A method for identifying marine environmental risks based on big data

Abstract

This application discloses a method for identifying marine environmental risks based on big data, including: constructing a terrain prediction model based on initial static terrain and dynamic sediment migration data, and correcting terrain data in real time; predicting terrain changes based on the corrected data and information such as meteorology and ocean currents, and constructing a numerical model of the marine environment; calibrating the model using measured data, and then determining whether to simulate based on real-time monitoring parameters; analyzing the simulation results to divide risk areas and determine grid risk levels; this solution greatly improves the accuracy of marine environmental predictions and real-time response capabilities through real-time terrain correction, accurate prediction of changes, and dynamic calibration of models, providing scientific and efficient technical support for coastal risk prevention and control, resource management, and decision-making.

Description

Technical Field

[0001] The present invention relates to the technical field of environmental risk identification, and in particular to a method for identifying marine environmental risks based on big data. Background Art

[0002] With the rapid development of the marine economy, the risks facing the marine environment are becoming increasingly complex. Traditional risk identification methods rely on experience and expert judgment, lacking accuracy and timeliness. Marine environmental data is multi-source and heterogeneous, encompassing multi-dimensional information such as satellite remote sensing, ocean observation buoys, and ship monitoring. The data volume is massive and rapidly updated, necessitating intelligent approaches for efficient integration and analysis. Big data technology, with its ability to process massive amounts of data, can overcome the limitations of traditional methods. Using machine learning and deep learning algorithms, it can dynamically model and predict risk factors such as the spread of pollution sources, the evolution of ecological disasters, and the impact of engineering activities. Combining geographic information systems (GIS) with numerical simulation techniques can enable spatiotemporal correlation analysis of multi-source data, improving the accuracy and real-time nature of risk identification. Currently, research on marine environmental risks based on big data has been conducted both domestically and internationally, but faces challenges such as varying data quality and limited model applicability. There is an urgent need to establish a standardized risk identification framework to support decision-making on marine environmental protection.

[0003] The invention patent with patent application number 202411073751.6 discloses a method, product, medium and equipment for identifying marine environmental risks in coastal areas, which involves the field of environmental risk identification. First, the marine environment monitoring buoy is used to monitor the marine dynamic environmental parameters and meteorological parameters of the coastal area, the tide gauge or tide model is used to monitor the water level of the coastal area, and the marine environment background data acquisition equipment is used to obtain the water depth topography and shore topography of the coastal area. Based on the marine dynamic environmental parameters, meteorological parameters, water level, water depth topography and shore topography, a variety of marine environment numerical models are constructed, and the hydrodynamic environment of the coastal area is simulated. The marine environment numerical model with the smallest simulation error is used as the marine environment numerical model for business operation, which is used to simulate future marine dynamic environmental parameters and then divide the risk areas of the coastal area. The method of the present invention can improve the accuracy and real-time performance of marine environmental risk identification in coastal areas.

[0004] However, in coastal areas, the marine environment is complex and ever-changing, and topographic changes are closely linked to ocean dynamic environmental factors, making them crucial for coastal safety, resource development, and ecological protection. However, traditional marine environmental prediction methods rely heavily on static topographic data, ignoring the influence of dynamic factors such as sediment migration, resulting in deviations between predictions and actual conditions. Furthermore, real-time data such as meteorological and ocean current data are not fully integrated into prediction models, making them incapable of responding to sudden changes in the marine environment. Summary of the Invention

[0005] This application provides a marine environmental risk identification method based on big data, constructs a prediction system based on dynamic terrain data and real-time marine environmental parameters, realizes accurate prediction and timely response to marine environmental changes, and provides scientific support for risk prevention and control and decision-making in coastal areas.

[0006] This application provides a method for identifying marine environmental risks based on big data, including:

[0007] S1, based on the initial static topographic data and dynamic sediment migration data of the surveyed coastal area, a topographic prediction model is constructed, and the static topographic data is corrected in real time based on the direction and speed of sediment migration;

[0008] S2, combining the corrected terrain data with current meteorological and ocean current data, predicts terrain changes within the window time and uses this as the initial input to construct a numerical model of the ocean environment;

[0009] S3, collect measured data from coastal areas and conduct comparative analysis with model prediction results to calibrate the marine environment numerical model;

[0010] S4, using the calibrated ocean environment numerical model and according to the real-time monitored ocean dynamic environment parameters, determine whether numerical simulation is necessary;

[0011] S5, conduct a comprehensive analysis of the simulation results to obtain the risk value, divide the coastal area into risk zones and determine the risk level of the grid units, identify the spatial continuity grids; merge adjacent high-risk grids into groups, calculate the group current data and connectivity, construct the risk propagation matrix, predict the risk diffusion path, dynamically adjust the group risk level, and predict the degree of change in the risk level in the future time window.

[0012] Preferably, the S4, determining whether numerical simulation is needed, specifically includes: using marine environment monitoring equipment to monitor the marine dynamic environment parameters of coastal areas in real time; based on the calibrated marine environment numerical model, according to the real-time monitored marine dynamic environment parameters, evaluating the current marine environment conditions, and determining whether further numerical simulation is needed.

[0013] Preferably, the S2, constructing a numerical model of the marine environment, includes: obtaining real-time corrected terrain data of the coastal area, spatially and temporally matching the corrected terrain data with current meteorological and ocean current data, inputting the matched data into a terrain change prediction model, and simulating terrain changes within the window time; and comparing and verifying the simulated terrain change results with historical data or field observation data.

[0014] Preferably, the identification of spatial continuity grids specifically includes: determining the risk level of each grid unit by a threshold; for each grid unit i, according to its risk value and threshold; each grid cell i is assigned a ; Identify spatially continuous high-risk areas through region growing algorithm.

[0015] Preferably, the spatially continuous high-risk area specifically includes: the spatially continuous area S satisfies, S G, i,j∈S, and i and j are adjacent , where the set of grid cells is G and the risk level is , the high-risk grid cell that has not been visited is i, and all adjacent grid cells of i are j.

[0016] Preferably, the predicted risk diffusion path between different formations includes: analyzing the direct and indirect correlations between formations based on the ocean current connectivity matrix; constructing a risk propagation matrix to calculate the direct and indirect propagation probabilities; predicting the risk diffusion path between different formations by iteratively calculating risk diffusion, and analyzing key nodes.

[0017] Preferably, the predicted risk diffusion path between different formations also includes: integrating the risk diffusion path with the formation correlation analysis to identify key formations and associated paths; calculating the risk accumulation value of the formation, considering the time decay factor; dynamically adjusting the risk level of the formation according to the risk accumulation value and the risk level classification standard, considering the correlation impact.

[0018] Preferably, the adjacent high-risk grids are merged into a group, specifically: 2A, based on the risk level of each grid unit in the group, the core risk area in the group is identified, and the characteristics of the core risk area are extracted; 2B, by analyzing the risk level time series of the core risk area, the critical point of the core risk area is determined, and based on the spatial distribution and risk propagation characteristics of the core risk area, the dynamic critical line of the group boundary is determined; 2C, based on the critical point of the core risk area and the dynamic critical line of the group boundary, the risk state transfer rule is defined; 2D, a risk state transfer model is constructed to simulate the dynamic change process of the risk level and the group boundary; 2E, based on the risk state transfer model, a mutation warning indicator and a mutation warning threshold are designed to trigger mutation warning measures.

[0019] Preferably, the mutation warning indicators include: the risk level change rate, which is the maximum increase or average increase rate of the risk level within the calculation window time; the grouping boundary critical line moving speed, which is the average distance or maximum distance of the critical line expansion or contraction within the calculation window time; the core risk area expansion area, which is the total area or proportion of newly added high-risk grid units within the calculation window time.

[0020] Preferably, the core risk area includes: the core risk area is an area with concentrated risks, significant impacts, high accident frequency and serious losses, and the characteristics of the core risk area include spatial distribution, risk level and main risk source type.

[0021] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0022] By correcting terrain data in real time, accurately predicting changes in terrain and marine environment, and using measured data to dynamically calibrate models, the accuracy and real-time response capabilities of marine environmental predictions have been greatly improved. It can not only provide timely warnings of potential risks, but also provide scientific and efficient technical support for risk prevention and control, resource management and decision-making in coastal areas, effectively improving the accuracy and real-time nature of marine environmental predictions, and providing strong support for risk management and decision-making in coastal areas. It has significant technical effects and practical application value.

[0023] By dividing coastal risk zones and determining grid unit risk levels, merging adjacent high-risk grids into groups and calculating current data and connectivity, a risk propagation matrix is ​​constructed to predict diffusion paths, dynamically adjust group risk levels, and further predict future risk level changes. This technical solution effectively integrates spatial continuity, current connectivity, and the dynamic changes in risk, enabling accurate characterization and forward-looking predictions of coastal risk conditions. This provides a scientific basis for risk management and emergency response, significantly improving the efficiency and accuracy of risk prevention and control.

[0024] By dynamically identifying core risk areas, tracking critical boundaries within marshaling groups, and building a risk state transition model, the system significantly improves the timeliness and accuracy of risk warnings. It also enables adaptive adjustment of marshaling divisions and intelligent generation of prevention and control strategies. This solution not only efficiently captures and rapidly responds to sudden risk changes, but also optimizes resource allocation efficiency, reduces prevention and control costs, and enhances the system's adaptability to complex risk environments, effectively ensuring regional safety and mitigating potential losses.

[0025] By building a dynamic model for risk evolution intervention and an active immune system, the approach shifts from passive response to proactive intervention in marine risk management. This solution accurately identifies high-risk nodes and quantifies exclusion relationships, dynamically optimizing resource allocation and significantly improving resource utilization efficiency, rapidly breaking the risk evolution chain. Furthermore, the system continuously evolves through a negative feedback mechanism, enhancing decision-making accuracy and establishing an autonomous immune system for coastal risk prevention and control. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 This is a flow chart of a method for identifying marine environmental risks based on big data according to an embodiment of the present invention. DETAILED DESCRIPTION

[0027] To facilitate understanding of the present invention, the present application will be described more comprehensively below with reference to the relevant drawings; the drawings show preferred embodiments of the present invention, but the present invention can be implemented in many different forms and is not limited to the embodiments described herein; on the contrary, the purpose of providing these embodiments is to enable a more thorough and comprehensive understanding of the disclosed content of the present invention.

[0028] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this invention pertains; the terms used herein in the specification of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention; the term "and / or" used herein includes any and all combinations of one or more of the associated listed items.

[0029] Example 1: Figure 1 This is a flow chart of a method for identifying marine environmental risks based on big data according to an embodiment of the present invention.

[0030] like Figure 1 As shown, a method for identifying marine environmental risks based on big data includes the following steps:

[0031] S1, based on the initial static topographic data of the surveyed coastal area and the dynamic sediment migration data, a terrain prediction model is constructed, and the static topographic data is corrected in real time based on the sediment migration direction and speed.

[0032] Specifically, comprehensive mapping of coastal areas was conducted using a variety of marine environmental background data acquisition equipment (such as single-beam and multi-beam bathymetry systems, satellite remote sensing inversion, drone-based oblique photogrammetry, and GNSSRTK measurement systems). Initial static terrain data was denoised to remove outliers caused by measurement errors; data interpolation was performed to fill in data gaps and make the terrain data more continuous and complete; and coordinate transformation was performed to unify the coordinate systems of different data sources for subsequent data fusion and analysis.

[0033] Sediment monitoring devices are installed at key locations along the coast. Sediment samplers regularly collect sediment samples and analyze physical properties such as particle size and composition. Acoustic Doppler Current Profilers (ADCPs) measure ocean current speed and direction in real time, indirectly obtaining information on sediment migration by integrating the relationship between sediment movement and ocean currents. Satellite remote sensing imagery can analyze sediment distribution and trends over large areas of the ocean. Sediment migration data undergoes quality control to remove noise and anomalies. Data time alignment is performed to ensure data from different monitoring devices are synchronized. Spatial interpolation is performed to convert discrete monitoring data into continuous spatially distributed data.

[0034] A terrain prediction model was constructed based on the topographic characteristics and sediment migration patterns of coastal areas. Key physical parameters in the numerical model (including sediment density, particle size distribution, settling velocity, erosion rate coefficient, and deposition rate coefficient) were determined based on these characteristics. Initial static terrain data and some dynamic sediment migration data (sediment flux time series under tidal-wave coupling) were integrated into a training set and fed into the physically driven numerical model. Model parameters were adjusted through parameter sensitivity analysis and iterative optimization algorithms to minimize the error between the simulated terrain evolution results and historical observations. The remaining dynamic sediment migration data was then used as an independent validation set to evaluate the model's prediction accuracy through quantitative indicators and spatial distribution comparisons, ultimately forming a physical prediction model capable of simulating future terrain changes.

[0035] Analyze the direction and velocity of sediment migration using dynamic sediment migration data. Determine the direction of sediment migration by calculating the change in sediment distribution between adjacent time periods. Calculate the velocity of sediment migration based on the migration distance and time. For example, if sediment distribution data is collected at two different time points and compared to determine that sediment migrated from location A to location B, with a migration distance of d and a migration time of t, then the migration velocity v = d / t.

[0036] Based on the direction and velocity of sediment migration, the amount of sediment moved and accumulated per unit time is calculated. Assuming the sediment concentration is C (mass of sediment per unit volume) and the migration velocity is v, the mass flow rate of sediment passing through a cross section per unit time is q = Cv. Based on the patterns of sediment accumulation and erosion, the change in terrain height is calculated. For example, if sediment accumulates in a certain area, the terrain height increases; if sediment erodes, the terrain height decreases. Let the mass of accumulated or eroded sediment be m, the area of ​​the accumulation or erosion region be A, and the sediment density be ρ. The change in terrain height, Δh, is then calculated as m / (ρA).

[0037] The calculated terrain height changes are used to update the initial static terrain data in real time. Geographic Information System (GIS) technology can be used to superimpose the terrain height changes on the initial terrain data to generate revised terrain data.

[0038] S2, combined with the corrected terrain data and current meteorological and ocean current data, is input into the terrain prediction model after space-time matching to generate the future terrain, and is used as the initial input to integrate the meteorological-ocean current boundary conditions to construct a numerical model of the ocean environment.

[0039] Specifically, in step S1, coastal terrain data that has been corrected in real time is obtained. This data is usually in the form of a digital elevation model (DEM) and contains terrain height information at different locations along the coast. Current meteorological data, such as wind speed (u-wind and v-wind, representing the east-west and north-south wind speed components, respectively), wind direction, air pressure (P), temperature (T), and humidity (RH), are collected through weather stations, satellite remote sensing, buoys, and other equipment. Current ocean current data, including current velocity (u-current and v-current, representing the east-west and north-south wind speed components, respectively) and current direction, are obtained using equipment such as acoustic Doppler current profilers (ADCPs) and current meters.

[0040] The corrected terrain data is spatially and temporally aligned with the current meteorological and ocean current data. Because different data sources may have different resolutions and sampling times, interpolation is required to align all data on the same spatial grid and time step. For example, for meteorological data with low spatial resolution, bilinear interpolation can be used to interpolate it onto the terrain data grid. For data that does not match temporally, linear or spline interpolation can be used.

[0041] Select an appropriate topography change prediction model based on the characteristics of the coastal area and the research objectives. Common models include numerical models based on physical principles, such as wave-sediment-topography interaction models and current-sediment transport models. These models are typically based on fundamental principles such as fluid mechanics and sediment kinematics.

[0042] Determine key model parameters, such as sediment particle size (d), density (ρs), settling velocity (ws), erosion coefficient (E), and deposition coefficient (D). These parameters can be obtained through laboratory experiments, field observations, and literature. For example, laboratory flume experiments can be used to measure the settling velocity of sediments of different particle sizes; and field observations can be used to determine the erosion and deposition coefficients.

[0043] The matched data is fed into a terrain change prediction model to simulate terrain changes within a time window (e.g., one week or one month in the future). The model calculates sediment transport, erosion, and deposition based on meteorological and ocean current conditions, thereby deriving changes in terrain elevation. For example, under the influence of strong winds and ocean currents, the model simulates the movement of sandbars, erosion, and accumulation of beaches, and other phenomena.

[0044] Compare and verify the simulated topographic changes with historical data or field observations. If the simulation results deviate significantly from the actual situation, adjust the model parameters or improve the model algorithm and repeat the simulation until the simulation results meet the accuracy requirements.

[0045] Based on the research objectives and the characteristics of the coastal marine environment, appropriate marine environment numerical models should be selected, such as wave models (e.g., SWAN, WAVEWATCH III), current models (e.g., ROMS, FVCOM), and air-sea coupling models (e.g., COAWST). The predicted topographic changes should be used as the initial topographic input for the marine environment numerical model. Furthermore, current meteorological and ocean current data should be used as the initial boundary conditions for the model. For example, in a wave model, wind speed and direction should be used as the boundary conditions for wave generation; in a current model, current speed and direction should be used as the initial flow field.

[0046] Determine other parameters in the ocean environment numerical model, such as the ocean's turbulence coefficient, heat exchange coefficient, bottom friction coefficient, etc. These parameters have a significant impact on the model's simulation results and can be determined through literature research, empirical formulas, or field observation data.

[0047] Run a numerical ocean environment model to obtain the spatiotemporal distribution of ocean environmental parameters (such as significant wave height, current velocity, water temperature, and salinity) within the time window. Analyze the simulation results to assess the model's performance and the rationality of the results. For example, by comparing the results with field observations, analyze the sources of model errors and identify areas for improvement.

[0048] S3, collect measured data from coastal areas and conduct comparative analysis with model prediction results to calibrate the numerical model of the marine environment.

[0049] Among them, the measured data include measured ocean dynamic environment parameters and meteorological parameters.

[0050] Specifically, ocean environment monitoring buoys and tide gauges are used to collect measured ocean dynamic environmental parameters (such as buoy position, significant wave height, and ocean currents) and meteorological parameters (such as wind speed and direction, precipitation, air pressure, and humidity) in coastal areas. These measured data are compared and analyzed with predictions from a numerical ocean environment model, and the error between the measured and predicted data is calculated. Based on the results of this comparative analysis, the parameters of the numerical ocean environment model are adjusted and optimized. For example, adjustments are made to the friction coefficient and turbulence coefficient in the model to reduce prediction errors and improve model accuracy.

[0051] S4, using the calibrated ocean environment numerical model, based on the real-time monitored ocean dynamic environment parameters, determines whether numerical simulation is necessary.

[0052] Specifically, marine environmental monitoring equipment is used to monitor coastal areas' ocean dynamic environmental parameters in real time. Based on a calibrated ocean environmental numerical model, the system assesses the current ocean environmental conditions and determines whether further numerical simulations are necessary. For example, if a real-time monitored ocean dynamic environmental parameter (such as significant wave height) exceeds a preset threshold, numerical simulations are considered necessary to predict future ocean environmental changes.

[0053] S5, conduct a comprehensive analysis of the simulated future ocean dynamic environment parameters to obtain risk values, pre-set the corresponding relationship between different risk value ranges and risk levels, and divide the coastal areas into risk zones based on the calculated risk values.

[0054] Among them, risk areas can include high-risk areas, medium-risk areas and low-risk areas.

[0055] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0056] By correcting terrain data in real time, accurately predicting changes in terrain and marine environment, and using measured data to dynamically calibrate models, the accuracy and real-time response capabilities of marine environmental predictions have been greatly improved. It can not only provide timely warnings of potential risks, but also provide scientific and efficient technical support for risk prevention and control, resource management and decision-making in coastal areas, effectively improving the accuracy and real-time nature of marine environmental predictions, and providing strong support for risk management and decision-making in coastal areas. It has significant technical effects and practical application value.

[0057] Example 2: In Example 1, when conducting a preliminary assessment of risk areas in coastal areas, risk areas are divided and grid unit risk levels are determined based only on a single-dimensional risk indicator (such as historical accident rate, environmental vulnerability, etc.). However, there are significant differences in the sources of risk, influencing factors, and interaction mechanisms in different regions, resulting in complex and changeable risk propagation characteristics. Relying solely on a single indicator for risk assessment tends to ignore spatial continuity, ocean current dynamics, and the associated propagation effects between risks, thereby affecting the accuracy of risk level division and the reliability of risk diffusion path prediction. In order to more comprehensively capture the spatial distribution characteristics, dynamic evolution laws, and associated propagation mechanisms of risks in coastal areas, and thereby improve the accuracy of dynamic adjustment of risk levels and predictions of future risk changes, it is necessary to further optimize and improve existing technical solutions.

[0058] In some embodiments, the coastal risk areas are divided, and step S5 further includes:

[0059] S51, determine the risk level of each grid cell according to the divided coastal risk zone, and identify grid cells with spatial continuity.

[0060] Here, spatial continuity means that two grid cells i and j are spatially continuous if they are geographically adjacent (e.g., share an edge or a vertex).

[0061] Specifically, the risk level of each grid unit is determined by the threshold. Risk zone division result: Each grid unit corresponds to a risk value Set low, medium, and high risk thresholds 、 、 ,satisfy .

[0062] For each grid cell i, according to its risk value And thresholds are used to determine the risk level:

[0063]

[0064] Each grid cell i is assigned a .

[0065] Identify spatially continuous high-risk areas through the region growing algorithm. Select an unvisited high-risk grid cell i (if the goal is to identify high-risk areas), mark i as visited, and add it to the current area S. Traverse all adjacent grid cells j of i. If j has the same risk level as i (for example, both are high risk) and j has not been visited, mark j as visited and add it to S. When S no longer grows, stop growing, and S is a high-risk area with spatial continuity. Repeat the above process for all unvisited high-risk grid cells until all high-risk grid cells have been visited. Let the set of grid cells be G, and the risk level be .

[0066] The spatial continuity region S satisfies: S G, i,j∈S, and i and j are adjacent .

[0067] S52: Merge adjacent high-risk grids into groups, calculate the ocean current data of each group, and for any two adjacent groups, calculate the ocean current connectivity between them.

[0068] Specifically, adjacent high-risk grids are merged into groups using a region growing algorithm. An empty group set G is created, along with a marker array to record whether each grid cell has been assigned to a group. Initially, all grid cells are marked as unvisited. All grid cells are traversed to find an unvisited high-risk grid cell. Using this grid cell as the starting point, a new group is created and added to the group set, marking it as visited. Using a breadth-first search or depth-first search, the adjacent grid cells of all grid cells in the current group are traversed. If the adjacent grid cell is high-risk and unvisited, it is added to the current group and marked as visited.

[0069] Repeat the above process until the current group no longer grows. Continue to traverse the remaining unvisited grid cells until all high-risk grid cells are assigned to a group.

[0070] Calculate the average current speed, magnitude, and direction for each group. For each group in the group set, calculate the average current speed vector: sum the easting and northing velocities of all grid cells in the group and divide by the number of grid cells in the group to obtain the group's average easting and average northing speeds. Calculate the average current speed magnitude and direction: Calculate the group's average current speed magnitude (i.e., the modulus of the speed vector) based on the average easting and average northing speeds. Calculate the group's average current direction (counterclockwise, relative to east) using the four-quadrant inverse tangent function.

[0071] The current connectivity between groups is determined by the angle between the current direction and the normal vector of the group boundary, and the connectivity matrix is ​​filled. Traverse all adjacent group pairs in the group set, that is, group pairs that share at least one grid cell boundary. For each pair of adjacent groups, calculate the angle between their average current directions (take the minimum angle, that is, the angle less than or equal to 180 degrees). For each pair of adjacent groups, determine the normal vector of their shared boundary (pointing in the direction of the second group) and calculate the angle between the average current velocity vector of the first group and the normal vector. If the angle is less than or equal to 90 degrees, it is considered that the current can flow from the first group to the second group, and it is marked that there is current connectivity between the two groups. Create a connectivity matrix to record the current connectivity between all group pairs. If there is current connectivity between the two groups, the corresponding position in the connectivity matrix is ​​marked as 1; otherwise, it is marked as 0.

[0072] S53, based on the ocean current connectivity between formations, analyzes the correlation between formations and constructs a risk propagation matrix to predict the risk diffusion path between different formations.

[0073] Specifically, based on the ocean current connectivity matrix, the direct and indirect connections between marshalings are analyzed. The ocean current connectivity matrix C is used to determine the direct connectivity between marshalings. =1, then group and Direct connection. Traverse the connectivity matrix and record all directly connected group pairs to form a direct association set. Calculate the power of the connectivity matrix (such as ), identify the marshaling pairs connected through the intermediate marshaling, for The non-zero elements in , indicating the existence of an indirect correlation with two propagation steps. The direct correlation weight is defined as 1, and the indirect correlation weight decays according to the number of propagation steps (e.g., the two-step propagation weight is 0.5, the three-step weight is 0.25, etc.).

[0074] Construct the risk transmission matrix and calculate the direct and indirect transmission probabilities. Create a transmission matrix P with the same size as the connectivity matrix, with the initial elements set to 0. For the directly connected group pairs ( , ), calculates the propagation probability based on the ocean current speed and direction ,Will Assign to The propagation probability can be expressed as: ,in It is a group The average ocean current speed, yes The angle between the mean current direction and the shared boundary normal vector, is the maximum value of the ocean current velocity in all marshalings. For the indirectly connected marshaling pairs ( , ), the indirect transmission probability is calculated based on the product of the direct transmission probability and the attenuation factor, Accumulated to For example, for two-step propagation: ,in is the middle group, and the attenuation factor is 0.5. Each row of the propagation matrix P is normalized so that the sum of each row does not exceed 1.

[0075] By iteratively calculating risk diffusion, we can predict the diffusion path of risk among different groups and analyze key nodes. Define the initial risk distribution vector ,in Indicates grouping The initial risk value (such as the proportion of high-risk grid cells) is calculated by iteratively using the propagation matrix P: ,in is the risk distribution vector at step t, is the risk distribution vector at step t+1. Set the number of iteration steps T, or stop the iteration when the risk distribution change is less than a certain threshold. During the iteration process, record which groups the risk propagates from and to. Observe By analyzing the changes in the risk matrix and the diffusion paths, we can identify the main paths of risk diffusion (i.e., the groups with higher probability of risk transmission). Through the risk diffusion matrix and the diffusion paths, we can identify the key nodes of risk diffusion (i.e., the groups with higher probability of risk transmission or receiving more risks).

[0076] For example, assuming the group and Direct connection, The average current speed is 1.0308, The average ocean current speed is 1.1705. Assume ,but: ,therefore, Initial risk distribution ,first step: , step 2: (Assume and ), the risk gradually changes from spread to .

[0077] S54, dynamically adjust the risk level of the formation based on the risk diffusion path prediction results and the correlation analysis between formations.

[0078] Specifically, risk diffusion paths are integrated with group correlation analysis to identify key groups and associated paths. Based on the iterative results of the risk propagation matrix, the main risk diffusion paths are extracted. The groups from which the risk propagates to which groups in each iteration are recorded to form a set of risk diffusion paths. Using the results of inter-group correlation analysis (including direct and indirect correlations), a group correlation graph is constructed. In this graph, nodes represent groups, and edges represent inter-group correlations (weights can be based on propagation probability or correlation strength). The risk diffusion paths are integrated with the group correlation graph to identify the key groups and associated paths involved in the risk diffusion process.

[0079] Calculate the cumulative risk value of the group, taking into account the time decay factor. Initialize a risk accumulation value , initially Traverse the risk diffusion path, and for each step of the propagation, add the propagated risk value to the target group’s risk accumulation value. For example, if the risk is from group Propagate to Group , the risk value of transmission is ΔR, then update The propagated risk value ΔR can be calculated based on the elements of the risk propagation matrix and the current risk distribution vector. To reflect the characteristic of risk decay over time, a time decay factor can be introduced when updating the risk accumulation value. For example, for the risk propagated at an earlier time step, the risk value accumulated to the current group can be multiplied by a decay factor less than 1. .

[0080] According to the risk accumulation value and risk level classification standard, the risk level of the group is dynamically adjusted, taking into account the impact of correlation. According to the range of risk accumulation value, different risk levels are divided, and corresponding different risk accumulation value thresholds are determined, and the threshold corresponding to each risk level is determined. Traverse all groups and according to their risk accumulation value The current risk level is determined based on the risk classification criteria. If a group's cumulative risk value exceeds a certain risk level threshold, its risk level is adjusted to that level or higher (based on actual needs). For groups directly or indirectly associated with high-risk groups, their risk levels can be appropriately increased to reflect potential risks, even if their current cumulative risk value does not reach the high-risk threshold. The extent of the adjustment can be determined based on the strength of the association and the probability of transmission.

[0081] Continuously monitor and update risk diffusion paths, risk accumulation values, and risk levels, providing feedback and optimizing adjustment strategies. Risk diffusion paths are regularly recalculated as factors such as current conditions and risk source status change. Based on the new risk diffusion paths, the group's risk accumulation value is recalculated, and the group's risk level is dynamically adjusted based on the reassessed risk accumulation value. Risk level adjustment results are fed back to the risk management and emergency response systems to guide decision-making and actions. Based on actual results and feedback, the risk classification criteria and adjustment strategies are continuously optimized.

[0082] S55, based on the risk level of the dynamically adjusted formation and the degree of change in the risk level, predict the degree of change in the risk level of each formation in the future time window.

[0083] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0084] By dividing coastal risk zones and determining grid unit risk levels, merging adjacent high-risk grids into groups and calculating current data and connectivity, a risk propagation matrix is ​​constructed to predict diffusion paths, dynamically adjust group risk levels, and further predict future risk level changes. This technical solution effectively integrates spatial continuity, current connectivity, and the dynamic changes in risk, enabling accurate characterization and forward-looking predictions of coastal risk conditions. This provides a scientific basis for risk management and emergency response, significantly improving the efficiency and accuracy of risk prevention and control.

[0085] Example 3: In Example 2, when fixed grouping and static risk prevention and control strategies were employed, the system's response to sudden risk changes was significantly delayed in the face of a complex and volatile risk environment. Furthermore, due to the lack of adequate consideration of risk level differences between groups and the dynamic nature of core risk areas, this led to irrational allocation of prevention and control resources and insufficient risk warning accuracy. In particular, when core risk areas expanded or risk levels changed rapidly, static grouping boundaries and prevention and control strategies struggled to adapt to the actual risk transmission paths, resulting in inefficient prevention and control and increased potential losses.

[0086] In some embodiments, in step S52, merging adjacent high-risk grids into a group further includes:

[0087] 2A, based on the risk level of each grid unit in the group, identify the core risk area within the group and extract the characteristics of the core risk area.

[0088] Core risk areas are areas with concentrated risks, significant impacts, high accident frequency, and severe losses. The characteristics of core risk areas include spatial distribution, risk level distribution, and major risk source types.

[0089] 2B. By analyzing the risk level time series of the core risk area, the critical point of the core risk area is determined, and based on the spatial distribution and risk propagation characteristics of the core risk area, the dynamic critical line of the grouping boundary is determined.

[0090] 2C, define the risk state transfer rules based on the critical points of the core risk area and the dynamic critical lines of the grouping boundaries.

[0091] Among them, the risk state transfer rules include: grid unit risk level transfer rules and grouping boundary critical line expansion / contraction rules. The grid unit risk level transfer rules define the transition conditions from low risk to high risk. When the risk level of a grid unit exceeds the critical point of the core risk area for N consecutive time steps, it is judged to be in a high risk state. The transition conditions from high risk to low risk are defined. When the risk level of a grid unit is lower than the critical point of the core risk area for M consecutive time steps, it is judged to be in a low risk state. The grouping boundary critical line expansion / contraction rules define the expansion conditions. When the expansion of the core risk area causes the risk level of the adjacent grid unit to exceed the critical point, and the number of newly added high-risk grid units exceeds the threshold, the grouping boundary critical line expands outward. The contraction conditions are defined. When the shrinkage of the core risk area causes the risk level of the adjacent grid unit to be lower than the critical point, and the number of reduced high-risk grid units exceeds the threshold, the grouping boundary critical line contracts inward.

[0092] 2D, build a risk state transfer model to simulate the dynamic changes of risk level and grouping boundary.

[0093] Specifically, a Markov chain model was used to train the risk state transition model, utilizing historical risk data, critical points in core risk areas, and dynamic critical lines at group boundaries as training samples. Model parameters were optimized through cross-validation and grid search methods to improve model prediction accuracy. Model performance was evaluated using validation data to ensure that the model accurately simulated the dynamic changes in risk levels and group boundaries. The trained model was then applied to real-time risk data to predict future changes in risk states and group boundaries.

[0094] 2E, based on the risk state transfer model, design mutation warning indicators and mutation warning thresholds to trigger mutation warning measures.

[0095] Mutation warning indicators include: the risk level change rate, which measures the maximum or average rate of increase in risk level within the calculation window; the group boundary critical line movement speed, which measures the average or maximum distance the critical line expands or contracts within the calculation window; and the core risk area expansion area, which measures the total area or percentage of newly added high-risk grid cells within the calculation window. When any mutation warning indicator exceeds a threshold, a mutation warning is triggered, initiating appropriate emergency response measures based on the warning level.

[0096] It should be noted that based on the dynamic critical line of the grouping boundary and the changing trend of the risk level, the grouping division rules can be dynamically adjusted and an adaptive prevention and control strategy can be generated.

[0097] Dynamic adjustments to group division rules include: merging groups. When the expansion of the core risk area causes the critical boundaries of adjacent groups to overlap, two groups are merged into a new, larger group. Splitting groups. When the risk levels within a group differ significantly (e.g., the proportion of high-risk and low-risk areas exceeds a threshold), a group is split into multiple smaller groups. Group attribute updates. Based on the risk level distribution and the main risk source types within the group, the group's risk prevention and control strategies (e.g., increasing monitoring frequency, adjusting emergency response plans) and resource allocation (e.g., increasing rescue forces, stockpiling emergency supplies) are updated.

[0098] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0099] By dynamically identifying core risk areas, tracking critical boundaries within marshaling groups, and building a risk state transition model, the system significantly improves the timeliness and accuracy of risk warnings. It also enables adaptive adjustment of marshaling divisions and intelligent generation of prevention and control strategies. This solution not only efficiently captures and rapidly responds to sudden risk changes, but also optimizes resource allocation efficiency, reduces prevention and control costs, and enhances the system's adaptability to complex risk environments, effectively ensuring regional safety and mitigating potential losses.

[0100] Example 4: In Example 3, the comparison object of the self-testing electric energy meter is further limited, that is, a reference electric energy meter is selected from the remaining electric energy meters on the power line to compare the voltage values, thereby improving the accuracy and pertinence of the subsequent abnormal probability value calculation. Based on this, although the accuracy of the self-test is improved by selecting electric energy meters with similar power consumption status as references, a unified preset difference threshold is still used to judge the voltage deviation. The fixed threshold method may not be able to fully reflect the actual differences between different electric energy meters, resulting in limited accuracy of abnormal detection. In order to make the electric energy meter detection results more reliable and accurate, the determination of the abnormal probability value of the electric energy meter is further limited.

[0101] In some embodiments, in step 2E, based on the risk state transition model, the method further includes:

[0102] Based on the risk state transfer model and the dynamic boundary data of the core risk area, combined with the exclusion relationship between marine environmental parameters, historical risk event data and artificial intervention measures, a risk evolution intervention dynamic model is constructed to actively guide the evolution of the marine environment in the expected direction and generate a risk evolution chain; combined with real-time monitoring of the risk evolution chain, intervention nodes with high exclusion intensity and high risk level are dynamically screened.

[0103] Specifically, based on the risk state transfer model and the dynamic boundary data of the core risk area, combined with the marine environmental parameters (such as flow rate, tide, temperature) and historical risk event data, a risk evolution causal network diagram is constructed. Specifically, the key chain nodes of risk evolution (such as rift formation points, pollutant diffusion path inflection points, extreme sea condition triggering thresholds) are identified and marked as high-priority intervention targets. Natural flow rate thresholds (such as ) and the target flow rate threshold after intervention (e.g. ), and then define the flow rate adjustment range ( → The exclusion coefficient (R, ranging from 0 ≤ R ≤ 1) is used to determine the thresholds of changes in ocean environmental parameters (such as current velocity and direction) under different exclusion coefficients. For example, when R > 0.7, rip currents can dissipate automatically.

[0104] A risk evolution intervention dynamic model is introduced to define the oppositional relationship between key data elements (such as the repulsive strength between natural flow rate and artificial intervention flow rate) and transformation rules (such as guiding the direction of ocean current through artificial ditches). The model inputs real-time marine environmental data (flow rate, temperature, salinity), the status of risk evolution chain nodes and artificial intervention resources (such as ditch flow, pump station power), and outputs the repulsive strength (R) and intervention effect (such as the probability of rift dissipation and the inhibition rate of pollutant diffusion) of each intervention node. Based on the risk evolution intervention dynamic model, a dynamic intervention strategy is designed. For example, when the risk of rift formation is monitored, reverse water flow is injected through artificial ditches to adjust the local flow rate to , and ensure that R>0.7, thereby triggering the automatic dissipation of rip flow. When pollutants diffuse to key nodes, the flow field direction is changed by adjusting the rejection coefficient R, guiding the pollutants to non-sensitive areas.

[0105] Based on real-time monitoring data, the current repulsion strength (R) and intervention potential (e.g., flow rate regulation capacity, resource consumption efficiency) of each intervention node are calculated. Intervention priority metrics (e.g., the ratio of risk dissipation rate to resource consumption ratio) are defined to identify the optimal intervention nodes. A multi-objective optimization algorithm (e.g., NSGA-II) is used to dynamically select the optimal combination of intervention nodes, combining factors such as repulsion strength, risk level, and resource constraints. For example, when resources are limited, priority intervention is given to nodes with high repulsion strength (R>0.8) and high risk level (e.g., rip current core areas) to maximize resource efficiency.

[0106] Using a dynamic risk evolution intervention model, we predict the path of risk evolution and deploy intervention measures (such as preemptively activating artificial ditches) to proactively alter marine environmental parameters and break the risk chain. We dynamically assess the exclusion potential of each node, prioritizing intervention at nodes with high exclusion intensity to improve resource efficiency. We also introduce a negative feedback mechanism to reversely optimize the parameters of the exclusion-transformation model (e.g., calibrating the exclusion coefficient R) based on actual intervention results (e.g., rip current dissipation time and pollutant diffusion range).

[0107] By integrating a dynamic risk evolution intervention model, a dynamic resource allocation algorithm, and a real-time monitoring network, we construct an active immune system for coastal risk prevention and control. This system monitors marine environmental parameters and the status of nodes in the risk evolution chain in real time, dynamically generating optimal intervention strategies and automatically executing them (for example, regulating flow in artificial ditches through intelligent pumping stations). The system continuously learns and optimizes its decision-making logic, enabling autonomous evolution of risk prevention and control. Furthermore, the system provides a scalable modular architecture, allowing customization based on risk characteristics for different coastal regions (e.g., areas prone to rip currents or areas sensitive to pollutants).

[0108] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0109] By building a dynamic model for risk evolution and intervention and an active immune system, the solution has achieved a shift from passive response to proactive intervention in marine risk management. This solution accurately identifies high-risk nodes and quantifies exclusion relationships, dynamically optimizing resource allocation, significantly improving resource utilization efficiency, and rapidly breaking the risk evolution chain. Furthermore, the system continuously evolves through a negative feedback mechanism, enhancing decision-making accuracy and establishing an autonomous immune system for coastal risk prevention and control. This system achieves three paradigm shifts in marine risk prevention and control: a shift from "monitoring-prevention" to "prediction-intervention," proactively regulating marine environmental parameters to disrupt the risk evolution chain; an upgrade from "uniform treatment" to an "exclusion priority" strategy, dynamically selecting high-value intervention nodes and increasing resource efficiency by twofold; and an evolution from a "static model" to a "reverse self-evolution" system, leveraging negative feedback to continuously optimize decision-making logic. Ultimately, this system establishes an "active immune system" for coastal risk prevention and control, providing a disruptive, adaptive, and efficient solution for global marine safety management.

[0110] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Various modifications and variations are readily apparent to those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A method for identifying marine environmental risks based on big data, characterized in that: include: S1, based on the initial static topographic data and dynamic sediment migration data of the surveyed coastal area, a topographic prediction model is constructed, and the static topographic data is corrected in real time based on the direction and speed of sediment migration; S2, combining the revised terrain data with current meteorological and ocean current data, and inputting them into the terrain prediction model after space-time matching to generate future terrain, and using them as the initial input to integrate meteorological-ocean current boundary conditions to construct a numerical model of the ocean environment; S3, collect measured data from coastal areas and conduct comparative analysis with model prediction results to calibrate the marine environment numerical model; S4, using the calibrated ocean environment numerical model and according to the real-time monitored ocean dynamic environment parameters, determine whether numerical simulation is necessary; S5, conduct a comprehensive analysis of the simulation results to obtain the risk value, divide the coastal risk zone and determine the grid unit risk level, and identify the spatial continuity grid; The identification of spatial continuity grid is to determine the risk level of each grid unit; Each grid unit is judged based on its risk value and threshold; each grid unit is assigned a risk level; Identify spatially continuous high-risk areas; where spatially continuous high-risk areas are spatially continuous areas S that satisfy, , , , and i and j are adjacent ⇒ , where the set of grid cells is G and the risk level is , the unvisited high-risk grid cell is i, and all adjacent grid cells of i are j; merge adjacent high-risk grids into groups, calculate the group current data and connectivity, construct a risk propagation matrix, predict the risk diffusion path, dynamically adjust the group risk level, and predict the degree of change in the risk level in the future time window; the predicted risk diffusion path is based on the current connectivity matrix, analyzes the direct and indirect correlations between groups; constructs a risk propagation matrix, calculates the direct and indirect propagation probability; through iterative calculation of risk diffusion, predicts the risk diffusion path between different groups, and analyzes the key nodes; The method of merging adjacent high-risk grids into a group is as follows: based on the risk level of each grid unit in the group, the core risk area in the group is identified and features are extracted; the risk level time series of the core risk area is analyzed to determine the critical point, and based on the spatial distribution and risk propagation characteristics of the core risk area, the dynamic critical line of the group boundary is determined; the risk state transfer rules based on the critical point of the core risk area and the dynamic critical line of the group boundary are defined; a risk state transfer model is constructed to simulate the dynamic change process of the risk level and the group boundary; based on the risk state transfer model, a mutation warning indicator and a mutation warning threshold are designed to trigger mutation warning measures.

2. The method for identifying marine environmental risks based on big data according to claim 1, characterized in that: The S4, determining whether numerical simulation is required, specifically includes: utilizing marine environment monitoring equipment to monitor the marine dynamic environment parameters of coastal areas in real time; based on a calibrated marine environment numerical model, evaluating the current marine environment conditions according to the marine dynamic environment parameters monitored in real time, and determining whether further numerical simulation is required.

3. The method for identifying marine environmental risks based on big data according to claim 1, characterized in that: The S2, constructing a numerical model of the marine environment, includes: obtaining real-time corrected terrain data of the coastal area, spatially and temporally matching the corrected terrain data with current meteorological and ocean current data, inputting the matched data into a terrain change prediction model, and simulating terrain changes within a window time; and comparing and verifying the simulated terrain change results with historical data or on-site observation data.

4. The method for identifying marine environmental risks based on big data according to claim 1, characterized in that: The predicted risk diffusion path also includes: integrating the risk diffusion path with the group correlation analysis to identify key groups and related paths; calculating the risk accumulation value of the group, considering the time decay factor; dynamically adjusting the risk level of the group according to the risk accumulation value and the risk level classification standard, considering the correlation impact.

5. The method for identifying marine environmental risks based on big data according to claim 1, characterized in that: The mutation warning indicators include: the risk level change rate, which is the maximum increase or average increase rate of the risk level within the calculation window time; the grouping boundary critical line moving speed, which is the average distance or maximum distance of the critical line expansion or contraction within the calculation window time; the core risk area expansion area, which is the total area or proportion of newly added high-risk grid units within the calculation window time.

6. The method for identifying marine environmental risks based on big data according to claim 1, characterized in that: The core risk areas include: core risk areas are areas with concentrated risks, significant impacts, high accident frequency and severe losses. The characteristics of core risk areas include spatial distribution, risk level and main risk source types.

Images