Method for tracing the source and monitoring of marine pollution based on satellite remote sensing and unmanned aerial vehicle

Abstract

The method for tracing and monitoring marine pollution based on satellite remote sensing and unmanned aerial vehicles provided by this application relates to the technical field of marine pollution tracing and monitoring, and includes: configuring a pollution diffusion path prediction model and a data-driven model based on the range pollution data, precise pollution data, and deep-water data after multi-modal data fusion to obtain a pollution prediction result and pollution error data, performing hybrid-driven fusion based on the pollution prediction result and pollution error data, and tracing the initial source of pollutants to obtain pollution source tracing data. Through the preprocessing and fusion of multi-modal data, combined with range pollution data, precise pollution data, and deep-water data, the pollution diffusion path prediction model and the data-driven model are dynamically adjusted, thereby generating more accurate pollution prediction results and pollution error data. At the same time, through reverse diffusion simulation and pollution source tracing analysis, the efficiency of pollution source positioning and decision support is improved, providing accurate and decision-making basis for dealing with marine pollution incidents.

Description

Technical Field

[0001] The present invention relates to the technical field of marine pollution source tracing and monitoring. Specifically, it relates to a method for marine pollution source tracing and monitoring based on satellite remote sensing and unmanned aerial vehicles. Background Art

[0002] With the increasing severity of marine pollution problems, tracking and monitoring pollution sources have become particularly important. Existing technologies mostly adopt single monitoring means, such as satellite remote sensing or unmanned aerial vehicle data collection methods. Although they can cover large-area monitoring requirements, they have limitations in terms of data accuracy, deep pollution detection, and real-time performance. The spread of marine pollution is comprehensively affected by hydrodynamic forces, pollution source intensity, and environmental parameters. Therefore, it is necessary to combine multiple data sources and analysis means to construct an accurate pollution source tracing and prediction system.

[0003] In practical applications, existing methods have independent and scattered data processing processes, and fail to fully integrate multi-modal data, resulting in low accuracy and reliability of prediction models. Finally, traditional decision-making systems fail to dynamically adjust strategies by combining historical risk data and real-time monitoring results, making it difficult to adapt to the rapid changes in the pollution situation, resulting in a lag in treatment effects. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for marine pollution source tracing and monitoring based on satellite remote sensing and unmanned aerial vehicles in view of the deficiencies of the prior art. The aim is to construct an all-round multi-modal pollution monitoring system by integrating satellite remote sensing, unmanned aerial vehicle data, and deep-water monitoring data, and use a hybrid drive model to achieve high-precision pollution diffusion prediction and source tracing. Combining dynamic evaluation and real-time optimization decision-making mechanisms, this method significantly improves the comprehensiveness of monitoring, the accuracy of prediction, and the flexibility of decision-making, providing a scientific basis for rapid and effective marine pollution treatment.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A method for marine pollution source tracing and monitoring based on satellite remote sensing and unmanned aerial vehicles includes the following:

[0007] Step S100: Obtain range pollution data, precise pollution data, and deep-water data.

[0008] Step S200: Perform preprocessing and multi-modal data fusion based on the obtained range pollution data, precise pollution data, and deep-water data.

[0009] Step S300: Configure a pollution diffusion path prediction model and a data-driven model based on the range pollution data, precise pollution data, and deep-water data after multi-modal data fusion, obtain pollution prediction results and pollution error data, perform hybrid-driven fusion based on the pollution prediction results and pollution error data, and track the initial source of pollutants to obtain pollution source tracing data.

[0010] Step S400: Generate three-dimensional volume data based on the obtained pollution source tracing data, the pollution prediction results, and the pollution error data after hybrid-driven fusion, and obtain a three-dimensional visual model through time series processing. Based on the obtained pollution source tracing data, mark the position of the pollution source through the three-dimensional visual model.

[0011] Step S500: Calculate preset classification thresholds and risk indices based on the pollution prediction results and pollution error data after hybrid-driven fusion to obtain risk index data and potential impact area data. Make a preset decision and optimize the preset decision based on the obtained risk index data and potential impact area data, and perform visualization processing and generate report content.

[0012] Step S600: Obtain the latest risk index data, compare the historical risk index data with the latest risk index data to obtain a comparison result, analyze the currently executed preset decision to obtain an analysis result, and generate a final report result based on the obtained report content, report result, comparison result, and analysis result.

[0013] As a preferred solution of the present invention, obtaining range pollution data, precise pollution data, and deep-water data specifically includes:

[0014] Step S100.1: Obtain range pollution data, precise pollution data, and deep-water data.

[0015] Conduct large-scale coverage monitoring of marine surface pollutants through satellite remote sensing to obtain range pollution data, which includes: pollutant distribution range, pollutant concentration, pollutant type, potential pollution source location, ocean current speed and direction, surface water temperature, and surface salinity.

[0016] The range pollution data preliminarily identifies the distribution and concentration of pollutants through high-resolution multi-spectral images of satellite remote sensing, extracts the spectral characteristics of pollutants using hyperspectral technology for substance classification, and captures temperature anomaly regions using thermal infrared to locate pollution sources.

[0017] Conduct high-precision monitoring of hotspots and areas difficult to cover by satellites through unmanned aerial vehicles to obtain precise pollution data, which includes: refined pollutant distribution, pollutant concentration, pollutant characteristics, specific pollution source types, biological pollution sources, local water temperature, local water flow speed, and local water transparency.

[0018] Precise pollution data is obtained by using a drone carrying a spectral sensor to identify the optical characteristics of different pollutants, obtaining a clear pollutant distribution map by using a high-resolution camera, and collecting water samples with an environmental DNA sampling device for biological source tracing analysis.

[0019] By using a buoy to monitor and collect the dissolved organic matter, heavy metal content, and pH value of columnar water bodies, deep water data is obtained. The deep water data includes: deep pollutant concentration, pollutant vertical distribution, deep water pollution source type, pollution source intensity, deep water temperature, deep water salinity, deep water flow velocity and direction, and water body density.

[0020] The deep water data is obtained by an underwater robot moving in the target area to collect deep water samples of pollutants.

[0021] As a preferred solution of the present invention, preprocessing and multi-modal data fusion are performed based on the obtained range pollution data, precise pollution data, and deep water data. Specifically:

[0022] Step S200.1: Perform preprocessing based on the obtained range pollution data, precise pollution data, and deep water data.

[0023] The preprocessing includes: image denoising, numerical data cleaning, and spectral data normalization.

[0024] For image denoising, a convolutional neural network is used to denoise the range pollution data, identify and repair the noise area, and perform geometric distortion correction on the precise pollution data. For numerical data cleaning, an RNN-based time series anomaly detection model is used to identify and remove sudden error data points. For spectral data normalization, normalization is used to uniformly adjust the spectral data collected by multiple sensors.

[0025] It should be noted here that since preprocessing is a well-known technology in the field, it will not be elaborated too much here.

[0026] Step S200.2: Perform multi-modal data fusion based on the preprocessed range pollution data, precise pollution data, and deep water data.

[0027] The multi-modal data fusion includes: spatio-temporal alignment, feature correlation enhancement, and weighted fusion.

[0028] Spatial-temporal alignment is based on the dynamic time warping algorithm to perform time series alignment on multi-source data. The data in the uncovered area is filled in using the geospatial interpolation algorithm. Feature correlation enhancement is achieved by mapping spectral data, thermal infrared data, and deep water physical parameters to a unified feature space through multi-modal feature maps. A correlation model between data is established using a graph neural network to improve the accuracy and robustness of data fusion. Weighted fusion dynamically assigns weights to the range pollution data, precise pollution data, and deep water data quality after spatial-temporal alignment and feature correlation enhancement. Specifically:

[0029]

[0030] Where: is the initial weight of the data, is the resolution factor, which is adjusted according to the relative quality of the resolution, is the timeliness factor, which is adjusted according to the impact of data delay, is the data integrity factor, which is adjusted according to the missing rate, are the weight impact factors corresponding to resolution, timeliness, and integrity.

[0031] It should be noted here that since multi-modal data fusion is a well-known technology in this field, it will not be elaborated too much here.

[0032] The experiment is set as: has a low impact on the resolution weight, has a medium impact on timeliness, has a large impact on data integrity.

[0033] The weight factors are set as: the resolution factor , the timeliness factor , and the integrity factor .

[0034] The resolution is lower than 1 meter, , the resolution is from 1 meter to 10 meters, , the resolution is higher than 10 meters, .

[0035] The data delay is less than 30 minutes, , the data delay is from 30 to 60 minutes, , the data delay exceeds 60 minutes, .

[0036] The data missing rate is lower than , the data missing rate to , the data missing rate is higher than .

[0037] As a preferred solution of the present invention, a pollution diffusion path prediction model and a data-driven model are configured based on the range pollution data, precise pollution data, and deep water data after multi-modal data fusion, and pollution prediction results and pollution error data are obtained. Based on the pollution prediction results and pollution error data, hybrid-driven fusion is performed, and the initial source of the pollutant is traced to obtain pollution source tracing data. Specifically:

[0038] Step S300.1: Configure a pollution diffusion path prediction model based on the range pollution data, precise pollution data, and deep water data after multi-modal data fusion. Specifically:

[0039]

[0040] In the formula: is the pollutant concentration, is the fluid velocity field, is the diffusion coefficient, dynamically adjusted to adapt to water quality differences, is the pollutant source term, representing the intensity of newly added pollutants.

[0041] The finite volume method is used to discretize the partial differential equation, and parallel computing is combined to accelerate the solution to obtain the pollution prediction results.

[0042] Step S300.2: Configure a data-driven model based on the range pollution data, precise pollution data, and deep water data after multi-modal data fusion.

[0043] Based on deep learning, a physically enhanced neural network is adopted, and the diffusion conservation law is added during the training process. By combining the physical error and data error through the loss function, pollution error data is obtained.

[0044] Step S300.3: Perform hybrid-driven fusion based on the obtained pollution prediction results and pollution error data. Specifically:

[0045]

[0046] In the formula: is the predicted value of the finally fused pollutant concentration, is the pollutant concentration predicted by the physical-driven model, is the pollutant concentration predicted by the data-driven model and are the weight factors of the physical-driven model and the data-driven model respectively.

[0047] Step S300.4: Based on the pollution prediction results and pollution error data after hybrid-driven fusion, trace the initial source of the pollutant through reverse diffusion simulation. Specifically:

[0048]

[0049] In the formula: is the geographical location of the pollution source, is the distribution location of the current pollutant, is the fluid velocity field, describing the path of pollutant movement with the environmental dynamics is the current time, is the initial time when the pollution occurred.

[0050] Extract the spectral characteristics of the pollution data within the extraction range and the precise pollution data, match them with the known pollutant characteristic library, extract the water sample DNA information in the precise pollution data, compare it with the biological database in the polluted area, identify possible biological pollution sources, and obtain the pollution source tracing data.

[0051] As a preferred solution of the present invention, based on the obtained pollution source tracing data, the pollution prediction results and pollution error data after hybrid drive fusion, three-dimensional volume data is generated, and through time series processing, a three-dimensional visual model is obtained. Based on the obtained pollution source tracing data, the pollution source location is marked through the three-dimensional visual model, specifically:

[0052] Step S400.1: Generate three-dimensional volume data based on the obtained pollution source tracing data, the pollution prediction results and pollution error data after hybrid drive fusion.

[0053] The three-dimensional volume data takes a grid cell of 10m×10m×1m as the basic unit to obtain a three-dimensional grid, and the pollutant concentration value is directly mapped into the three-dimensional grid, and a color gradient is used to represent the concentration range.

[0054] The color gradient includes: blue, yellow, and red.

[0055] Blue represents a low concentration of 0 - 10mg / L, yellow represents a medium concentration of 10–50mg / L, and red represents a high concentration greater than 50mg / L.

[0056] Arrange blue at the bottom, yellow in the middle, and red at the top, generate a frame-by-frame animation of pollutant diffusion through time series, and combine with the fluid velocity field , superimpose vector arrows in the three-dimensional grid, visually display the diffusion direction, and obtain a three-dimensional visual model.

[0057] In the three-dimensional grid, the length of the vector arrow represents the flow velocity magnitude, and the direction indicates the fluid movement direction, which is dynamically superimposed on the three-dimensional visual model to show the trend and direction of pollutant diffusion.

[0058] Step S400.2: Based on the obtained pollution source tracing data, mark the pollution source location through the three-dimensional visual model to obtain the marked pollution source result.

[0059] The annotations include: the location of the pollution source, the intensity of the pollution source, and the type of the pollution source.

[0060] When the user clicks on the pollution source annotation, detailed information is provided, including: spectral characteristics, environmental DNA, and pollution prediction results.

[0061] As a preferred solution of the present invention, based on the pollution prediction results and pollution error data after hybrid drive fusion, preset classification thresholds and risk indices are calculated to obtain risk index data and potential impact area data. Based on the obtained risk index data and potential impact area data, decision presets and decision preset optimizations are carried out, and visualization processing and report content generation are performed, specifically including:

[0062] Step S500.1: Based on the pollution prediction results and pollution error data after hybrid drive fusion, preset classification thresholds are determined.

[0063] The classification thresholds include: low pollution area, medium pollution area, and high pollution area.

[0064] When , it is a low pollution area. When , it is a medium pollution area. When , it is a high pollution area.

[0065] The high pollution area indicates a serious threat to the ecosystem, the medium pollution area needs to be monitored, and the low pollution area is basically harmless.

[0066] Based on the three-dimensional visual model after the pollution source location annotation, the pollution area is divided by the preset classification threshold to generate a pollution distribution zoning map, and the pollutant types are marked and classified in combination with the pollution source traceability data.

[0067] Step S500.2: Based on the pollution prediction results and pollution error data after the preset classification threshold, risk indices are calculated to obtain risk index data, specifically:

[0068]

[0069] In the formula: is the risk index, quantifying the threat degree of the pollutant to the environment. is the pollutant concentration. is the environmental sensitivity factor. is the pollution source intensity.

[0070] Based on the obtained risk index data, priority areas are set.

[0071] The priority areas are set as: when , it is a high priority area. When , it is a medium priority area. When , it is a low priority area, and combined with the pollution distribution zoning map, a priority map is generated to mark the high-risk areas.

[0072] Based on the obtained risk index data and the pollution prediction results and pollution error data after the preset classification threshold, the pollution diffusion path prediction model is used to predict the pollution diffusion path in the next N hours and obtain the potential impact area data.

[0073] Step S500.3: Decision presetting is performed based on the obtained risk index data and potential impact area data.

[0074] In high-priority areas, adsorption equipment will be deployed to treat oil films and garbage salvage equipment will be deployed. In medium-priority areas, chemical neutralization agents will be used to decompose pollutants. In low-priority areas, ecological recovery will be promoted by releasing repair bacteria.

[0075] Assign decision presets to clean-up equipment and monitoring personnel based on priority maps.

[0076] Step S500.4: Use the gradient boosting tree to perform decision preset optimization.

[0077] The gradient boosting tree is trained using the historical risk index data and pollutant concentrations in the database as the training set. The model accuracy is evaluated using the cross-validation method, and the gradient boosting tree hyperparameters are adjusted through grid search.

[0078] Step S500.5: Perform visualization processing based on the obtained risk index data and the decision preset results.

[0079] High priority areas are painted red, medium priority areas are painted yellow, and low priority areas are painted green.

[0080] Dynamically display the impact of different decision preset results on pollutant concentrations and generate report content.

[0081] The report content includes: risk index data, priority map, 3D visual model, pollution source marking results and decision preset results.

[0082] As a preferred solution of the present invention, the latest risk index data is obtained, and the historical risk index data is compared with the latest risk index data to obtain the comparison result, and the current execution decision preset is analyzed to obtain the analysis result. The final report result is generated based on the obtained report content, report results, comparison results and analysis results, which specifically includes:

[0083] Step S600.1, obtain the latest risk index data.

[0084] The risk index data is listed as historical risk index data, and the historical risk index data is compared with the latest risk index data, specifically:

[0085]

[0086]

[0087] In the formula: represents the change in the risk index, then it indicates an increase in risk, then it indicates a decrease in risk, then it indicates that the risk remains stable, represents the change in the pollution concentration, then it indicates an increase in the pollutant concentration and the need to strengthen the treatment measures, then it indicates a decrease in the pollutant concentration and the current treatment is effective.

[0088] According to and the change values, the treatment effects are classified into three categories: high - efficiency, medium - efficiency, and low - efficiency, and a report result is generated. The report result includes: the concentration reduction rate, the control effect of the diffusion range, and the risk reduction amplitude.

[0089] Step S600.2: Analyze the current execution decision preset.

[0090] The analysis includes: resource input and treatment coverage.

[0091] If or when, then increase the resource input or modify the decision preset means. When when, then increase the monitoring points and closely track the diffusion path. When when, then the current strategy is initially effective and maintain the existing intensity.

[0092] Step S600.3: Generate the final report result based on the obtained report content, report result, the comparison result between the historical risk index data and the latest risk index data, and the analysis result of the decision preset.

[0093] It is generated through office and output in docx format. The final report result includes: the comparative analysis result between the historical risk index data and the latest risk index data, the pollution concentration distribution, the pollution source, the analysis result of the decision preset, the change in the risk index, and the pollution prediction result.

[0094] Compared with the prior art, the beneficial effects of the present invention are:

[0095] 1. Through the multi - modal data fusion of satellite remote sensing, unmanned aerial vehicle monitoring, and deep - water sampling, the full - coverage monitoring of pollutants from the surface layer to the deep - water area is realized, overcoming the technical bottleneck that a single data source cannot accurately reflect the three - dimensional pollution characteristics, and greatly improving the comprehensiveness and accuracy of the monitoring system.

[0096] 2. By introducing the multi-modal data fusion technology and using the dynamic time warping algorithm, feature correlation enhancement, and weighted fusion method, multi-source heterogeneous data is effectively integrated, improving the robustness and consistency of the data, providing high-quality input data for the subsequent pollution diffusion prediction model, and thus significantly improving the accuracy of the model.

[0097] 3. Based on the hybrid-driven model, combining the advantages of the physical-driven model and the data-driven model, the present invention can achieve high-precision pollution diffusion prediction and pollution source tracing. At the same time, the model parameters are dynamically adjusted to adapt to real-time monitoring data, ensuring the reliability of the pollution diffusion path prediction and the tracing results.

[0098] 4. Through the preprocessing and fusion of multi-modal data, combining range pollution data, precise pollution data, and deep water data, the pollution diffusion path prediction model and the data-driven model are dynamically adjusted, thereby generating more accurate pollution prediction results and pollution error data. At the same time, through reverse diffusion simulation and pollution source tracing analysis, the efficiency of pollution source location and decision support is significantly improved, providing accurate and real-time decision-making basis for quickly responding to marine pollution incidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0099] Figure 1 It is a flowchart of the method for marine pollution source tracing and monitoring based on satellite remote sensing and unmanned aerial vehicle provided by the embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0100] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0101] Please refer to Figure 1 , Figure 1 It is a flowchart of the method for marine pollution source tracing and monitoring based on satellite remote sensing and unmanned aerial vehicle provided by the embodiment of the present application.

[0102] In this embodiment, the method for marine pollution source tracing and monitoring based on satellite remote sensing and unmanned aerial vehicle may include steps S100, S200, S300, S400, S500, and S600;

[0103] Step S100: Obtain range pollution data, precise pollution data, and deep water data.

[0104] Step S200: Perform preprocessing and multi-modal data fusion based on the obtained range pollution data, precise pollution data, and deep water data.

[0105] Step S300: Configure a pollution diffusion path prediction model and a data-driven model based on the range pollution data, precise pollution data, and deep-water data after multi-modal data fusion, obtain a pollution prediction result and pollution error data, perform hybrid-driven fusion based on the pollution prediction result and pollution error data, and track the initial source of pollutants to obtain pollution source tracing data.

[0106] Step S400: Generate three-dimensional volume data based on the obtained pollution source tracing data, the pollution prediction result, and the pollution error data after hybrid-driven fusion, and obtain a three-dimensional visual model through time series processing. Based on the obtained pollution source tracing data, mark the position of the pollution source through the three-dimensional visual model.

[0107] Step S500: Calculate a preset classification threshold and a risk index based on the pollution prediction result and pollution error data after hybrid-driven fusion to obtain risk index data and potential impact area data. Based on the obtained risk index data and potential impact area data, perform decision preset and decision preset optimization, and perform visualization processing and generate report content.

[0108] Step S600: Obtain the latest risk index data, compare the historical risk index data with the latest risk index data to obtain a comparison result, analyze the current execution decision preset to obtain an analysis result, and generate a final report result based on the obtained report content, report result, comparison result, and analysis result.

[0109] In the specific implementation manner, obtaining the range pollution data, precise pollution data, and deep-water data specifically includes:

[0110] Step S100.1: Obtain the range pollution data, precise pollution data, and deep-water data.

[0111] Conduct large-scale coverage monitoring of ocean surface pollutants through satellite remote sensing to obtain range pollution data, which includes: pollutant distribution range, pollutant concentration, pollutant type, potential pollution source location, ocean current speed and direction, surface water temperature, and surface salinity.

[0112] The range pollution data initially identifies the distribution and concentration of pollutants through satellite remote sensing high-resolution multi-spectral images, uses hyperspectral to extract the spectral characteristics of pollutants for substance classification, and uses thermal infrared to capture temperature anomaly areas for locating pollution sources.

[0113] High-precision monitoring is carried out on hotspots and areas difficult to cover by satellites through drones to obtain accurate pollution data, which includes: refined pollutant distribution, pollutant concentration, pollutant characteristics, specific pollution source types, biological pollution sources, local water temperature, local water flow velocity, and local water transparency.

[0114] The accurate pollution data is obtained by the drone carrying a spectral sensor to identify the optical characteristics of different pollutants, using a high-resolution camera to obtain a clear pollutant distribution map, and using an environmental DNA sampling device to collect water samples for biological traceability analysis.

[0115] Dissolved organic matter, heavy metal content, and pH value of columnar water bodies are monitored and collected through buoys to obtain deep water data, which includes: deep layer pollutant concentration, pollutant vertical distribution, deep water pollution source type, pollution source intensity, deep layer water temperature, deep layer salinity, deep layer water flow velocity and direction, and water body density.

[0116] The deep water data is obtained by an underwater robot moving in the target area to collect deep water samples of pollutants.

[0117] In the specific implementation manner, preprocessing and multi-modal data fusion are performed based on the obtained range pollution data, accurate pollution data, and deep water data. Specifically:

[0118] Step S200.1: Perform preprocessing based on the obtained range pollution data, accurate pollution data, and deep water data.

[0119] The preprocessing includes: image denoising, numerical data cleaning, and spectral data standardization.

[0120] Image denoising uses a convolutional neural network to denoise the range pollution data, identify and repair noise areas, and perform geometric distortion correction on the accurate pollution data. Numerical data cleaning uses an RNN-based time series anomaly detection model to identify and eliminate sudden error data points. Spectral data standardization uses normalization to uniformly adjust the spectral data collected by multiple sensors.

[0121] It should be noted here that since preprocessing is a well-known technology in this field, it will not be elaborated too much here.

[0122] Step S200.2: Perform multi-modal data fusion based on the preprocessed range pollution data, accurate pollution data, and deep water data.

[0123] Multi-modal data fusion includes: spatio-temporal alignment, feature correlation enhancement, and weighted fusion.

[0124] Spatio-temporal alignment is based on the dynamic time warping algorithm to perform time series alignment on multi-source data, uses the geospatial interpolation algorithm to fill in the data in the uncovered area, strengthens feature correlation, maps spectral data, thermal infrared data, and deep water physical parameters to a unified feature space through multi-modal feature maps, and uses graph neural networks to establish a correlation model between data to improve the accuracy and robustness of data fusion. Weighted fusion dynamically assigns weights to the range pollution data, precise pollution data, and deep water data quality after spatio-temporal alignment and feature correlation strengthening. Specifically:

[0125]

[0126] Where: is the initial weight of the data, is the resolution factor, adjusted according to the relative superiority or inferiority of the resolution, is the timeliness factor, adjusted according to the impact of data delay, is the data integrity factor, adjusted according to the missing rate, is the weight impact factor corresponding to resolution, timeliness, and integrity.

[0127] It should be noted here that since multi-modal data fusion is a well-known technology in this field, it will not be elaborated too much here.

[0128] The experiment is set as: is for low impact of resolution weight, is for medium impact of timeliness, is for high impact of data integrity.

[0129] The weight factors are set as: resolution factor 、timeliness factor and integrity factor .

[0130] The resolution is lower than 1 meter, , the resolution is from 1 meter to 10 meters, , the resolution is higher than 10 meters, .

[0131] The data delay is less than 30 minutes, , the data delay is from 30 to 60 minutes, , the data delay exceeds 60 minutes, .

[0132] The data missing rate is lower than , the data missing rate to , the data missing rate is higher than .

[0133] If the satellite data has a resolution of 30 meters (low resolution), a delay of 30 minutes, and a completeness of 100%, the drone data has a resolution of 10 meters (medium resolution), a delay of 15 minutes, and a completeness of 95%, and the deep - water collected data has no resolution, a delay of 1 hour, and a completeness of 92%, then the calculated weights are as follows:

[0134] Satellite data weight :

[0135]

[0136] Drone data weight :

[0137]

[0138] Deep - water collected data weight :

[0139]

[0140] The final fused eigenvalue is:

[0141]

[0142] In the specific implementation manner, based on the range pollution data, precise pollution data, and deep - water data after multi - modal data fusion, a pollution diffusion path prediction model and a data - driven model are configured, and pollution prediction results and pollution error data are obtained. Based on the pollution prediction results and pollution error data, hybrid - drive fusion is performed, and the initial source of the pollutant is traced to obtain pollution source tracing data. Specifically:

[0143] Step S300.1: Configure a pollution diffusion path prediction model based on the range pollution data, precise pollution data, and deep - water data after multi - modal data fusion. Specifically:

[0144]

[0145] In the formula: is the pollutant concentration, is the fluid velocity field, is the diffusion coefficient, dynamically adjusted to adapt to water quality differences, is the pollutant source term, representing the intensity of newly added pollutants.

[0146] The partial differential equation is discretized using the finite - volume method, and parallel computing is combined to accelerate the solution to obtain the pollution prediction results.

[0147] Step S300.2: Configure a data - driven model based on the range pollution data, precise pollution data, and deep - water data after multi - modal data fusion.

[0148] Based on deep learning, a physically enhanced neural network is adopted. During the training process, the diffusion conservation law is introduced. By combining the physical error and the data error through the loss function, the contaminated error data is obtained, specifically as follows:

[0149]

[0150] In the formula: is used to optimize the data-driven model, is the residual of the diffusion equation, is the error between the contaminated prediction result and the observation, and are the weight factors of the physical error and the data error respectively.

[0151] Step S300.3: Perform hybrid-driven fusion based on the obtained contaminated prediction result and the contaminated error data, specifically as follows:

[0152]

[0153] In the formula: is the predicted value of the final fused pollutant concentration, is the pollutant concentration predicted by the physical-driven model, is the pollutant concentration predicted by the data-driven model and are the weight factors of the physical-driven model and the data-driven model respectively.

[0154] Step S300.4: Based on the contaminated prediction result and the contaminated error data after hybrid-driven fusion, through reverse diffusion simulation, trace the initial source of the pollutant, specifically as follows:

[0155]

[0156] In the formula: is the geographical location of the pollution source, is the distribution location of the current pollutant, is the fluid velocity field, describing the path of the pollutant moving with the environmental dynamics is the current time, is the initial time when the pollution occurred.

[0157] Extract the spectral features of the range pollution data and the precise pollution data, match them with the known pollutant feature library, extract the water sample DNA information in the precise pollution data, compare it with the biological database in the polluted area, identify the possible biological pollution sources, and obtain the pollution source tracing data.

[0158] In the specific implementation manner, three-dimensional volume data is generated based on the obtained pollution source tracing data, the pollution prediction results and pollution error data after hybrid drive fusion, and through time series processing, a three-dimensional visual model is obtained. Based on the obtained pollution source tracing data, the positions of pollution sources are marked through the three-dimensional visual model, specifically as follows:

[0159] Step S400.1: Generate three-dimensional volume data based on the obtained pollution source tracing data, the pollution prediction results and pollution error data after hybrid drive fusion.

[0160] The three-dimensional volume data takes a grid cell of 10m×10m×1m as the basic unit to obtain a three-dimensional grid. The pollutant concentration values are directly mapped into the three-dimensional grid, and color gradients are used to represent the concentration ranges.

[0161] The color gradients include: blue, yellow, and red.

[0162] Blue represents a low concentration of 0 - 10mg / L, yellow represents a medium concentration of 10–50mg / L, and red represents a high concentration greater than 50mg / L.

[0163] Arrange blue at the bottom, yellow in the middle, and red at the top, and generate a frame-by-frame animation of pollutant diffusion through time series. Combine with the fluid velocity field , and superimpose vector arrows in the three-dimensional grid to visually display the diffusion direction, and obtain a three-dimensional visual model.

[0164] In the three-dimensional grid, the length of the vector arrow represents the flow velocity magnitude, and the direction indicates the fluid movement direction, which is dynamically superimposed on the three-dimensional visual model to show the trend and direction of pollutant diffusion.

[0165] Step S400.2: Based on the obtained pollution source tracing data, mark the positions of pollution sources through the three-dimensional visual model to obtain the marked pollution source results.

[0166] The marking includes: the position of the pollution source, the intensity of the pollution source, and the type of the pollution source.

[0167] When the user clicks on the pollution source mark, detailed information is provided, and the detailed information includes: spectral characteristics, environmental DNA, and pollution prediction results.

[0168] In the specific implementation manner, preset classification thresholds and risk indices are calculated based on the pollution prediction results and pollution error data after hybrid drive fusion to obtain risk index data and potential impact area data. Based on the obtained risk index data and potential impact area data, decision presets and decision preset optimizations are carried out, and visualization processing and report content generation are carried out, specifically including:

[0169] Step S500.1: Set preset classification thresholds based on the pollution prediction results and pollution error data after hybrid drive fusion.

[0170] The classification thresholds include: low pollution area, medium pollution area, and high pollution area.

[0171] When , it is a low pollution area. When , it is a medium pollution area. When , it is a high pollution area.

[0172] The high pollution area indicates a serious threat to the ecosystem. The medium pollution area requires monitoring, and the low pollution area is basically harmless.

[0173] Based on the three-dimensional visual model after marking the pollution source location, divide the pollution area through the preset classification thresholds, generate a pollution distribution zoning map, and combine with the pollution source tracing data to mark and classify the pollutant types.

[0174] Step S500.2: Calculate the risk index based on the pollution prediction results and pollution error data after the preset classification thresholds to obtain risk index data. Specifically:

[0175]

[0176] In the formula: is the risk index, which quantifies the threat degree of pollutants to the environment. is the pollutant concentration. is the environmental sensitivity factor. is the pollution source intensity.

[0177] Set the priority areas based on the obtained risk index data.

[0178] The priority areas are set as: when , it is a high priority area. When , it is a medium priority area. When , it is a low priority area. Combine with the pollution distribution zoning map to generate a priority map and mark the high-risk areas.

[0179] Based on the obtained risk index data and the pollution prediction results and pollution error data after the preset classification thresholds, use the pollution diffusion path prediction model to predict the pollution diffusion path within the next N hours to obtain the potential impact area data.

[0180] Step S500.3: Make a preset decision based on the obtained risk index data and potential impact area data.

[0181] In high-priority areas, adsorption equipment will be deployed to treat oil films and garbage salvage equipment will be deployed. In medium-priority areas, chemical neutralization agents will be used to decompose pollutants. In low-priority areas, ecological recovery will be promoted by releasing repair bacteria.

[0182] Assign decision presets to clean-up equipment and monitoring personnel based on priority maps.

[0183] Step S500.4: Use the gradient boosting tree to perform decision preset optimization.

[0184] The gradient boosting tree is trained using the historical risk index data and pollutant concentrations in the database as the training set. The model accuracy is evaluated using the cross-validation method, and the gradient boosting tree hyperparameters are adjusted through grid search.

[0185] Step S500.5: Perform visualization processing based on the obtained risk index data and the decision preset results.

[0186] High priority areas are painted red, medium priority areas are painted yellow, and low priority areas are painted green.

[0187] Dynamically display the impact of different decision preset results on pollutant concentrations and generate report content.

[0188] The report content includes: risk index data, priority map, 3D visual model, pollution source marking results and decision preset results.

[0189] In a specific implementation method, the latest risk index data is obtained, and the historical risk index data is compared with the latest risk index data to obtain a comparison result, and the current execution decision preset is analyzed to obtain an analysis result. The final report result is generated based on the obtained report content, report results, comparison results and analysis results, which specifically includes:

[0190] Step S600.1, obtain the latest risk index data.

[0191] The risk index data is listed as historical risk index data, and the historical risk index data is compared with the latest risk index data, specifically:

[0192]

[0193]

[0194] Where: Indicates the change in risk index, , it means the risk is increased, , it means the risk is reduced. , it means the risk remains stable. Indicates the change in pollution concentration, , it indicates that the pollutant concentration increases and the treatment measures need to be strengthened. , it indicates that the pollutant concentration decreases and the current treatment is effective.

[0195] According to and 's change values, the treatment effects are classified into three categories: high - efficiency, medium - efficiency, and low - efficiency, and a report result is generated. The report result includes: the concentration reduction rate, the control effect of the diffusion range, and the risk reduction amplitude.

[0196] Step S600.2: Analyze the current execution decision preset.

[0197] The analysis includes: resource input and treatment coverage.

[0198] If or , then increase the resource input or modify the decision preset means. When , then increase the monitoring points and closely track the diffusion path. When , then the current strategy begins to show results and maintain the existing intensity.

[0199] Step S600.3: Generate the final report result based on the obtained report content, report result, the comparison result of historical risk index data and the latest risk index data, and the analysis result of the decision preset.

[0200] It is generated through office and output in docx format. The final report result includes: the comparative analysis result of historical risk index data and the latest risk index data, the pollution concentration distribution, the pollution source, the analysis result of the decision preset, the risk index change, and the pollution prediction result.

[0201] In the above content, in practical applications, first, obtain the range pollution data, precise pollution data, and deep-water data. Through satellite remote sensing, a large-scale coverage monitoring of marine surface pollutants is carried out to obtain the range pollution data, including the distribution range of pollutants, pollutant concentration, pollutant type, location of potential pollution sources, ocean current speed and direction, surface water temperature, and surface salinity. The distribution and concentration of pollutants are initially identified through high-resolution multispectral images of satellite remote sensing. The spectral characteristics of pollutants are extracted using hyperspectral technology to complete substance classification, and the pollution source is located by capturing temperature anomaly areas through thermal infrared. Subsequently, high-precision monitoring is carried out on hot spots and areas difficult to cover by satellites using drones to obtain precise pollution data, including refined information on pollutant distribution, concentration, characteristics, specific pollution source types, biological pollution sources, local water temperature, local water flow speed, and local water transparency. Finally, data on dissolved organic matter, heavy metal content, and pH value of columnar water bodies are collected through buoy monitoring, and deep-water samples are collected in combination with underwater robots to obtain deep-water data, including deep pollutant concentration, vertical distribution of pollutants, pollution source type and intensity, as well as deep water temperature, salinity, water flow speed and direction, and water density.

[0202] Then, preprocess and perform multimodal data fusion on the obtained range pollution data, precise pollution data, and deep-water data. The preprocessing includes image denoising, numerical data cleaning, and spectral data standardization. Image data is denoised and the noise area is repaired through a convolutional neural network. Geometric distortion correction is performed on the precise pollution data. Numerical data cleaning uses a method based on a time-series anomaly detection model to eliminate sudden error data points. Spectral data is standardized through normalization technology to eliminate differences in data collected by different sensors. Multimodal data fusion includes spatio-temporal alignment, feature association enhancement, and weighted fusion. The time series of multi-source data is aligned through the dynamic time warping algorithm, and data in uncovered areas is filled in by combining geospatial interpolation methods. Furthermore, the relevance of multimodal features is enhanced through a graph neural network, and dynamic weighted fusion is performed according to the resolution, timeliness, and integrity of the data to generate high-quality fusion data.

[0203] Next, configure a pollution diffusion path prediction model and a data-driven model based on the multimodal data fusion results to generate pollution prediction results and pollution error data. The pollution diffusion path prediction model is based on the diffusion equation:

[0204]

[0205] where represents the pollutant concentration, is the fluid velocity field, is the diffusion coefficient, is the pollution source intensity. By discretely solving partial differential equations through the finite volume method and accelerating with parallel computing, the pollution diffusion prediction results are obtained. The data-driven model combines a physics-enhanced neural network with the law of conservation of pollution diffusion. During the training process, physical errors and data errors are combined in the loss function for optimization to generate pollution error data.

[0206] Then, based on the pollution prediction results and pollution error data, a hybrid-driven fusion prediction of pollutant concentration is carried out. Combining the prediction results of the physical-driven model and the data-driven model, through the weighted formula:

[0207]

[0208] where, and are dynamically adjusted weight factors. Combining the advantages of the physical model and the data model, the final fused pollutant concentration prediction value is generated.

[0209] Next, source tracing of the pollution source is carried out through reverse diffusion simulation and feature analysis. According to the prediction results of the pollution diffusion path, the source location of the pollutant is traced back:

[0210]

[0211] where, is the pollution source location, is the current pollutant distribution location, is the fluid velocity field. On this basis, combining the spectral characteristics of the pollutant and environmental DNA analysis, matching the pollution feature library with the regional biological database, the type and specific source of the pollution source are further verified.

[0212] Finally, based on the results of pollution source tracing and pollution diffusion prediction, optimization suggestions for pollution control are generated. According to the division of high, medium, and low-risk areas, the allocation of control resources is adjusted, and appropriate control measures are recommended in combination with the type of pollutant, including: mechanical cleaning, chemical neutralization, or ecological restoration. The prediction model is dynamically updated, the decision-making strategy is optimized, and the pollution distribution, diffusion path, and control effect are displayed through a three-dimensional visualization platform, providing comprehensive support for subsequent pollution monitoring and control.

[0213] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for tracing and monitoring marine pollution based on satellite remote sensing and unmanned aerial vehicles, characterized in that, It includes the following steps: S100. Obtain range pollution data, precise pollution data, and deep water data; S200. Perform preprocessing and multi-modal data fusion based on the obtained range pollution data, precise pollution data, and deep water data; S300. Configure a pollution diffusion path prediction model and a data-driven model based on the range pollution data, precise pollution data, and deep water data after multi-modal data fusion, obtain pollution prediction results and pollution error data, perform hybrid-driven fusion based on the pollution prediction results and pollution error data, and trace the initial source of pollutants to obtain pollution source tracing data; S400. Generate three-dimensional volume data based on the obtained pollution source tracing data and the pollution prediction results and pollution error data after hybrid-driven fusion, and obtain a three-dimensional visual model through time series processing. Mark the position of the pollution source based on the obtained pollution source tracing data through the three-dimensional visual model; S500. Calculate preset classification thresholds and risk indices based on the pollution prediction results and pollution error data after hybrid-driven fusion to obtain risk index data and potential impact area data. Perform decision preset and decision preset optimization based on the obtained risk index data and potential impact area data, and perform visualization processing and generate report content; S600. Obtain the latest risk index data, compare the historical risk index data with the latest risk index data to obtain a comparison result, analyze the current execution decision preset to obtain an analysis result, and generate a final report result based on the obtained report content, report result, comparison result, and analysis result; Obtain range pollution data, precise pollution data, and deep water data, specifically including: S100.

1. Obtain range pollution data, precise pollution data, and deep water data; Conduct large-scale coverage monitoring of marine surface pollutants through satellite remote sensing to obtain range pollution data, which includes: pollutant distribution range, pollutant concentration, pollutant type, potential pollution source location, ocean current speed and direction, surface water temperature, and surface salinity; Conduct high-precision monitoring of hotspots and areas difficult to cover by satellites through unmanned aerial vehicles to obtain precise pollution data, which includes: refined pollutant distribution, pollutant concentration, pollutant characteristics, specific pollution source type, biological pollution source, local water temperature, local water flow speed, and local water transparency; Collect dissolved organic matter, heavy metal content, and pH value of columnar water through buoy monitoring to obtain deep water data, which includes: deep pollutant concentration, pollutant vertical distribution, deep water pollution source type, pollution source intensity, deep water temperature, deep water salinity, deep water flow speed and direction, and water density; Perform preprocessing and multi-modal data fusion based on the obtained range pollution data, precise pollution data, and deep water data, specifically: S200.

1. Perform preprocessing based on the obtained range pollution data, precise pollution data, and deep water data; The preprocessing includes: image denoising, numerical data cleaning, and spectral data standardization; Image denoising uses a convolutional neural network to denoise range contaminated data, identify and repair noise areas, and perform geometric distortion correction on precisely contaminated data. Numerical data cleaning uses an RNN-based time series anomaly detection model to identify and remove burst error data points. Spectral data normalization uses normalization to uniformly adjust spectral data collected by multiple sensors; S200.

2. Perform multi-modal data fusion based on the pre-processed range contaminated data, precisely contaminated data, and deep water data; Multi-modal data fusion includes: spatio-temporal alignment, feature correlation enhancement, and weighted fusion; Spatio-temporal alignment is based on the dynamic time warping algorithm to align the time series of multi-source data, and uses the geospatial interpolation algorithm to fill in the data in the uncovered areas. Feature correlation enhancement maps spectral data, thermal infrared data, and deep water physical parameters to a unified feature space through multi-modal feature maps, and uses a graph neural network to establish a correlation model between data to improve the accuracy and robustness of data fusion. Weighted fusion dynamically assigns weights based on the quality of the range contaminated data, precisely contaminated data, and deep water data after spatio-temporal alignment and feature correlation enhancement; Configure a pollution diffusion path prediction model and a data-driven model based on the range contaminated data, precisely contaminated data, and deep water data after multi-modal data fusion to obtain pollution prediction results and pollution error data. Perform hybrid-driven fusion based on the pollution prediction results and pollution error data, and trace the initial source of the pollutant to obtain source tracing data of the pollution source. Specifically: S300.

1. Configure a pollution diffusion path prediction model based on the range contaminated data, precisely contaminated data, and deep water data after multi-modal data fusion; Use the finite volume method to discretize the partial differential equation, and combine parallel computing to accelerate the solution to obtain the pollution prediction result; S300.

2. Configure a data-driven model based on the range contaminated data, precisely contaminated data, and deep water data after multi-modal data fusion; Based on deep learning, use a physically enhanced neural network, add the diffusion conservation law during the training process, and combine physical error and data error through the loss function to obtain pollution error data; S300.

3. Perform hybrid-driven fusion based on the obtained pollution prediction results and pollution error data. Specifically: In the formula: is the predicted value of the finally fused pollutant concentration, is the pollutant concentration predicted by the physically driven model, is the pollutant concentration predicted by the data-driven model and are the weight factors of the physically driven model and the data-driven model respectively; S300.

4. Based on the pollution prediction results and pollution error data after hybrid-driven fusion, trace the initial source of the pollutant through reverse diffusion simulation; Extract the spectral features of the range contaminated data and precisely contaminated data, match them with the known pollutant feature library, extract the water sample DNA information in the precisely contaminated data, compare it with the biological database in the polluted area, identify possible biological pollution sources, and obtain source tracing data of the pollution source; Generate three-dimensional volume data based on the obtained source tracing data of the pollution source, the pollution prediction results, and pollution error data after hybrid-driven fusion, and obtain a three-dimensional visual model through time series processing. Based on the obtained source tracing data of the pollution source, mark the location of the pollution source through the three-dimensional visual model. Specifically: S400.

1. Generate three-dimensional volume data based on the obtained source tracing data of the pollution source, the pollution prediction results, and pollution error data after hybrid-driven fusion; The three-dimensional volume data uses a grid cell of 10m×10m×1m as the basic unit to obtain a three-dimensional grid. The pollutant concentration values are directly mapped into the three-dimensional grid, and a color gradient is used to represent the concentration range. Color gradients include: blue, yellow, and red; Place the blue arrangement below, the yellow arrangement in the middle, and the red arrangement above. Generate frame-by-frame animations of pollutant diffusion through time series, and combine with the fluid velocity field , overlay vector arrows in the three-dimensional grid to visually display the diffusion direction, and obtain a three-dimensional visual model; S400.

2. Based on the obtained pollution source tracing data, the pollution source locations are marked through a three-dimensional visual model to obtain the pollution source marking results; The annotations include: location of pollution source, intensity of pollution source and type of pollution source; When the user clicks on the pollution source label, detailed information is provided, including: spectral characteristics, environmental DNA and pollution prediction results; Based on the pollution prediction results and pollution error data after hybrid drive fusion, preset classification thresholds and risk index calculations are performed to obtain risk index data and potential impact area data. Based on the obtained risk index data and potential impact area data, decision presets and decision preset optimization are performed, and visualization processing and report content are generated, including: S500.

1. Preset classification thresholds based on the pollution prediction results and pollution error data after hybrid drive fusion; Classification thresholds include: low pollution area, medium pollution area and high pollution area; Highly polluted areas represent a serious threat to the ecosystem, medium polluted areas require monitoring, and low polluted areas are basically harmless; Based on the three-dimensional visual model with the pollution source location marked, the pollution area is divided by preset classification thresholds to generate a pollution distribution zoning map, and the pollutant types are marked and classified in combination with the pollution source tracing data; S500.

2. Calculate the risk index based on the pollution prediction results after the preset classification threshold and the pollution error data to obtain the risk index data, specifically: In the formula: is the risk index, quantifying the threat degree of pollutants to the environment, is the pollutant concentration, is the environmental sensitivity factor, is the pollution source intensity; set the priority area based on the obtained risk index data; Based on the obtained risk index data and the pollution prediction results and pollution error data after the preset classification threshold, the pollution diffusion path prediction model is used to predict the pollution diffusion path in the next N hours and obtain the potential impact area data; S500.

3. Make decision presets based on the obtained risk index data and potential impact area data; In high-priority areas, adsorption equipment will be deployed to treat oil film and garbage salvage equipment. In medium-priority areas, chemical neutralization agents will be used to decompose pollutants. In low-priority areas, restoration bacteria will be deployed to promote ecological recovery. Assign decision presets to cleanup equipment and monitoring personnel based on priority maps; S500.4, using gradient boosting tree to optimize decision presets; The gradient boosting tree is trained using the historical risk index data and pollutant concentrations in the database as the training set, the model accuracy is evaluated using the cross-validation method, and the hyperparameters of the gradient boosting tree are adjusted through grid search; S500.

5. Perform visualization based on the obtained risk index data and decision preset results; Draw high priority areas in red, medium priority areas in yellow, and low priority areas in green; Dynamically display the impact of different decision-making preset results on pollutant concentrations and generate report content; The report content includes: risk index data, priority map, 3D visual model, pollution source marking results and decision preset results.

2. The method for tracing and monitoring marine pollution based on satellite remote sensing and unmanned aerial vehicles according to claim 1, wherein, Obtain the latest risk index data, compare the historical risk index data with the latest risk index data, obtain the comparison results, analyze the current execution decision presets, obtain the analysis results, and generate the final report results based on the obtained report content, report results, comparison results and analysis results, including: S600.

1. Obtain the latest risk index data; List the risk index data as historical risk index data, and compare the historical risk index data with the latest risk index data. Specifically: Wherein: represents the change in the risk index, if it is [value], it indicates an increase in risk, if it is [value], it indicates a decrease in risk, if it is [value], it indicates that the risk remains stable, represents the change in the pollution concentration, if it is [value], it indicates an increase in the pollutant concentration and the treatment measures need to be strengthened, if it is [value], it indicates a decrease in the pollutant concentration and the current treatment is effective; According to and Based on the change values, the treatment effects are classified into three categories: high efficiency, medium efficiency, and low efficiency, and a report result is generated. The report result includes: the concentration reduction rate, the control effect of the diffusion range, and the risk reduction amplitude; S600.

2. Analyze the current execution decision preset; The analysis includes: resource investment and governance coverage; If or occurs, increase resource investment or modify the decision-making preset means. When occurs, increase the monitoring points and closely track the diffusion path. When occurs, the current strategy begins to show results, maintain the existing intensity; S600.

3. Generate the final report result based on the obtained report content, report result, the comparison result between the historical risk index data and the latest risk index data, and the analysis result of the decision preset.

Images