data. It solves the problems of strong subjectivity, incomplete coverage of risk-causing factors, and weak correlation with accident risks in the current maritime meteorological risk level interval division, and improves the scientificity, accuracy and practicality of risk division.

[0117] The following is a specific example to illustrate this.

[0118] Example 1: Dividing risk level ranges based on wind speed.

[0119] (1) Extraction of wind speed-related risk factors. By analyzing the IMO SOLAS Convention and the "Code for General Design of Seaports", wind speed was identified as the core risk factor. Combined with ship dynamics theory, wind speed change rate was added as an auxiliary risk factor. 188 maritime accident reports from the Chinese coast were collected, and wind speed data at the time of the accidents was extracted in a structured manner. After review and confirmation by 5 experts, wind speed was identified as a key risk factor and included in the indicator system.

[0120] (2) Quantitative scoring of accident levels.

[0121] The 188 accidents were graded and scored according to the "Statistical Method for Water Traffic Accidents". Among them, there were 8 particularly serious accidents (4 points), 26 serious accidents (3 points), 72 major accidents (2 points), and 82 general accidents (1 point).

[0122] (3) Basic wind speed interval division.

[0123] Based on the strict principles of relevant regulations, the basic wind speed ranges are determined as follows: 0-5.4m / s, 5.5-10.7m / s, 10.8-17.1m / s, 17.2-24.4m / s, and ≥24.5m / s.

[0124] (4) Experts refine the interval.

[0125] A team of 22 experts was assembled, and after three rounds of iterative consultation, the wind speed ranges were refined into nine categories: 0-3.0 m / s, 3.1-5.4 m / s, 5.5-8.0 m / s, 8.1-10.7 m / s, 10.8-14.0 m / s, 14.1-17.1 m / s, 17.2-20.0 m / s, 20.1-24.4 m / s, and ≥24.5 m / s.

[0126] (5) Data matching and risk value calculation.

[0127] The 188 accidents were categorized into 9 intervals based on wind speed, and the average accident level for each interval was calculated as follows: 0-3.0 m / s (y1=1.02), 3.1-5.4 m / s (y2=1.05), 5.5-8.0 m / s (y3=1.18), 8.1-10.7 m / s (y4=1.35), 10.8-14.0 m / s (y5=1.82), 14.1-17.1 m / s (y6=2.45), 17.2-20.0 m / s (y7=2.98), 20.1-24.4 m / s (y8=3.52), and ≥24.5 m / s (y9=3.86).

[0128] (6) Expert interval analysis and cluster initialization.

[0129] Based on the eight expert-refined segmentation points (3.0, 5.4, 8.0, 10.7, 14.0, 17.1, 20.0, 24.4), the K-means clustering algorithm (n_clusters=4, random seed 42) was applied to compress them into four cluster centers as initial segmentation points, forming five initial wind speed intervals. The Pearson correlation coefficient r between the median of the expert-refined intervals and the mean of the corresponding accident level was calculated to be 0.9712, reaching a significant correlation level. However, an optimization process was still initiated to improve sample balance.

[0130] (7) Genetic algorithm optimization.

[0131] The population size was set to 50, with a maximum iteration of 100 generations. K-means clustering results were used as the initial population base, and interval partitioning was optimized through selection, crossover, and mutation operations. The fitness function comprehensively considered the correlation coefficient between the interval median and the accident level (weight 0.9), the balance of sample proportions in each interval (weight 0.05), and compliance with regulations (weight 0.05).

[0132] (8) Risk level generation.

[0133] After optimization using a genetic algorithm, the optimal five wind speed range division schemes with a fitness F=0.90 (correlation coefficient r=0.93, balance α=1.0, compliance β=1.0) were obtained. The sample proportion of each range was calculated, and all met the sample size requirement of ≥5%. The five optimized wind speed ranges were mapped to five risk levels: slight risk (0-5.4m / s), low risk (5.5-10.7m / s), moderate risk (10.8-17.1m / s), high risk (17.2-24.4m / s), and high risk (≥24.5m / s), forming the final wind speed risk classification table.

[0134] It should be noted that the above example uses wind speed as an example to illustrate the technical solution of this application, but the method of this application is also applicable to other meteorological hazards (such as visibility, wave height, ocean currents, swells, etc.). When applied to other meteorological hazards, only the following should be adjusted according to the characteristics of the corresponding factors: The basic interval division threshold in step (3) (according to the corresponding specifications); the initial interval setting for multi-round iterative solicitation in step (4); the meteorological parameter type of the accident case when matching data in step (5); the input data range of clustering and genetic algorithm in steps (6)-(7) (such as the range of visibility values ​​being 0-10 nautical miles, the range of wave height values ​​being 0-10 meters, etc.).

[0135] Apart from the parameter adjustments mentioned above, the core algorithm process of this application (indicator system construction, interval partitioning, clustering, genetic optimization, and risk level generation) remains unchanged.

[0136] This embodiment has significant advantages compared to the following three types of defects currently existing.

[0137] Three types of defects: (1) The extraction of risk factors is singular. Most technologies rely only on regulations, standards or single theoretical derivations and do not integrate empirical accident data, resulting in incomplete factor coverage and omission of hidden risk factors; (2) The division of intervals is highly subjective. Existing technologies either rely only on expert experience judgment or only on standard thresholds, lacking a dual verification mechanism of "expert consensus + data support", which is prone to deviation; (3) The interval optimization lacks data drive. It has not established a quantitative correlation between meteorological intervals and accident risks, making it difficult to improve the objectivity of experience judgment through historical data, ultimately resulting in insufficient accuracy of risk warning and failure to meet the actual needs of navigation safety assessment.

[0138] Advantages of this embodiment: (1) Comprehensiveness of risk factors: Meteorological risk factors are extracted through the triple integration of norms, theories and empirical evidence, covering both explicit and implicit risk points, avoiding the limitations of single-dimensional extraction, and ensuring the integrity and scientific nature of the factor system. (2) Objectivity of interval division: Combining a multi-round iterative consultation mechanism and historical accident data, the dual verification of "expert experience + data support" is achieved, reducing subjective judgment errors; through intelligent optimization of genetic algorithms, the correlation between intervals and accident risks is further improved. (3) Accuracy of risk assessment: A quantitative correlation model between meteorological intervals and accident levels is established, and the accuracy of risk level division is ensured through correlation coefficient verification and multi-constraint optimization, providing a reliable basis for maritime navigation safety early warning. (4) Strong practicality: The final five-level risk level system is simple and intuitive, and can be directly applied to scenarios such as ship navigation decision-making, port scheduling, and maritime supervision, supporting dynamic risk assessment and improving the level of maritime safety management.

[0139] Based on the same inventive concept, this application also provides a meteorological risk level interval intelligent division device for implementing the above-mentioned intelligent division method for meteorological risk level intervals. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more embodiments of the meteorological risk level interval intelligent division device provided below can be found in the limitations of the meteorological risk level interval intelligent division method above, and will not be repeated here.

[0140] In one exemplary embodiment, such as Figure 2 As shown, a smart device for dividing meteorological risk level intervals is provided, which includes the following modules.

[0141] Document acquisition module 201 is used to acquire meteorological standard documents and maritime accident reports.

[0142] The risk factor extraction module 202 is used to extract meteorological risk factors according to the meteorological standard documents to obtain standard factors, and to mine meteorological risk factors that induce maritime accidents based on the principles of ship dynamics and marine meteorology to obtain theoretical supplementary factors.

[0143] The risk factor indicator system construction module 203 is used to screen accident cases related to the normative factors and the theoretical supplementary factors from the maritime accident report, and to verify and analyze the normative factors and the theoretical supplementary factors based on the meteorological data corresponding to the accident cases, combined with manual review and expert evaluation, to determine the meteorological risk factor indicator system.

[0144] The accident level quantification module 204 is used to classify and quantify the accident cases, and determine the accident level and corresponding quantification score of each accident case.

[0145] The meteorological interval division module 205 is used to divide the value of each meteorological risk factor in the meteorological risk factor index system into intervals according to the risk threshold corresponding to each meteorological risk factor and the multi-round iterative solicitation mechanism, so as to obtain the meteorological intervals corresponding to each meteorological risk factor.

[0146] The meteorological interval clustering module 206 is used to cluster the meteorological intervals of any meteorological risk factor using a clustering algorithm to obtain the meteorological clustering intervals.

[0147] The interval risk value calculation module 207 is used to calculate the average accident level of any meteorological cluster interval corresponding to any meteorological risk factor, based on the level quantification score of accident cases within the meteorological cluster interval.

[0148] The interval optimization module 208 is used to optimize the meteorological clustering interval of any meteorological risk factor based on the average accident level using a genetic algorithm, thereby obtaining the meteorological clustering optimization interval.

[0149] The meteorological risk level interval division module 209 is used to generate the meteorological risk level interval of each meteorological risk factor in the meteorological risk factor index system based on the meteorological clustering optimization interval; the meteorological risk level interval is used for maritime navigation early warning and risk assessment.

[0150] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 3 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs in the non-volatile storage media to run. The database stores the meteorological risk level ranges for each meteorological risk factor in the meteorological risk factor index system. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements an intelligent method for classifying meteorological risk level ranges.

[0151] Those skilled in the art will understand that Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer equipment to which the present application is applied. Specific computer equipment may include, for example, [the following is a list of possible additional structures]. Figure 3 The embodiments show more or fewer components, combinations of certain components, or different component arrangements. In one exemplary embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, which the processor executes to implement the steps in the above-described method embodiments.

[0152] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0153] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0154] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0155] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0156] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, etc., and are not limited to these.

[0157] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0158] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for intelligently dividing meteorological risk level intervals, characterized in that, The intelligent method for dividing meteorological risk level intervals includes: Obtain meteorological regulations and maritime accident reports; Meteorological risk factors are extracted from the meteorological standard documents to obtain standard factors. Based on the principles of ship dynamics and marine meteorology, meteorological risk factors that induce maritime accidents are explored to obtain theoretical supplementary factors. Accident cases related to the normative factors and the theoretical supplementary factors are selected from the maritime accident reports. Based on the meteorological data corresponding to the accident cases, the normative factors and the theoretical supplementary factors are verified and analyzed by combining manual review and expert evaluation to determine the meteorological risk factor index system. Each of the aforementioned accident cases is classified and quantified to determine the accident level and corresponding quantitative score for each accident case. Based on the risk thresholds corresponding to each meteorological risk factor and the multi-round iterative solicitation mechanism, the values ​​of each meteorological risk factor in the meteorological risk factor index system are divided into intervals to obtain the meteorological intervals corresponding to each meteorological risk factor. For any meteorological risk factor, a clustering algorithm is used to cluster the meteorological intervals of the meteorological risk factor to obtain meteorological cluster intervals; For any meteorological risk factor corresponding to any meteorological cluster interval, the average accident level of the meteorological cluster interval is calculated based on the level quantification score of the accident cases within the meteorological cluster interval. For any meteorological risk factor, based on the average accident level, a genetic algorithm is used to optimize the meteorological clustering interval of the meteorological risk factor to obtain the optimized meteorological clustering interval. The meteorological risk level range of each meteorological risk factor in the meteorological risk factor index system is generated based on the meteorological clustering optimization range; the meteorological risk level range is used for maritime navigation early warning and risk assessment.

2. The intelligent method for dividing meteorological risk level intervals according to claim 1, characterized in that, Based on the risk thresholds corresponding to each meteorological risk factor and a multi-round iterative solicitation mechanism, the values ​​of each meteorological risk factor in the meteorological risk factor index system are divided into intervals to obtain the meteorological intervals corresponding to each meteorological risk factor, specifically including: Based on the risk threshold corresponding to each meteorological risk factor, the values ​​of each meteorological risk factor in the meteorological risk factor index system are divided into intervals to obtain the basic intervals corresponding to each meteorological risk factor. The basic intervals are adjusted based on a multi-round iterative solicitation mechanism to obtain the meteorological intervals corresponding to each meteorological risk factor in the meteorological risk factor index system.

3. The intelligent method for dividing meteorological risk level intervals according to claim 2, characterized in that, The basic intervals are adjusted based on a multi-round iterative solicitation mechanism to obtain the meteorological intervals corresponding to each meteorological risk factor in the meteorological risk factor index system, specifically including: An expert team was formed, and multiple rounds of anonymous consultations were conducted among the experts in the team to obtain their suggested values ​​based on the base intervals, the meteorological standard documents, the accident cases, and reference values. The suggested values ​​included the number of base intervals and the interval boundary thresholds. The reference values ​​were the suggested values ​​from the other experts in the expert team. Based on the mean and dispersion of the suggested values, the meteorological intervals corresponding to each meteorological risk factor in the meteorological risk factor index system are obtained.

4. The intelligent method for dividing meteorological risk level intervals according to claim 1, characterized in that, Clustering algorithms were used to cluster the meteorological risk factors into meteorological intervals, resulting in meteorological cluster intervals, specifically including: The K-means clustering method is used to cluster the meteorological intervals of the meteorological risk factor to obtain meteorological cluster intervals; the number of clusters in the K-means clustering method is set according to the number of meteorological risk levels; the number of clusters is 4; the meteorological cluster intervals are used to determine the initial division boundary of 5 meteorological risk intervals.

5. The intelligent method for dividing meteorological risk level intervals according to claim 1, characterized in that, For any meteorological risk factor, based on the average accident level, a genetic algorithm is used to optimize the meteorological clustering interval of that risk factor, resulting in an optimized meteorological clustering interval, specifically including: The population is initialized by treating each segmentation point of the meteorological cluster interval as an individual; the population size is determined based on the total number of accident cases in the maritime accident report; the population size is 50. Based on the initial population, multiple genetic operations and elite retention are performed with the goal of maximizing the fitness function value, and the optimal population is obtained after reaching the set maximum number of iterations; wherein, the maximum number of iterations is set according to the convergence stability requirements of the meteorological clustering interval; the maximum number of iterations is 100 generations; Optimal intervals are generated based on the split points corresponding to each individual in the optimal population, and effective intervals are selected as meteorological clustering optimization intervals based on the optimal intervals; the effective intervals include: intervals where the proportion of accident cases in the optimal interval is greater than a set proportion, the set proportion being set according to the sample statistical significance requirements; the set proportion is 1%; The fitness function value is calculated based on the absolute value of the Pearson correlation coefficient between the median of the meteorological cluster interval and the corresponding average accident level at each iteration, the sample balance performance score of the meteorological cluster interval, and the set compliance score; the sample balance performance score of the meteorological cluster interval is the normalized value of the inverse of the variance of the proportion of accident cases in the meteorological cluster interval.

6. The intelligent method for dividing meteorological risk level intervals according to claim 5, characterized in that, The formula for calculating the fitness function value is: F=0.9×r_score+0.05×α+0.05×β; Where F represents the fitness function value; r_score represents the absolute value of the Pearson correlation coefficient between the median of the meteorological cluster interval and the corresponding average accident level; α represents the sample balance performance score of the meteorological cluster interval; and β represents the setting compliance score. Among them, r_score is required to pass the significance test, and the significance level of r_score needs to be less than 0.01; otherwise, r_score is counted as 0. α is obtained by taking the reciprocal of the variance of the proportion of accident cases and then performing Min-Max normalization.

7. The intelligent method for dividing meteorological risk level intervals according to claim 1, characterized in that, Based on the meteorological clustering optimization interval, the meteorological risk level intervals for each meteorological risk factor in the meteorological risk factor index system are generated, specifically including: For any meteorological risk factor, each meteorological clustering optimization interval is taken as a meteorological risk level interval; the meteorological risk level interval includes five levels: slight risk, low risk, moderate risk, high risk and high risk, and the meteorological parameter threshold range corresponding to each level is determined by the boundary value of the meteorological clustering optimization interval.

8. A smart device for dividing meteorological risk level intervals, characterized in that, The intelligent meteorological risk level interval division device includes: The document acquisition module is used to acquire meteorological standard documents and maritime accident reports; The risk factor extraction module is used to extract meteorological risk factors according to the meteorological standard documents to obtain standard factors, and to mine meteorological risk factors that induce maritime accidents based on the principles of ship dynamics and marine meteorology to obtain theoretical supplementary factors. The risk factor indicator system construction module is used to screen accident cases related to the normative factors and the theoretical supplementary factors from the maritime accident reports, and to verify and analyze the normative factors and the theoretical supplementary factors based on the meteorological data corresponding to the accident cases, combined with manual review and expert evaluation, to determine the meteorological risk factor indicator system. The accident level quantification module is used to classify and quantify the accident cases, and determine the accident level and corresponding quantification score of each accident case. The meteorological interval division module is used to divide the value of each meteorological risk factor in the meteorological risk factor index system into intervals based on the risk threshold corresponding to each meteorological risk factor and the multi-round iterative solicitation mechanism, so as to obtain the meteorological intervals corresponding to each meteorological risk factor. The meteorological interval clustering module is used to cluster the meteorological intervals of any meteorological risk factor using a clustering algorithm to obtain meteorological cluster intervals. The interval risk value calculation module is used to calculate the average accident level of any meteorological cluster interval corresponding to any meteorological risk factor, based on the level quantification score of accident cases within the meteorological cluster interval. The interval optimization module is used to optimize the meteorological clustering interval of any meteorological risk factor based on the average accident level using a genetic algorithm, thereby obtaining the meteorological clustering optimization interval. The meteorological risk level interval division module is used to generate the meteorological risk level interval of each meteorological risk factor in the meteorological risk factor index system based on the meteorological clustering optimization interval; the meteorological risk level interval is used for maritime navigation early warning and risk assessment.

9. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that the processor executes the computer program to implement the intelligent division method for meteorological risk level intervals according to any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the intelligent division method for meteorological risk level intervals as described in any one of claims 1-7.

Images