A method, device, medium and product for defining a single-factor threshold of navigation weather risk

Abstract

The application discloses a navigation meteorological risk single-factor threshold definition method, device, medium and product, relates to the field of water navigation safety, and comprises the following steps: determining the number of accidents in the accident sample set and the number of occurrences or the time length of the corresponding meteorological condition in the background climate set of each meteorological factor in each interval based on a double-source adversarial data set; determining a risk enrichment factor and an accident severity weight of each meteorological factor in each interval; determining a relative risk rate curve of each meteorological factor according to the risk enrichment factor and the corresponding accident severity weight of each meteorological factor in each interval; determining the risk classification threshold corresponding to each meteorological factor based on the Jenks natural breakpoint method and the relative risk rate curve of each meteorological factor; and determining an independent risk classification table of each meteorological factor according to the risk classification threshold and the corresponding relative risk rate curve of each meteorological factor. The application can improve the accuracy of maritime accident risk assessment.

Description

Technical Field

[0001] This application relates to the field of maritime navigation safety, and in particular to a method, device, medium and product for defining single-factor thresholds for navigation meteorological risks. Background Technology

[0002] As ship tonnage continues to increase and shipping routes become increasingly dense, the impact of maritime weather conditions on navigation safety is becoming more and more apparent. Weather factors such as wind, waves, and visibility not only affect ship maneuverability but also largely determine the likelihood and consequences of maritime accidents.

[0003] Currently, research and applications on maritime meteorology and navigation safety, both domestically and internationally, are largely based on international conventions, national regulations, and practical experience, managing meteorological indicators in a tiered manner. The International Maritime Organization (IMO) has established principled requirements for navigation behavior under adverse weather conditions through documents such as COLREGs and SOLAS, while domestic guidelines and industry standards also provide reference thresholds for key meteorological elements such as wind and waves. Existing methods mainly rely on the interpretation of regulations, accident case analysis, and expert experience. Some studies have begun to introduce statistical analysis to explore the relationship between meteorological factors and maritime accidents, but overall, the classification remains primarily empirical.

[0004] The above-mentioned technologies are mainly based on normative clauses and limited experience summaries, or rely on conventional statistical models to build correlations, and are mainly applicable to risk identification under general meteorological conditions. There are still obvious deficiencies in the construction of indicator systems, the characterization of risk boundaries and the way data is used, making it difficult to support refined and scenario-based navigation risk assessment and control. Specifically, this is reflected in the following aspects: (1) The existing meteorological indicator system is incomplete, the division logic is coarse, the optimization algorithm is lacking, and the data and experience are not fully utilized, making it difficult to meet the needs of refined risk assessment and control. (2) The impact of a single meteorological factor on navigation safety varies significantly in different value ranges, but the existing research still does not clearly characterize its risk boundaries. Safety thresholds mostly rely on experience division, which easily leads to interval definition bias, thereby affecting the accuracy of risk assessment results. (3) The research model driven by pure accident data has a significant "survivor bias". Under normal weather conditions, the high frequency of navigation leads to a higher total number of accidents but a lower actual accident rate. Under extreme weather conditions, the frequency of navigation is low and the accident samples are scarce. Relying solely on accident data can easily misjudge the true risk level of different meteorological scenarios.

[0005] To address the aforementioned issues, a scientific method for defining risk zones for maritime meteorological indicators is urgently needed to improve navigation safety and reduce the likelihood of maritime accidents. Summary of the Invention

[0006] The purpose of this application is to provide a method, device, medium, and product for defining single-factor thresholds for navigation meteorological risks, which can improve the accuracy of maritime accident risk assessment.

[0007] To achieve the above objectives, this application provides the following solution: Firstly, this application provides a method for defining a single-factor threshold for navigation weather risk, the method comprising: A dual-source adversarial dataset is constructed based on an accident sample set and a background climate set; the accident sample set includes meteorological factors at the time of the maritime accident; the background climate set includes historical meteorological factors of the target sea area. The value range of each meteorological factor is divided into multiple intervals; and based on the dual-source adversarial dataset, the number of accidents occurring in the accident sample set and the number of occurrences or duration of the corresponding meteorological conditions in the background climate set for each meteorological factor in each interval are determined. Based on the number of accidents occurring in the accident sample set of each meteorological factor in each interval and the frequency or duration of occurrence of the corresponding meteorological conditions in the background climate set, the risk enrichment factor of each meteorological factor in each interval is determined; the risk enrichment factor is used to quantify the degree of enrichment of the risk of accident occurrence relative to the random level. Accident loss indicators are constructed based on the accident sample set, and the conditional expected loss of each meteorological factor in each interval is determined. The conditional expected loss is normalized and monotonically smoothed and compressed to obtain the accident severity weight. The relative risk rate curve of each meteorological factor is determined based on the risk enrichment factor of each meteorological factor in each interval and the corresponding accident severity weight. Based on Jenks' natural breakpoint method and the relative risk rate curve of each meteorological factor, the risk classification threshold corresponding to each meteorological factor is determined. An independent risk classification table for each meteorological factor is determined based on the risk classification threshold and the corresponding relative risk rate curve; the independent risk classification table is used to quantify the accident risk level under different meteorological conditions.

[0008] Secondly, this application provides a single-factor threshold determination device for navigation weather risk, the single-factor threshold determination device for navigation weather risk comprising: A dataset construction module is used to construct a dual-source adversarial dataset based on an accident sample set and a background climate set; the accident sample set includes meteorological factors at the time of the maritime accident; the background climate set includes historical meteorological factors of the target sea area. The feature extraction module is used to divide the value range of each meteorological factor into multiple intervals; and based on the dual-source adversarial dataset, to determine the number of accidents in the accident sample set and the number of occurrences or duration of the corresponding meteorological conditions in the background climate set for each meteorological factor in each interval. The risk enrichment factor determination module is used to determine the risk enrichment factor of each meteorological factor in each interval based on the number of accidents in the accident sample set and the number or duration of the corresponding meteorological conditions in the background climate set for each meteorological factor; the risk enrichment factor is used to quantify the degree of enrichment of the risk of accident occurrence relative to the random level. The relative risk rate curve determination module is used to construct accident loss indicators based on the accident sample set and determine the conditional expected loss of each meteorological factor in each interval; and to normalize and monotonically smooth the conditional expected loss to obtain the accident severity weight; and to determine the relative risk rate curve of each meteorological factor based on the risk enrichment factor and the corresponding accident severity weight of each meteorological factor in each interval. The risk classification threshold determination module is used to determine the risk classification threshold corresponding to each meteorological factor based on Jenks' natural breakpoint method and the relative risk rate curve of each meteorological factor. The independent risk grading table determination module is used to determine the independent risk grading table for each meteorological factor based on the risk grading threshold of each meteorological factor and the corresponding relative risk rate curve; the independent risk grading table is used to quantify the accident risk level under different meteorological conditions.

[0009] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the aforementioned single-factor threshold determination method for navigation weather risk.

[0010] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the aforementioned method for defining single-factor thresholds for navigation meteorological risks.

[0011] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the aforementioned single-factor threshold determination method for navigation weather risk.

[0012] According to the specific embodiments provided in this application, this application has the following technical effects: This application provides a method, device, medium, and product for defining single-factor thresholds for navigation meteorological risks. Through statistical analysis based on historical accident and background climate data, risk classification and threshold definition have clear statistical basis, avoiding the arbitrariness of traditional experience-based classifications and improving data-driven approach and objectivity. Through level mapping, discrete risk levels are transformed into comparable numerical indicators, providing an operable quantitative tool for navigation risk early warning, dynamic route planning, and decision support. This application scientifically quantifies the impact of key meteorological factors such as wind speed, visibility, and wave height on accident risks by integrating historical maritime accident data and long-term meteorological data, achieving objective classification and threshold definition of single-factor risk levels. It is applicable to different sea areas, different meteorological factors, and long-term historical data, and can be continuously updated and optimized, supporting the construction of refined navigation safety management and early warning systems. In practical applications, this method can receive real-time meteorological data as input and output the corresponding single-factor risk level or risk value, thereby serving shipping companies, ship drivers and maritime regulatory departments. It is widely used in scenarios such as navigation risk warning, dynamic route planning and auxiliary decision-making, providing quantitative basis for route adjustment, speed control and departure decisions, and supporting automated decision-making in accident risk analysis and intelligent shipping systems, thereby improving the refinement and data-driven level of navigation safety management. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, 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.

[0014] Figure 1 This is a schematic flowchart of a single-factor threshold definition method for navigation meteorological risk in one embodiment of this application. Detailed Implementation

[0015] 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.

[0016] 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.

[0017] In one exemplary embodiment, such as Figure 1As shown, a single-factor threshold definition method for navigation meteorological risk is provided, which includes the following steps S101 to S106. Wherein: S101, a dual-source adversarial dataset is constructed based on the accident sample set and the background climate set; Specifically, an accident sample set is constructed based on a multi-source maritime accident database, and a dual-source adversarial dataset is constructed by combining background climate data provided by the National Marine Data Center; the accident sample set includes meteorological factors at the time of the maritime accident; the background climate set includes historical meteorological factors of the target sea area. S101 specifically includes the construction of an accident sample set and the construction of a background climate set; The process of constructing the accident sample set is as follows: S111, obtain no fewer than 500 historical maritime accident investigation reports; S112, Preprocessing historical maritime accident investigation reports to obtain key information; the key information includes: the time, location, and type of the accident. Specifically, text extraction is performed on historical maritime accident investigation reports. Addressing the differences between text and image layers in different PDF types, pdfplumber (a PDF data extraction library) is used for text parsing, and PaddleOCR (a PaddlePaddle OCR tool library) is combined to perform image content recognition, thus uniformly converting different types of PDF data into searchable TXT format. Accident investigation reports are uniformly converted into standardized, processable text data. Based on accident information fields such as "accident time, accident location, accident type, and casualties," regular expressions are used to identify and extract time expressions and numerical information. Matching and retrieving accident types and related descriptions based on a predefined accident type thesaurus and domain keywords enables structured extraction of key information such as accident time, location, type, and casualties. S113. Based on the time of the accident, key moments are obtained. A meteorological description corpus covering meteorological and sea state elements such as waves, wind, and precipitation is constructed. The BM25 information retrieval model (a probability-based retrieval model) is used to perform relevance matching on the standardized accident report text to achieve automatic retrieval and extraction of potential meteorological information segments and obtain meteorological factor information corresponding to the key moments. In this process, the local time identified in the text is converted into a unified standard time. Finally, the key moment is written into the structured accident database for subsequent meteorological matching and risk analysis. Meteorological factors corresponding to critical moments are extracted. These meteorological factors include at least key maritime accident influencing factors such as wind speed, visibility, and wave height, and are used to characterize the marine environmental conditions at the time of the accident. The above meteorological factors are used as molecular data in the Risk Enrichment Factor (REF) of accident occurrence conditions to characterize the statistical characteristics of the actual occurrence of accidents under specific meteorological conditions.

[0018] S114, Construct an accident sample set based on meteorological factors corresponding to key moments; The process of constructing the background climate set is as follows: S121: Obtain historical meteorological data for the target sea area over the past year from the National Marine Data Center; the historical meteorological data is sampled at hourly intervals. S122, performs time alignment and linear interpolation on historical meteorological data to extract historical meteorological factors; Specifically, historical meteorological data is time-aligned. This time alignment eliminates time base differences by unifying the time zone standards of different data sources. Based on the relative positional relationship between the time of the accident and the meteorological observation time series, a linear interpolation method is used to continuously estimate adjacent observations, thereby achieving accurate matching between meteorological observation data and accident time. Meteorological factors such as wind speed, visibility, and wave height consistent with the accident sample set are extracted. S123, construct a background climate set based on historical meteorological factors.

[0019] Based on the background climate set, the frequency of occurrence of each meteorological factor within the study time range is statistically analyzed, and the corresponding natural probability distribution is constructed. The probability distribution is used to characterize meteorological factors in the normal navigation environment without considering the occurrence of accidents, and provides a probability denominator benchmark for accident risk analysis (risk enrichment factor).

[0020] S102, divide the value range of each meteorological factor into multiple intervals; and based on the dual-source adversarial dataset, determine the number of accidents in the accident sample set and the number of occurrences or duration of the corresponding meteorological conditions in the background climate set for each meteorological factor in each interval. S103, based on the number of accidents in the accident sample set of each meteorological factor in each interval and the number of occurrences or duration of meteorological conditions corresponding to the background climate set, determine the risk enrichment factor of each meteorological factor in each interval; the risk enrichment factor is used to quantify the degree of enrichment of the accident occurrence risk relative to the random level. S103 specifically includes: Using formula Determine the risk enrichment factor of meteorological factor x within the current interval. ; in, This represents the total number of accidents in the accident sample set. This indicates the total duration or number of days of the background climate concentration. This indicates the number of accidents that occurred in the accident sample set. This indicates the frequency or duration of occurrence of meteorological factors corresponding to the background climate concentration.

[0021] S104. Construct accident loss indicators based on the accident sample set and determine the conditional expected loss of each meteorological factor in each interval; normalize the conditional expected loss and perform monotonically smooth compression mapping to obtain the accident severity weight; determine the relative risk rate curve of each meteorological factor based on the risk enrichment factor of each meteorological factor in each interval and the corresponding accident severity weight. As a specific implementation, firstly, an accident loss index is constructed based on an accident sample set to characterize the differences in accident consequences under different wave height conditions. This accident loss index comprehensively considers the accident damage level and assigns preset weights. Based on the accident sample set, the average accident damage level is statistically calculated for each wave height interval, and the conditional expected loss for each interval is calculated. Subsequently, using the lowest wave height interval as a benchmark, the conditional expected loss for each interval is normalized to form a preliminary accident severity weight. Then, a monotonically smoothed compression mapping is applied to address weight fluctuations caused by the sparse sample size of high wave heights, resulting in a final stable weight value. Finally, this accident severity weight is incorporated into a risk enrichment factor to calculate the weighted risk rate. It is used to quantify the accident risk level of each wave height range and to provide a basis for threshold identification and independent risk classification of meteorological factors.

[0022] S104 specifically includes: Using formula Determined Expected Loss ; The conditional expected loss is normalized using the smallest interval; The normalized conditional expected loss is monotonically smoothed to obtain the accident severity weights for each interval. Using formula Determine the relative risk rate curve for each meteorological factor. ; in, As an indicator of accident losses, Weights for the severity of the accident.

[0023] By continuously plotting the WRF values ​​for each interval, the relative risk rate curve of the meteorological factor is obtained. The slope of the curve reflects the rate of deterioration of accident risk as meteorological conditions change.

[0024] S105. Based on Jenks' natural breakpoint method and the relative risk rate curve of each meteorological factor, the risk classification threshold corresponding to each meteorological factor is determined. S105 specifically includes: S51. Based on the relative risk rate curve of each meteorological factor, the Jenks natural breakpoint method is used to analyze the distribution pattern and changing trend. S52, by identifying the turning point where the risk enrichment factor transforms from steady growth to sudden growth as the meteorological factor value changes, the risk classification threshold of the corresponding meteorological factor is determined.

[0025] The above method was used to identify background correction threshold abrupt changes, including at least the following three types of key thresholds: (1) Threshold T1 (low risk → medium risk): This value corresponds to the stage where the accident risk transitions from being significantly lower than the background level to being close to the background level. It is the warning threshold before risk enrichment occurs. (2) Threshold T2 (medium risk → high risk): At this threshold, the slope of the WRF curve reaches its maximum; when the WRF exceeds T2, the accident risk enters the stage of accelerated deterioration. (3) Threshold T3 (high risk → extremely high risk): At this threshold, the probability of an accident increases significantly, that is, the indicator undergoes a statistically significant abrupt increase before and after the threshold. Its determination is made by combining WRF with the accident probability indicator, and the two together constitute the basis for threshold determination. When WRF exceeds T3, the probability of an accident increases significantly.

[0026] The threshold determination process relies entirely on the ratio between accident sample data and background climate data to achieve an objective classification of meteorological risk levels.

[0027] S106, determine an independent risk classification table for each meteorological factor based on the risk classification threshold of each meteorological factor and the corresponding relative risk rate curve; the independent risk classification table is used to quantify the accident risk level under different meteorological conditions.

[0028] For each meteorological factor, the relative risk rate curve construction and threshold identification process are performed independently to form an independent risk classification table corresponding to each meteorological factor, which describes the accident risk level of the factor in different value ranges. ; The risk classification table has the following characteristics: (1) Each meteorological factor corresponds to several consecutive intervals, and each interval is mapped to a risk level; (2) The risk levels include at least Level I (normal zone), Level II (attention zone), Level III (warning zone) and Level IV (no-fly zone), which respectively represent the changing trend of accident risk from low to high; As shown in Table 1, the grading table is automatically divided by the calculated WRF curve. The boundary of each interval is determined by the abrupt change point or slope change of the WRF curve, without the need for manual experience setting.

[0029] Table 1

[0030] The following section, using the risk level range classification based on wave height as an example, provides a further detailed explanation of the technical solution of this application: S1, Constructing a dual-source adversarial dataset We collected 626 historical maritime accident investigation reports as accident sample data sources, extracted meteorological accident factors such as wave height corresponding to the key moments of the accidents, and constructed an accident sample set.

[0031] Meanwhile, background meteorological data of the target sea area over the past year were collected, with a sampling interval of 1 hour, resulting in a total of 8,760 meteorological samples.

[0032] Based on the background climate data, the frequency of occurrence of each meteorological factor within the study time range was statistically analyzed, and the corresponding background climate natural probability distribution was constructed, as shown in Table 2.

[0033] Table 2

[0034] S2, Accident Severity Weighting Modeling (1) To characterize the differences in accident consequences under different wave heights, an accident loss index is introduced: ; Among them, the weighting coefficient is set. =1.0, =0.3, =1.2, Indicates the number of deaths. Indicates the number of injured. The extent of ship damage is shown in Table 3.

[0035] Table 3

[0036] (2) Constructing accident sample loss statistics The "average loss composition" was obtained by statistically analyzing 626 accidents according to wave height range, as shown in Table 4. Table 4

[0037] (3) Calculation of conditional expected loss ; The calculation results of the conditional expected loss are shown in Table 5.

[0038] Table 5

[0039] (4) Weight normalization calculation Based on the interval of 0–2m: ,Right now ;in, Preliminary accident severity weighting; The results of the normalized conditional expected loss calculation are shown in Table 6.

[0040] Table 6

[0041] (5) Monotonic smoothing correction Due to the high sparsity of high-volume samples, the weights are compressed and mapped: ; The accident severity weights for each interval are shown in Table 7.

[0042] Table 7

[0043] S3, Construction of the relative risk rate curve For wave height meteorological factors, the value range of 0m-12m is divided into 6 discrete intervals with preset equal interval widths: 0-2m, 2-4m, 4-6m, 6-8m, 8-10m, and 10-12m. Within each of these intervals, the following statistics are calculated: (1) Number of accidents in the accident sample set ; (2) Frequency of occurrence of meteorological conditions corresponding to the background climate concentration ; The distribution of accidents is shown in Table 8.

[0044] Table 8

[0045] First, the risk enrichment factor is defined as: ; The REF calculated for each wave height range is shown in Table 9.

[0046] Table 9

[0047] Based on this, an accident severity weight is introduced to construct a relative risk rate curve: ; Table 10

[0048] S4, Background Correction Threshold Change Point Identification The distribution pattern and trend of the constructed WRF curve were analyzed using Jenks' natural break method. By identifying the turning point of WRF from steady growth to sudden growth as the meteorological factor value changes, the risk classification threshold of the meteorological factor was determined.

[0049] (1) Threshold T1 (WRF=0.695): This value corresponds to the stage where the accident risk transitions from being significantly lower than the background level to being close to the background level. It is the warning threshold before the risk enrichment occurs. When WRF exceeds T1, the accident risk begins to show a clear enrichment trend.

[0050] (2) Threshold T2 (WRF=3.057): At this critical value, the slope of the WRF curve reaches its maximum value; when WRF exceeds T2, the accident risk enters the stage of accelerated deterioration.

[0051] (3) Threshold T3 (WRF=5.106): At this threshold, the probability of an accident increases significantly, that is, the indicator undergoes a statistically significant abrupt increase before and after the threshold. Its determination is made by combining WRF with the probability of an accident, and the two together constitute the basis for threshold determination. When WRF exceeds T3, the probability of an accident increases significantly.

[0052] S5 meteorological factor independent risk classification For wave height meteorological factors, by constructing a relative risk rate curve and performing threshold identification, risk classification results corresponding to the meteorological factors are generated one-to-one, thus forming an independent risk classification table for wave height meteorological factors, which is used to characterize the accident risk level corresponding to the meteorological factors in different value ranges, as shown in Table 11.

[0053] Table 11

[0054] Based on the same inventive concept, this application also provides a navigation weather risk single-factor threshold definition device for implementing the navigation weather risk single-factor threshold definition method described above. 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 navigation weather risk single-factor threshold definition device provided below can be found in the limitations of the navigation weather risk single-factor threshold definition method above, and will not be repeated here.

[0055] In one exemplary embodiment, a navigation meteorological risk single-factor threshold definition device is provided, comprising: A dataset construction module is used to construct a dual-source adversarial dataset based on an accident sample set and a background climate set; the accident sample set includes meteorological factors at the time of the maritime accident; the background climate set includes historical meteorological factors of the target sea area. The feature extraction module is used to divide the value range of each meteorological factor into multiple intervals; and based on the dual-source adversarial dataset, to determine the number of accidents in the accident sample set and the number of occurrences or duration of the corresponding meteorological conditions in the background climate set for each meteorological factor in each interval. The risk enrichment factor determination module is used to determine the risk enrichment factor of each meteorological factor in each interval based on the number of accidents in the accident sample set and the number or duration of the corresponding meteorological conditions in the background climate set for each meteorological factor; the risk enrichment factor is used to quantify the degree of enrichment of the risk of accident occurrence relative to the random level. The relative risk rate curve determination module is used to construct accident loss indicators based on the accident sample set and determine the conditional expected loss of each meteorological factor in each interval; and to normalize and monotonically smooth the conditional expected loss to obtain the accident severity weight; and to determine the relative risk rate curve of each meteorological factor based on the risk enrichment factor and the corresponding accident severity weight of each meteorological factor in each interval. The risk classification threshold determination module is used to determine the risk classification threshold corresponding to each meteorological factor based on Jenks' natural breakpoint method and the relative risk rate curve of each meteorological factor. The independent risk grading table determination module is used to determine the independent risk grading table for each meteorological factor based on the risk grading threshold of each meteorological factor and the corresponding relative risk rate curve; the independent risk grading table is used to quantify the accident risk level under different meteorological conditions.

[0056] In an exemplary embodiment, a computer device is provided, which may be a server or a terminal. The 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 connected to the system bus via the I / O interfaces. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The I / O interfaces of the computer device are used for exchanging information between the processor and external devices. The communication interface of the computer device is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a single-factor threshold definition method for navigation meteorological risk.

[0057] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0058] 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.

[0059] 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.

[0060] 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, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0061] Those skilled in the art will understand that all or part of the processes in 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. When executed, the computer program can include the processes of the embodiments described above. 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).

[0062] 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, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0063] In this application, all actions to acquire signals, information, or data are carried out in compliance with the relevant data protection laws and policies of the country where the location is situated, and with the authorization granted by the owner of the relevant device.

[0064] 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.

[0065] 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 defining a single-factor threshold for navigation meteorological risk, characterized in that, The method for defining the single-factor threshold of navigation meteorological risk includes: A dual-source adversarial dataset is constructed based on an accident sample set and a background climate set; the accident sample set includes meteorological factors at the time of the maritime accident; the background climate set includes historical meteorological factors of the target sea area. The value range of each meteorological factor is divided into multiple intervals; and based on the dual-source adversarial dataset, the number of accidents occurring in the accident sample set and the number of occurrences or duration of the corresponding meteorological conditions in the background climate set for each meteorological factor in each interval are determined. Based on the number of accidents occurring in the accident sample set of each meteorological factor in each interval and the frequency or duration of occurrence of the corresponding meteorological conditions in the background climate set, the risk enrichment factor of each meteorological factor in each interval is determined; the risk enrichment factor is used to quantify the degree of enrichment of the risk of accident occurrence relative to the random level. Accident loss indicators are constructed based on the accident sample set, and the conditional expected loss of each meteorological factor in each interval is determined. The conditional expected loss is normalized and monotonically smoothed and compressed to obtain the accident severity weight. The relative risk rate curve of each meteorological factor is determined based on the risk enrichment factor of each meteorological factor in each interval and the corresponding accident severity weight. Based on Jenks' natural breakpoint method and the relative risk rate curve of each meteorological factor, the risk classification threshold corresponding to each meteorological factor is determined. An independent risk classification table for each meteorological factor is determined based on the risk classification threshold and the corresponding relative risk rate curve; the independent risk classification table is used to quantify the accident risk level under different meteorological conditions. The method of determining the risk enrichment factor for each meteorological factor in each interval based on the number of accidents in the accident sample set for each meteorological factor within each interval and the frequency or duration of occurrence of the corresponding meteorological conditions in the background climate set includes: Using formula Determine the risk enrichment factor of meteorological factor x within the current interval. ; in, This represents the total number of accidents in the accident sample set. This indicates the total duration or number of days of the background climate concentration. This indicates the number of accidents that occurred in the accident sample set. This indicates the duration or number of days that meteorological factors corresponding to the background climate concentration occur; The process involves constructing an accident loss index based on an accident sample set and determining the conditional expected loss for each meteorological factor within each interval; normalizing the conditional expected loss and applying a monotonically smooth compression mapping to obtain the accident severity weight; and determining the relative risk rate curve for each meteorological factor based on its risk enrichment factor and corresponding accident severity weight within each interval. Specifically, this includes: Using formula Determined Expected Loss ; The conditional expected loss is normalized using the smallest interval; The normalized conditional expected loss is monotonically smoothed to obtain the accident severity weights for each interval. Using formula Determine the relative risk rate curve for each meteorological factor. ; in, As an indicator of accident losses, Weights for the severity of the accident.

2. The method for defining single-factor thresholds for navigation meteorological risk according to claim 1, characterized in that, The construction of a dual-source adversarial dataset based on an accident sample set and a background climate set specifically includes: Obtain historical maritime accident investigation reports; Preprocessing of historical maritime accident investigation reports yields key information, including the time, location, and type of the accident. The key moments are determined based on the time of the accident, and the meteorological factors corresponding to the key moments are extracted. An accident sample set was constructed based on meteorological factors corresponding to key moments. Obtain historical meteorological data for the target sea area; Historical meteorological data are time-aligned and linearly interpolated to extract historical meteorological factors; A background climate set was constructed based on historical meteorological factors.

3. The method for defining single-factor thresholds for navigation meteorological risk according to claim 2, characterized in that, The preprocessing includes: page text extraction, paragraph segmentation, noise removal, and standardization.

4. The method for defining single-factor thresholds for navigation meteorological risk according to claim 1, characterized in that, The method of determining the risk classification threshold for each meteorological factor based on Jenks' natural breakpoint method and the relative risk rate curve of each meteorological factor specifically includes: Based on the relative risk rate curve of each meteorological factor, the Jenks natural breakpoint method is used to analyze the distribution pattern and changing trend. By identifying the turning point where risk enrichment factors transform from steady growth to abrupt growth as meteorological factors change, the risk classification threshold of the corresponding meteorological factors can be determined.

5. A single-factor threshold determination device for navigation meteorological risk, used to implement the single-factor threshold determination method for navigation meteorological risk according to any one of claims 1-4, characterized in that, The navigation meteorological risk single-factor threshold definition device includes: A dataset construction module is used to construct a dual-source adversarial dataset based on an accident sample set and a background climate set; the accident sample set includes meteorological factors at the time of the maritime accident; the background climate set includes historical meteorological factors of the target sea area. The feature extraction module is used to divide the value range of each meteorological factor into multiple intervals; and based on the dual-source adversarial dataset, to determine the number of accidents in the accident sample set and the number of occurrences or duration of the corresponding meteorological conditions in the background climate set for each meteorological factor in each interval. The risk enrichment factor determination module is used to determine the risk enrichment factor of each meteorological factor in each interval based on the number of accidents in the accident sample set and the number or duration of the corresponding meteorological conditions in the background climate set for each meteorological factor; the risk enrichment factor is used to quantify the degree of enrichment of the risk of accident occurrence relative to the random level. The relative risk rate curve determination module is used to construct accident loss indicators based on the accident sample set and determine the conditional expected loss of each meteorological factor in each interval; and to normalize and monotonically smooth the conditional expected loss to obtain the accident severity weight; and to determine the relative risk rate curve of each meteorological factor based on the risk enrichment factor and the corresponding accident severity weight of each meteorological factor in each interval. The risk classification threshold determination module is used to determine the risk classification threshold corresponding to each meteorological factor based on Jenks' natural breakpoint method and the relative risk rate curve of each meteorological factor. The independent risk grading table determination module is used to determine the independent risk grading table for each meteorological factor based on the risk grading threshold of each meteorological factor and the corresponding relative risk rate curve; the independent risk grading table is used to quantify the accident risk level under different meteorological conditions.

6. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the single-factor threshold determination method for navigation meteorological risk according to any one of claims 1-4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the single-factor threshold definition method for navigation meteorological risk as described in any one of claims 1-4.

8. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the single-factor threshold definition method for navigation meteorological risk as described in any one of claims 1-4.

Images