Polar region ship navigation risk management and control method and system

Abstract

The invention relates to a polar region ship navigation risk management and control method and system, and the method comprises the following steps: obtaining polar region natural environment original data, predicting missing data points through a random forest, and obtaining a high-density environment risk data sample in a space-time dimension; a Bayesian network model of a polar region ship navigation process is constructed based on accident risk factors, model parameters are calculated based on environment observation data, and navigation safety assessment in a multi-risk mode is realized by constructing a risk assessment model; and for the most important risk type after navigation safety assessment, comprehensive risk management and control of polar region ship navigation are realized in combination with ship seaworthiness, ship navigation window period, air route and risk management and control measures. Compared with the prior art, the method has the advantages that environment and climate real-time characteristics are fully considered, risk assessment and comprehensive risk management and control are realized through subjective and objective combination, and the like.

Description

Technical Field

[0001] The present invention relates to the field of ship navigation risk prevention, and in particular to a polar ship navigation risk management and control method and system. Background Art

[0002] With global climate change and the increase in polar expeditions, ship navigation activities in polar waters are becoming increasingly frequent. Due to the unique natural geographical conditions of the Arctic and the scarcity of coastal ports, the navigation conditions for ships on the Arctic route are more complicated than those on ordinary waters, and the safety of the Arctic route has become a key issue of global concern. The polar water environment is complex and changeable. Compared with other sea areas, the main problems faced by ship navigation are extreme environments characterized by sea ice, low temperatures, and high latitudes. Existing research mainly shows the following deficiencies:

[0003] (1) The existing ship navigation risk prevention and control system is mainly aimed at general waters, and does not take into account the special environment and real-time climate characteristics of polar waters, and cannot meet the high-density analysis needs of navigation in polar waters.

[0004] (2) In the current ship navigation risk assessment, the subjective data obtained through expert questionnaires are relatively strong. With the enhancement of information technology and data collection capabilities, the ship navigation risk prevention and control system needs to be strengthened in terms of objectivity and real-time to meet the spatiotemporal information requirements of the ship navigation process.

[0005] (3) The risk prevention and control measures for the safe navigation of polar ships are not strong, and there is a lack of an overall risk management and control system for ship navigation, especially an overall plan such as risk response models based on ship characteristics, route characteristics, and temporal and spatial characteristics.

[0006] Therefore, it is necessary to develop a risk prevention and control system for ship navigation in polar waters to implement an operational and systematic ship polar navigation plan and improve the safety, economy and reliability of ship navigation in polar waters. Summary of the invention

[0007] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and provide a polar ship navigation risk management method and system to solve or partially solve the problem that the ship navigation risk prevention and control system in polar waters lacks consideration of the real-time characteristics of special environment and climate, and the evaluation means are highly subjective, lack objectivity and real-time performance, and lack an overall risk management system.

[0008] The purpose of the present invention can be achieved by the following technical solutions:

[0009] One aspect of the present invention provides a method for managing and controlling polar ship navigation risks, comprising the following steps:

[0010] Obtain original data on the polar natural environment, use random forests to predict missing data points, and obtain high-density environmental risk data samples in the time and space dimensions;

[0011] A Bayesian network model of polar ship navigation is constructed based on accident risk factors, model parameters are calculated based on environmental observation data, and navigation safety assessment under multiple risk modes is achieved by constructing a risk assessment model;

[0012] Based on the main risk types after the navigation safety assessment, comprehensive risk control of polar ship navigation is achieved by combining the ship's seaworthiness, ship navigation window period, routes and risk control measures.

[0013] As a preferred technical solution, the process of predicting missing data points through random forest to obtain high-density environmental risk data samples in the spatiotemporal dimension includes the following steps:

[0014] Converting the polar natural environment raw data into geographic coordinates to form high-density environmental risk data;

[0015] According to the types of missing data points in high-density environmental risk data, a random forest model is constructed and trained, and the trained random forest model is used to predict the values ​​of the missing data points.

[0016] As a preferred technical solution, the process of constructing a Bayesian network model of polar ship navigation process based on accident risk factors includes the following steps:

[0017] Based on the high-density environmental risk data samples, multiple risk types are analyzed by accident tree analysis to obtain accident risk factors. The accident risk factors are used as node variables to construct a Bayesian network model of the polar ship navigation process, wherein the multiple risk types include multiple of ice entrapment, collision, stranding, sinking, fire, and explosion.

[0018] As a preferred technical solution, the process of calculating model parameters based on environmental observation data includes the following steps:

[0019] Based on the visibility, wave speed, wind speed, sea ice density, sea ice thickness and temperature data of the route, the corresponding risk level is obtained according to the interval;

[0020] Taking visibility, wave speed, wind speed, sea ice density, sea ice thickness and temperature as dynamic variables, the Bayesian network model is trained to obtain the transfer probability of the dynamic variables and determine the model parameters.

[0021] As a preferred technical solution, the process of building a risk assessment model includes the following steps:

[0022] A BBN risk assessment model is established based on the accident tree, a DBN risk assessment model is constructed through the marginal effect of the hidden Markov calculation factor time series, and the navigation safety assessment under multiple risk modes is realized based on the high-density environmental risk data samples.

[0023] As a preferred technical solution, the process of achieving comprehensive risk control of polar ship navigation, combining ship seaworthiness, ship navigation window, route and risk control measures, includes the following steps:

[0024] Obtaining the seaworthiness of a ship based on its ice class: using the ice class classification standard to analyze the seaworthiness of ships of different ice classes on routes;

[0025] The seasonal window period for ship navigation is obtained according to environmental conditions: fog, waves, wind and sea ice are used to reflect the changes in system behavior over time, and the dynamic change representation of the system dynamics model is realized. The seasonal window period for ship navigation is obtained by simulating the risk of polar ship navigation;

[0026] Obtain navigation routes for ships based on environmental conditions: obtain channel environmental factor data and perform three-dimensional spatial interpolation for missing data, build an indicator system and use cloud model fusion to obtain a polar environmental risk portrait, and obtain safe and controllable routes by extracting risk equipotential lines.

[0027] As a preferred technical solution, the waterway environmental factor data include ice density, thickness, wave height, temperature and visibility.

[0028] As a preferred technical solution, the original data of the polar natural environment includes wind speed, wave height, temperature, sea ice density and sea ice thickness data.

[0029] Another aspect of the present invention provides a polar ship navigation risk management and control system, comprising:

[0030] A high-density environmental risk data sample construction module is used to obtain the original data of the polar natural environment, predict the missing data points through random forests, and obtain high-density environmental risk data samples in the spatiotemporal dimension;

[0031] The multi-risk mode navigation safety assessment module is used to build a Bayesian network model of the polar ship navigation process based on accident risk factors, calculate model parameters based on environmental observation data, and realize navigation safety assessment under multi-risk modes by building a risk assessment model;

[0032] The comprehensive risk management and control module for polar ship navigation is used to target the most important risk types after navigation safety assessment, and combines ship seaworthiness, ship navigation window period, route routes and risk management measures to achieve comprehensive risk management and control of polar ship navigation.

[0033] Another aspect of the present invention provides a computer-readable storage medium, comprising one or more programs for execution by one or more processors of an electronic device, wherein the one or more programs include instructions for executing the aforementioned polar ship navigation risk management method.

[0034] Compared with the prior art, the present invention has at least one of the following beneficial effects:

[0035] (1) Fully consider the real-time characteristics of the environment and climate: The present invention obtains the original data of the polar natural environment, predicts the missing data points through random forest, obtains high-density environmental risk data samples in the time and space dimensions, and obtains polar environmental big data to fully describe the characteristics of the polar navigation environment.

[0036] (2) Risk assessment by combining subjective and objective factors: Based on the polar waters ship navigation risk visualization platform driven by multi-source data, the present invention assesses the dynamic risk of ship navigation in polar waters by combining subjective and objective factors, and effectively prevents and controls navigation risks.

[0037] (3) Comprehensive risk control is achieved from four aspects: seaworthiness, navigation window, routes and control measures: The present invention ranks the risks faced based on the risk assessment results of the navigation process of ships in polar waters, and makes a four-in-one control decision on the most important risks, thereby improving the safety, economy and reliability of navigation in polar waters. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 This is a flow chart of the polar waters navigation risk prevention and control system in the embodiment;

[0039] Figure 2 The spatial distribution of sea ice density and sea ice thickness in June, August and October 2023 in the embodiment;

[0040] Figure 3 This is a schematic diagram of the visualization of the temperature from January to December 2023 in the embodiment;

[0041] Figure 4 This is a schematic diagram of wave height visualization from January to December 2023 in the embodiment;

[0042] Figure 5 This is a visualization diagram of wind speed from January to December 2023 in the embodiment;

[0043] Figure 6 This is a schematic diagram of the multi-risk Bayesian network for Arctic navigation in the embodiment;

[0044] Figure 7 This is the Northeast Route RIO distribution and node analysis diagram (PC7, IA Super) in the embodiment;

[0045] Figure 8This is the RIO distribution and node analysis diagram of the Northwest Route in the embodiment (PC7, IA Super);

[0046] Fig. 9 This is the Northeast Route RIO distribution and node analysis diagram (IA, IB) in the embodiment;

[0047] Fig.10 The Northwest Route RIO distribution and node analysis diagram (IA, IB) in the embodiment;

[0048] Fig.11 This is the Northeast Route RIO distribution and node analysis diagram (IC, No Strength Ice) in the embodiment;

[0049] Fig.12 This is the Northwest Route RIO distribution and node analysis diagram (IC, No Strength Ice) in the embodiment;

[0050] Fig.13 This is a risk diagram of the navigation process of ships on the 6th route of the Arctic Northwest Passage in the embodiment;

[0051] Fig.14 This is a schematic diagram of the route for optimizing navigation safety in the Northeast Passage in the embodiment;

[0052] Fig.15 This is a schematic diagram of the multi-risk Bayesian network for Arctic navigation in the embodiment;

[0053] Fig.16 Schematic diagram of DBN model for risk assessment of Arctic navigation process in the embodiment;

[0054] Fig.17 This is a schematic diagram of the navigation risks of the Northeast Passage for ice-free ships from January to December in the embodiment;

[0055] Fig.18 This is a schematic diagram of the BBN model for risk decision-making during Arctic navigation in the embodiment;

[0056] Fig.19 This is a schematic diagram of Arctic navigation decision-making in July in the embodiment;

[0057] Fig. 20 This is a visualization diagram of the navigation risk of ice-free ships from January to December in the embodiment. DETAILED DESCRIPTION

[0058] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0059] Example 1

[0060] In response to the problems existing in the aforementioned prior art, this embodiment provides a polar ship navigation risk management method, which includes a polar waters navigation risk prevention and control system. Based on a polar waters ship navigation risk visualization platform driven by multi-source data, it comprehensively evaluates the dynamic risks of ship navigation in polar waters, establishes targeted ship navigation risk control plans, effectively prevents and controls navigation risks, and improves the safety, economy and reliability of polar waters navigation.

[0061] See also Figure 1 , the risk management and control method for polar ship navigation includes the following steps:

[0062] Step S1, visualization of spatiotemporal high-density data of environmental risks for ships in polar waters.

[0063] Step S101, obtaining polar natural environment raw data from multiple data websites and performing information integration analysis.

[0064] The main sources of data are:

[0065] (1) The fifth-generation global climate reanalysis database ERA5 released by the European Centre for Medium-Range Weather Forecasts (ECMWF) integrates observational data and model data from all over the world to achieve global coverage of a unified data set. The data set contains data information such as wind speed, wave height, temperature, and sea ice density. The spatial accuracy of atmospheric data is 0.25°, and the spatial accuracy of ocean wave data is 0.5°.

[0066] (2) The National Snow and Ice Data Center (NSIDC) of the United States has obtained Arctic sea ice density data based on the scanning multi-channel / multi-band microwave radiometer of the Nimbus-7 satellite and the DMSP satellite. This dataset has been acquired since 1978, with a time accuracy ranging from daily, monthly to annual. The projection method uses polar stereographic projection, and the data spatial resolution is 25 km × 25 km.

[0067] (3) The Pan-Arctic Ice Ocean Modeling and Assimilation System (PIOMAS), developed by the Institute of Applied Physics at the University of Washington, is a numerical analysis model that can simulate and predict changes in Arctic sea ice. It can achieve full-time numerical simulation and prediction of Arctic sea ice conditions based on considerations of physical processes such as Arctic sea ice formation and melting, as well as changes in the global ocean and climate system, combined with measured data from satellite observation systems. Currently available sea ice meteorological observation satellites include the United States' ICESat-2, Europe's CryoSat-2, SMOS, and Japan's AMSR2. Although these satellites can provide high-resolution measured data on Arctic sea ice thickness, they are limited by the complex environmental conditions in the Arctic. Currently, these satellites can generally only provide ice thickness data from September to May of the following year, and cannot meet the needs of sea ice thickness impact analysis from June to August. The PIOMAS system provides us with reliable full-time ice thickness analysis data in a 25km×25km format.

[0068] Step S102, using a random forest algorithm to obtain high-density environmental risk data in a certain time and space dimension.

[0069] First convert it into geographic coordinates, and then form a data sample with high-density data (10km*10km) according to the geographic coordinates.

[0070] Due to the differences in different data types, the environmental data information provided by the data website is different in format, spatial accuracy and time accuracy. The present invention aims to establish a fusion risk assessment model based on environmental data in NetCDF format and achieve visualization effects, integrate other forms of data sets, and complete the format accuracy unification of environmental data files. Currently commonly used spatial interpolation methods include the most natural neighbor method, quadratic polynomial interpolation method, inverse distance weighted interpolation method, Kriging interpolation method, radial basis function interpolation method, etc.

[0071] The data processed in this step are meteorological data in NetCDF format for the Arctic region. The main data information is the longitude and latitude of the spatial data points and the meteorological values ​​under the corresponding longitude and latitude conditions. It is suitable for the interpolation algorithm to process three-dimensional data, and the longitude and latitude points of the data set are evenly distributed on the rectangular coordinate plane. Considering the scope of application of the above methods and other conditions, the random forest method is used as the application method of spatial interpolation.

[0072] In this embodiment, the Arctic waters are represented as 60°-90°N, -180°W-180°E in the latitude and longitude range, and the spatial accuracy of the sample data is 0.25°×0.25° (i.e. 27.8km*9.5km), that is, the spatial grid layout of the data points is 121×1440. In order to solve the problem of lack of data, the random forest method can predict and supplement the missing data points with meteorological data such as sea ice thickness, sea ice density, temperature, wave height, wind speed, etc. Random forest is a powerful ensemble learning algorithm that integrates the prediction results of decision trees to improve the overall prediction performance and stability. The mathematical expression is:

[0073]

[0074] Where i represents the i-th decision tree, x represents the feature information, and T(X) represents the predicted value.

[0075] The prediction process of random forest is as follows. Determine the feature variables used for interpolation, such as longitude and latitude, time, etc. The feature variables should be able to represent the spatial and temporal distribution of meteorological data. Build a random forest model. Including training: Use known data in the training set to train the random forest model. Random forest makes predictions by generating multiple decision trees. Each decision tree is trained on a random subset of features, and finally the final prediction result is determined by a voting mechanism. Given the meteorological data of Arctic waters, a random forest regression model is used to predict each missing data point in the data set. The mathematical expression is:

[0076]

[0077] Where X miss represents the data points with missing feature values, and RF(X) represents the prediction model.

[0078] Use the trained model to predict the missing data in the test set. Apply the model to the missing data points and get the interpolated results. The model will predict the missing values ​​based on the characteristics of the known data. Figure 2-Figure 5 Visualization of predicted sea ice density and thickness, temperature, wave height, and wind speed. Figure 2 In the figure, the left side is a schematic diagram of the sea ice thickness distribution, and the right side is the spatial distribution of sea ice density. From top to bottom, they are the distribution maps of sea ice density and thickness in June, August and October 2023. Figure 3 In the middle, from left to right and from top to bottom, this is a visualization diagram of the temperature for each month from January to December 2023. Figure 4 In the middle, from left to right and from top to bottom, this is a visualization diagram of the wave heights for each month from January to December 2023. Figure 5 In the middle, from left to right and from top to bottom, this is a visualization diagram of wind speed for each month from January to December 2023.

[0079] Step S103: Environmental risk map: Combining the above algorithms and steps, a visualized environmental risk image is obtained through computer processing.

[0080] Step S2: Dynamic risk assessment of ship navigation in polar waters.

[0081] This method is based on high-density polar environmental risk data and uses objective data visualization results as one of the analysis bases for the risk characteristics of polar ship navigation. It combines objective data such as accident reports, environmental information and ships with subjective data from crew experience to establish a polar waters ship navigation risk assessment model and a risk decision-making model, and analyzes the risk of ship navigation in Arctic waters. Based on the spatiotemporal characteristics of risk perception, risk control plans and risk decision-making recommendations are obtained.

[0082] Step S201: Model construction.

[0083] Combined with the accident tree analysis method, ice entrapment, collision, grounding, sinking and fire / explosion accidents were analyzed. After analysis, the final risk factors were selected and a preliminary Bayesian network model of the risk of navigation in the Arctic Northeast Passage was established. Through the analysis of Arctic accident report data and related literature, risk factors were established and used as node variables to build a Bayesian network model, such as Figure 6 shown.

[0084] Step S202: subjective and objective data processing.

[0085] This step introduces environmental observation data to obtain model parameters of environmental influencing factors. The visibility data on the route is collected through the NOAA website and then statistically analyzed to determine the parameters; the wave / wind speed, sea ice density, sea ice thickness and temperature data on the route are extracted from the EMWCF shared meteorological information website, and the state description data of the environmental data is determined by using interval division according to the relevant scoring rules. The scoring rules are shown in Table 1.

[0086] Table 1 Environmental factor scoring rules

[0087]

[0088] Step S203: Determination of the transfer matrix.

[0089] The difficulty of establishing the entire model is to obtain the transfer probability of dynamic nodes. By analyzing the process risks, sea ice density, sea ice thickness, visibility, wind, waves, and temperature variables are defined as dynamic variables. First, statistical analysis is performed on the data in the sample database to establish a sample set, and the model is input for parameter training to obtain the transfer probability of dynamic nodes.

[0090] Step S204: Risk decision.

[0091] Through accident tree analysis, the risk causes of ice entrapment, collision, grounding, sinking and fire are obtained, including environmental factors, human factors and ship factors. Then, the BBN risk assessment model is established based on the accident tree analysis. Then, the HMM algorithm is used to obtain the marginal effect of the factor time series, and the DBN risk assessment model is established. Finally, data samples are obtained to conduct navigation safety assessment under multiple risk modes.

[0092] Step S3, making a four-in-one management and control decision on the risks of ship navigation in polar waters.

[0093] According to the risk assessment results of ships' navigation in polar waters, the risks faced are ranked and control plans are adopted for the most important risks (ice entrapment, collision or grounding). The basic scope of the control plan includes: Ships: seaworthiness requirements for ships navigating in polar waters, especially the ability to resist ice levels. Time period: the window period for ships to navigate in polar waters, especially the navigation period with good safety and economy. Route: route selection for ships to navigate in polar waters, including the northeast and northwest routes. Measures: risk management plans and measures for ships' navigation in polar waters. The following work steps are formed:

[0094] Step S301, establishing the seaworthiness of the ship according to the ship's ice class.

[0095] On the basis of data fusion theory, the risk of sea ice navigation, which has a special and prominent impact on ships in the Arctic waters, is analyzed. The Polaris risk analysis system proposed by the International Maritime Organization (IMO) is introduced. The RIO value proposed in the system is calculated according to the ice class classification standard. The seaworthiness of ships of different ice classes (seasonal navigation) on the Arctic route can be analyzed, such as Figure 7-12 As shown in the figure. The classification of ship ice classes is shown in Table 2. The results show that during the navigable period from June to November in Arctic waters, ships of all ice classes on the Northeast Route can navigate normally, but the RIO values ​​in the waters near the Drang Strait, the East Siberian Sea, the Sannikov Strait and the Vinikits Strait are relatively low, so navigation in these areas requires attention; while the RIO value in the Northwest Route is generally lower than that in the Northeast Route. PC7, IA Super, and IA ships can navigate the entire section of the Northwest Route, but ships of IA ice class are already at a high risk level in the waters near Griffith Island and Prince of Wales Strait, so independent navigation can be avoided as much as possible. IB, IC and No Strength Ice ships cannot navigate independently in the waters near Griffith Island and Prince of Wales Strait, so independent navigation arrangements for IB, IC and No Strength Ice ships in relevant waters should be avoided in the design of navigation plans.

[0096] Table 2 Ship ice class classification

[0097]

[0098] Step S302, establishing a seasonal window period for ship navigation based on environmental conditions.

[0099] Through the amount of fog, waves, wind and sea ice that reflect the change of system behavior over time, the dynamic change characteristics of the system dynamics model are realized to characterize the process risk of Arctic ship navigation. Based on the analysis of the risk mechanism of Arctic ship navigation and the determination of variable properties, a system dynamics model of the process risk of Arctic ship navigation is established. With the help of simulation of Arctic ship navigation risks, the window period for Arctic ship navigation is identified. Taking the Northwest Passage as an example, Fig.13 The results show that the navigable time window for ships on the Arctic Northwest Passage is from late August to early September throughout the year, which is less than one month on average.

[0100] Step S303, establishing a route optimized for ship navigation safety based on environmental conditions.

[0101] Acquire data based on waterway environmental factors (including ice, wind, waves, temperature, and visibility), and use a three-dimensional spatial interpolation method to obtain complete and continuous environmental data information to address the problem of missing original data; consider five types of environmental information, build an indicator system, and use the cloud model fusion algorithm to obtain an Arctic environmental risk profile, extract risk equipotential lines, and propose a safe and controllable optimized route. Taking the Northeast Passage as an example, Fig.14 It is the optimized route for navigation safety in the Northeast Passage. The four sub-graphs illustrate the dynamic optimization process of route decision-making.

[0102] Step S304, establishing a ship navigation risk management and control mode according to environmental conditions.

[0103] The following is a further explanation of this method using the risk control of ship navigation in the Northeast Passage from January to December 2023 as an example.

[0104] First, the risk factors were analyzed, including environmental factors, human factors, and ship factors. By establishing a DBN model for the risk analysis of the Arctic route and verifying the model, the five risks (ice entrapment, collision, grounding, fire, and sinking) of ice-free ships throughout the entire process of sailing in the Arctic Northeast Passage in July were evaluated. Through the analysis of navigation risk factors, some risk factors were initially obtained as variable nodes of the Bayesian network. Combined with the accident tree analysis method, ice entrapment, collision, grounding, sinking, and fire / explosion accidents were analyzed, the final risk factors were selected, and a Bayesian network model for the risk of the Arctic Northeast Passage navigation process was built. Fig.15 .

[0105] Set the environment node as a dynamic node to obtain a dynamic Bayesian network model, see Fig.16In this example, environmental observation data are introduced to obtain the model parameters of environmental influencing factors. The scoring rules are shown in Table 1.

[0106] By introducing historical data information to determine the conditional probability, prior probability and other parameters of the network model, and statistically analyzing the data in the sample database, the sample set is established, and the model is input for parameter training to obtain the transfer probability of dynamic nodes. The sample set is input into the Bayesian model for reasoning, and the evaluation results of the Arctic Northeast Passage for ice-free ships from January to December are obtained, such as Fig.17 As shown, from left to right and from top to bottom, this is a schematic diagram of navigation risks for ice-free ships in the Northeast Passage from January to December.

[0107] Based on the risk assessment results, ice-free ships sailing in the Arctic from January to December face greater risks of ice entrapment, collision and grounding. The risk of ice entrapment is greatly affected by seasonality, which is mainly affected by sea ice. The risks of grounding and collision are greatly affected by spatial location, among which the risk of grounding is "M" shaped, with two peaks in the Sannikov Strait and the Velikytsky Strait. The risk of collision and ice entrapment present a slightly opposite trend, and are greater in the Kara Sea and the Barents Sea. The distribution of fire and sinking risks is relatively stable from January to December.

[0108] Secondly, based on Bayesian decision making, this example proposes three decision-making schemes for ships sailing in the Arctic route, and establishes the BBN model for risk decision-making in the Arctic route. Fig.18 By making risk decisions on the navigation of ice-free ships in July, the decision results of five risks of ice-free ships in the Northeast Passage in July were obtained.

[0109] Based on the dynamic Bayesian model of the risk of Arctic ship navigation process, the BBN model adds decision nodes and utility nodes. The decision node is one of the inputs of the system, while the utility node is the output of the system. The decision node represents a series of measures taken after the process risk value is obtained through the risk analysis of the model, including decision 1 (independent navigation, including pilotage), decision 2 (icebreaker assisted navigation, including escort and icebreaking), and decision 3 (following navigation).

[0110] In this example, utility node parameters are determined from two aspects: safety and economic cost.

[0111] (1) Safety assessment:

[0112] Independent navigation: Assess the pilot’s professional skills and experience, as well as his or her knowledge of local sea ice conditions.

[0113] Icebreaker-assisted navigation: Consider the icebreaking capability and stability of the icebreaker, as well as its support effect on trapped ships.

[0114] Following: Analyze the maneuverability of the following ship and its ability to deal with emergencies.

[0115] (2) Cost estimation:

[0116] Independent navigation: Calculate the cost of pilotage services and possible additional costs (such as delay losses).

[0117] Icebreaker Assisted Navigation: Estimate the cost of renting or operating an icebreaker, and the likely fuel consumption.

[0118] Car-following: Analyze the additional fuel consumption and labor costs that may be generated by the car-following solution.

[0119] After obtaining multiple decision utility values, the BBN model is trained to determine the interval. The final training results show that the decision utility value range is roughly [-100, 100] and is most sensitive. The larger the decision utility value, the better the decision-making scheme. Finally, the model parameter values ​​are shown in Table 3. Taking ice-free ships as the object, the control method of their risks in the Arctic Northeast Passage in July was decided, and the model was verified by comparing the model simulation results with the expert scoring results. Fig.19 , the solid line is the model output, and the dotted line is the weighted average result after the expert scoring. The results show that as the navigation process progresses, the utility values ​​of the decisions in each time slice are constantly fluctuating. It can be seen that when the ship is in the second time slice, the third time slice, and the sixth time slice, the utility value of decision 2 (icebreaker assisted navigation) is the highest among the three decisions; in the fourth time slice, the seventh time slice, and the eighth time slice, the utility value of decision 1 (independent navigation) reaches the highest in the same period, and in the third time slice, the utility value of decision 3 (following navigation) is the highest.

[0120] Table 3 Decision utility parameter table

[0121]

[0122] Finally, based on the assessment results and combined with ship risk visualization, see Fig. 20 , make auxiliary decisions, formulate navigation plans, and obtain the minimum requirements for icebreaker levels that meet the icebreaking capabilities of each node in the northeast section from January to December.

[0123] The risks faced by ships of different ice classes when sailing vary due to the difference in ice class. According to the International Code for Ships Operating in Polar Waters formulated by the International Maritime Organization (IMO), ships sailing in Arctic waters are divided into three categories: A, B, and C, with a total of 12 ice classes, as shown in Table 2. Taking Class C ships (ice-free, IC, IB, IA, and IA Super ships) as an example, due to the high risk of ice entrapment and no or weak icebreaking capabilities, they can only sail independently in September. In the rest of the months, icebreakers are required to assist in navigation, and the icebreaker levels that can meet safety requirements in different months and different sections are also different. In terms of time, from October of the previous year to March of the following year, the Arctic route is gradually covered by thick sea ice. If you want to sail, you should choose icebreakers to assist in navigation, strengthen sea ice monitoring and prediction, plan the navigation route in advance, and allocate emergency resources; in February and March, the ice conditions in the Drang Strait, the Sannikov Strait, and the Vilkitsky Strait are the most serious, and PC3 or even PC2 ice-class icebreakers are required to meet safety requirements. From April to June, the ice conditions in the Arctic route are still serious, but the ice class requirements for icebreakers are slightly lower, and PC3 icebreakers can meet safety needs for assisted navigation. From July to August, the sea ice in the Arctic Northeast Passage has decreased significantly, and wind and wave factors have become the main factors affecting navigation safety. Ships only need short-term assisted navigation or following navigation by icebreakers in the Drang Strait and the Vilkitsky Strait to avoid ice entrapment and stranding. Ships can navigate independently in other sections. It is necessary to pay attention to weather forecasts and warning information, adjust navigation plans in time, and choose to follow navigation in bad weather to ensure navigation safety. September is the period of the year with the least sea ice in the Northeast Passage, and non-ice-class ships can navigate independently throughout the entire journey. In terms of space, PC7-class ships can meet the needs of the Kara Strait except in February, and the rest require assisted navigation by icebreakers of higher ice classes. See Table 4 for detailed plans.

[0124] Table 4 Navigation plan decision of Class C ships from January to December

[0125]

[0126] In summary, this method first constructs a multi-risk assessment model for the Arctic navigation process based on a dynamic Bayesian network. Secondly, in view of the multi-mode risk prevention and control issues of ship navigation in the Arctic route, risk prevention and control methods such as icebreaking escort, pilotage, and following required for ships sailing on the Arctic route are proposed, and a risk decision-making model for the Arctic navigation process based on BBN is established. Finally, based on the objective environmental observation data obtained, the risk assessment of ships of different ice classes is carried out through the established model, and the risk distribution characteristics and navigation plans of ships of different ice classes from January to December are obtained. Based on the polar waters ship navigation risk visualization platform driven by multi-source data, this method comprehensively evaluates the dynamic risks of ship navigation in polar waters, establishes targeted ship navigation risk control plans, effectively prevents and controls navigation risks, and improves the safety, economy and reliability of navigation in polar waters.

[0127] Example 2

[0128] Based on Example 1, this example provides a polar ship navigation risk management and control system, including:

[0129] A high-density environmental risk data sample construction module is used to obtain the original data of the polar natural environment, predict the missing data points through random forests, and obtain high-density environmental risk data samples in the spatiotemporal dimension;

[0130] The multi-risk mode navigation safety assessment module is used to build a Bayesian network model of the polar ship navigation process based on accident risk factors, calculate model parameters based on environmental observation data, and realize navigation safety assessment under multi-risk modes by building a risk assessment model;

[0131] The comprehensive risk management and control module for polar ship navigation is used to target the most important risk types after navigation safety assessment, and combines ship seaworthiness, ship navigation window period, route routes and risk management measures to achieve comprehensive risk management and control of polar ship navigation.

[0132] Example 3

[0133] This embodiment provides an electronic device, including: one or more processors and a memory, wherein the memory stores one or more programs, and the one or more programs include instructions for executing the polar ship navigation risk management method as described in Example 1.

[0134] Example 4

[0135] This embodiment provides a computer-readable storage medium, including one or more programs for execution by one or more processors of an electronic device, wherein the one or more programs include instructions for executing the polar ship navigation risk management method as described in Example 1.

[0136] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for managing and controlling polar ship navigation risks, characterized in that: The steps include: Obtain original data on the polar natural environment, use random forests to predict missing data points, and obtain high-density environmental risk data samples in the time and space dimensions; A Bayesian network model of polar ship navigation is constructed based on accident risk factors, model parameters are calculated based on environmental observation data, and navigation safety assessment under multiple risk modes is achieved by constructing a risk assessment model; Based on the main risk types after the navigation safety assessment, comprehensive risk control of polar ship navigation is achieved by combining the ship's seaworthiness, ship navigation window period, routes and risk control measures.

2. A method for managing and controlling polar ship navigation risks according to claim 1, characterized in that: The process of predicting missing data points through random forest to obtain high-density environmental risk data samples in the spatiotemporal dimension includes the following steps: Converting the polar natural environment raw data into geographic coordinates to form high-density environmental risk data; According to the types of missing data points in high-density environmental risk data, a random forest model is constructed and trained, and the trained random forest model is used to predict the values ​​of the missing data points.

3. A method for controlling polar ship navigation risks according to claim 1, characterized in that: The process of constructing a Bayesian network model of polar ship navigation based on accident risk factors includes the following steps: Based on the high-density environmental risk data samples, multiple risk types are analyzed by accident tree analysis to obtain accident risk factors. The accident risk factors are used as node variables to construct a Bayesian network model of the polar ship navigation process, wherein the multiple risk types include multiple of ice entrapment, collision, stranding, sinking, fire, and explosion.

4. A method for managing and controlling polar ship navigation risks according to claim 1, characterized in that: The process of calculating model parameters based on environmental observation data includes the following steps: Based on the visibility, wave speed, wind speed, sea ice density, sea ice thickness and temperature data of the route, the corresponding risk level is obtained according to the interval; Taking visibility, wave speed, wind speed, sea ice density, sea ice thickness and temperature as dynamic variables, the Bayesian network model is trained to obtain the transfer probability of the dynamic variables and determine the model parameters.

5. A method for managing and controlling polar ship navigation risks according to claim 1, characterized in that: The process of building a risk assessment model includes the following steps: A BBN risk assessment model is established based on the accident tree, a DBN risk assessment model is constructed through the marginal effect of the hidden Markov calculation factor time series, and the navigation safety assessment under multiple risk modes is realized based on the high-density environmental risk data samples.

6. A method for controlling polar ship navigation risks according to claim 1, characterized in that: Combining the ship's seaworthiness, the ship's navigation window, the route and risk control measures, the process of achieving comprehensive risk control for polar ship navigation includes the following steps: Obtaining the seaworthiness of a ship based on its ice class: using the ice class classification standard to analyze the seaworthiness of ships of different ice classes on routes; The seasonal window period for ship navigation is obtained according to environmental conditions: fog, waves, wind and sea ice are used to reflect the changes in system behavior over time, and the dynamic change representation of the system dynamics model is realized. The seasonal window period for ship navigation is obtained by simulating the risk of polar ship navigation; Obtain navigation routes for ships based on environmental conditions: obtain channel environmental factor data and perform three-dimensional spatial interpolation for missing data, build an indicator system and use cloud model fusion to obtain a polar environmental risk portrait, and obtain safe and controllable routes by extracting risk equipotential lines.

7. A method for managing and controlling polar ship navigation risks according to claim 6, characterized in that: The waterway environmental factor data include ice density, thickness, wave height, temperature and visibility.

8. A method for managing and controlling polar ship navigation risks according to claim 1, characterized in that: The polar natural environment raw data include wind speed, wave height, temperature, sea ice density and sea ice thickness data.

9. A polar ship navigation risk management and control system, characterized in that: include: A high-density environmental risk data sample construction module is used to obtain the original data of the polar natural environment, predict the missing data points through random forests, and obtain high-density environmental risk data samples in the spatiotemporal dimension; The multi-risk mode navigation safety assessment module is used to build a Bayesian network model of the polar ship navigation process based on accident risk factors, calculate model parameters based on environmental observation data, and realize navigation safety assessment under multi-risk modes by building a risk assessment model; The comprehensive risk management and control module for polar ship navigation is used to target the most important risk types after navigation safety assessment, and combines ship seaworthiness, ship navigation window period, route routes and risk management measures to achieve comprehensive risk management and control of polar ship navigation.

10. A computer-readable storage medium, characterized in that: It comprises one or more programs executed by one or more processors of an electronic device, wherein the one or more programs comprise instructions for executing the polar ship navigation risk management method as claimed in any one of claims 1 to 7.

Images