Marine oil spill pollution diffusion early warning and prediction system and method

Abstract

The invention discloses a marine oil spill pollution diffusion early warning prediction system and method, and relates to the technical field of marine environment prediction, the system comprises a data acquisition and preprocessing unit, a model construction and path simulation unit, a path prediction and anomaly detection unit, a model parameter optimization unit and a visual display unit; and the model construction and path simulation unit is used for constructing an offshore oil spill transportation model based on an ocean numerical model and simulating an oil spill diffusion path in combination with the preprocessed multi-source data. According to the invention, the multi-source data in the marine environment is collected and preprocessed, so that the monitoring range of the marine oil spill event is widened, and the comprehensiveness and timeliness of the data are ensured; meanwhile, by constructing a new characteristic variable, the precision of model simulation is enhanced, and the prediction result is more in line with the actual situation, so that the accuracy and reliability of prediction are greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of marine environment prediction, and in particular to a marine oil spill pollution diffusion early warning prediction system and method. Background Art

[0002] Marine oil spill pollution refers to the leakage of crude oil or refined oil into the sea, beaches or coastal areas due to accidents or operational errors during the exploration, extraction, refining, transportation, loading and unloading and use of oil, which causes serious damage to the marine environment and its ecosystem. This pollution has become a serious global environmental problem, which not only threatens the health of the marine ecosystem, but also has a profound impact on fishery resources, tourism and economic development of coastal areas. With the growth of the global shipping industry and the increase in offshore oil extraction activities, the risk of marine oil spill accidents is also increasing. These accidents will not only cause large amounts of oil to leak into the marine environment, but also have a lasting negative impact on the marine ecosystem through processes such as drift, diffusion and sedimentation.

[0003] At present, the early warning and prediction of marine oil spill pollution mainly rely on satellite remote sensing monitoring, aircraft cruise reconnaissance, ship field inspections and buoy fixed-point monitoring. Although these methods can provide information about oil spills to a certain extent, they generally have problems such as limited monitoring range, low data update frequency and monitoring accuracy that needs to be improved. In addition, the existing mathematical models often fail to fully consider the complex interactions between multiple environmental factors such as wind fields, wave fields, tides and circulations in the process of simulating the drift and diffusion of oil spills, which directly affects the accuracy and reliability of the prediction results.

[0004] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0005] In view of the problems in the related technology, the present invention proposes a marine oil spill pollution spread early warning prediction system and method to overcome the above-mentioned technical problems existing in the existing related technology.

[0006] To this end, the specific technical solution adopted by the present invention is as follows:

[0007] According to one aspect of the present invention, there is provided a marine oil spill pollution diffusion early warning prediction system, the system comprising: a data acquisition and preprocessing unit, a model building and path simulation unit, a path prediction and anomaly detection unit, a model parameter optimization unit and a visualization display unit;

[0008] A data acquisition and preprocessing unit is used to collect multi-source data in the marine environment and preprocess the collected multi-source data;

[0009] The model building and path simulation unit is used to build an offshore oil spill transport model based on the ocean numerical model, and simulate the path of oil spill diffusion in combination with pre-processed multi-source data;

[0010] The path prediction and anomaly detection unit is used to predict the oil spill diffusion path based on the simulation results of the oil spill diffusion path and historical multi-source data using a causal convolutional neural network model, and detect outliers in the multi-source data in real time;

[0011] A model parameter optimization unit is used to feed back the detected abnormal values ​​to the causal convolutional neural network model, and optimize the oil spill diffusion path prediction result by adjusting the causal convolutional neural network model parameters;

[0012] The visualization unit is used to visualize the optimized oil spill diffusion path prediction results using graph neural network technology.

[0013] In one embodiment, the data acquisition and preprocessing unit includes: a data acquisition module, a data cleaning module and a feature engineering module;

[0014] A data acquisition module for collecting multi-source data related to oil spill spread in the marine environment from several data sources;

[0015] The data cleaning module is used to clean the collected multi-source data to remove duplicate values ​​and process missing values;

[0016] The feature engineering module is used to reduce the dimensionality of the multi-source data after data cleaning using a nonlinear dimensionality reduction algorithm, and to establish a multi-source data set based on the physical mechanism of oil spill diffusion.

[0017] In one embodiment, the multi-source data includes: meteorological data, ocean dynamics data, ship trajectories, historical oil spill case data, and physical and chemical properties of the spilled oil.

[0018] In one embodiment, the feature engineering module includes: a parameter setting submodule, a data dimension reduction submodule, a feature calculation submodule and a data set establishment submodule;

[0019] The parameter setting submodule is used to set the relevant parameters of the nonlinear dimensionality reduction algorithm based on the multi-source data after data cleaning;

[0020] The data dimension reduction submodule is used to use a nonlinear dimension reduction algorithm to perform dimension reduction processing on the multi-source data after data cleaning, output the multi-source data after dimension reduction, and establish an initial data set;

[0021] The feature calculation submodule is used to analyze and calculate the kinetic potential energy and pollutant drift index of the oil spill point according to the physical mechanism of oil spill diffusion;

[0022] The dataset establishment submodule is used to construct new feature variables based on the calculation results, and add the new feature variables to the initial dataset to establish a multi-source dataset.

[0023] In one embodiment, the model building and path simulation unit includes: a parameter setting module, a model building module and a diffusion path simulation module;

[0024] The parameter setting module is used to set the ocean area grid, discretize the control equations, and configure the boundary conditions to simulate the dynamic process of the ocean area based on the ocean numerical model;

[0025] Model building module, which is used to build an offshore oil spill transport model by combining the physical and chemical characteristics of the oil spill;

[0026] The diffusion path simulation module is used to simulate the motion trajectory of oil spill particles based on the constructed offshore oil spill transport model and adopt the Lagrangian particle tracking algorithm, and generate the oil spill diffusion path by combining the pre-processed multi-source data.

[0027] In one embodiment, the model building module includes: a characteristic analysis submodule, a horizontal transport model building submodule, an initial model building submodule, and a transport model building submodule;

[0028] The characteristic analysis submodule is used to analyze the pre-processed multi-source data to obtain the physical and chemical characteristics of the oil spill during transportation;

[0029] The horizontal transport model construction submodule is used to establish an oil spill horizontal transport model based on the two-dimensional diffusion equation and combined with the physical and chemical characteristics of the oil spill during the transport process;

[0030] The initial model building submodule is used to combine the oil spill horizontal transport model with the three-dimensional diffusion equation to establish the initial offshore oil spill transport model;

[0031] The transport model construction submodule is used to perform spatial discretization processing on the initial offshore oil spill transport model using the finite difference method, and to construct the offshore oil spill transport model based on the spatial discretization processing results.

[0032] In one embodiment, the two-dimensional diffusion equation is formulated as:

[0033]

[0034] In the formula, represents partial differential; C represents oil film concentration; t represents time; λ represents the component of flow velocity in the x direction; x represents the spatial coordinate axis along the east-west direction; v represents the component of flow velocity in the y direction; y represents the spatial coordinate axis along the north-south direction; K represents the diffusion coefficient; represents the Laplace operator.

[0035] In one embodiment, the three-dimensional diffusion equation is formulated as:

[0036]

[0037] In the formula, represents partial differential; C represents oil film concentration; t represents time; represents the divergence operator; γ represents the three-dimensional velocity vector; N represents the diffusion tensor; S represents the source term.

[0038] In one embodiment, the path prediction and anomaly detection unit includes: a training set building module, a model building and training module, a path prediction module and an outlier detection module;

[0039] The training set establishment module is used to collect and preprocess historical multi-source data and establish a training set in combination with the simulation results of the oil spill diffusion path;

[0040] The model building and training module is used to build a causal convolutional neural network model based on the convolutional neural network algorithm, and use the training set to train the built causal convolutional neural network model;

[0041] The path prediction module is used to collect real-time multi-source data and input the real-time multi-source data into the trained causal convolutional neural network model to predict the oil spill diffusion path;

[0042] The outlier detection module is used to detect outliers on the input real-time multi-source data stream in combination with the isolation forest algorithm.

[0043] According to another aspect of the present invention, there is also provided a method for early warning and prediction of oil spill pollution spread at sea, the method comprising:

[0044] S1. Collect multi-source data in the marine environment and pre-process the collected multi-source data;

[0045] S2. Based on the ocean numerical model, a marine oil spill transport model is constructed, and the oil spill diffusion path is simulated by combining the pre-processed multi-source data;

[0046] S3, based on the simulation results of the oil spill diffusion path and historical multi-source data, the causal convolutional neural network model is used to predict the oil spill diffusion path and detect outliers in the multi-source data in real time;

[0047] S4, feeding back the detected abnormal values ​​to the causal convolutional neural network model, and optimizing the oil spill diffusion path prediction results by adjusting the causal convolutional neural network model parameters;

[0048] S5. Use graph neural network technology to visualize the optimized oil spill diffusion path prediction results.

[0049] The beneficial effects of the present invention are:

[0050] 1. The present invention broadens the monitoring scope of marine oil spill incidents by collecting and preprocessing multi-source data in the marine environment, ensuring the comprehensiveness and timeliness of the data; at the same time, by constructing new characteristic variables, it not only enhances the accuracy of model simulation, but also makes the prediction results more in line with the actual situation, thereby greatly improving the accuracy and reliability of the prediction.

[0051] 2. By utilizing the causal convolutional neural network model, the present invention can quickly and accurately predict the oil spill diffusion path, thereby making full use of historical data and simulation results, capturing the dynamic characteristics of oil spill diffusion, and further realizing accurate prediction of the oil spill diffusion trend.

[0052] 3. The present invention introduces graph neural network technology to visualize the optimized prediction results of the oil spill diffusion path, making the prediction results more intuitive and easy to understand, which helps decision makers quickly understand the trend and scope of oil spill pollution diffusion, so as to take timely response measures to reduce environmental pollution and economic losses. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0054] Figure 1 This is a principle block diagram of a marine oil spill pollution diffusion early warning prediction system according to an embodiment of the present invention;

[0055] Figure 2 It is a flow chart of a method for early warning and prediction of marine oil spill pollution diffusion according to an embodiment of the present invention.

[0056] In the figure:

[0057] 1. Data collection and preprocessing unit; 2. Model building and path simulation unit; 3. Path prediction and anomaly detection unit; 4. Model parameter optimization unit; 5. Visual display unit. DETAILED DESCRIPTION

[0058] To further illustrate each embodiment, the present invention provides drawings, which are part of the disclosure of the present invention and are mainly used to illustrate the embodiments. They can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these contents, ordinary technicians in this field should be able to understand other possible implementation methods and advantages of the present invention.

[0059] According to an embodiment of the present invention, a system and method for early warning and prediction of oil spill pollution spread at sea are provided.

[0060] The present invention is further described with reference to the accompanying drawings and specific embodiments. Figure 1 As shown, according to the marine oil spill pollution diffusion early warning prediction system of the embodiment of the present invention, the system includes: a data acquisition and preprocessing unit 1, a model building and path simulation unit 2, a path prediction and anomaly detection unit 3, a model parameter optimization unit 4 and a visualization display unit 5;

[0061] The data acquisition and preprocessing unit 1 is used to acquire multi-source data in the marine environment and preprocess the acquired multi-source data.

[0062] In this optional embodiment, the data acquisition and preprocessing unit 1 includes: a data acquisition module, a data cleaning module and a feature engineering module;

[0063] The data acquisition module is used to collect multi-source data related to oil spill diffusion in the marine environment from several data sources.

[0064] In this optional embodiment, the multi-source data includes: meteorological data, ocean dynamics data, ship trajectories, historical oil spill case data and physical and chemical properties of the spilled oil.

[0065] It should be noted that meteorological data is one of the important parameters for predicting the spread of oil spills at sea, including key information such as wind speed, wind direction, temperature, humidity, etc. These factors will directly affect the drift and spread speed of oil spills in the ocean. For example, strong winds and specific wind directions will accelerate the spread of oil spills, while temperature and humidity may affect the evaporation rate of oil spills.

[0066] Ocean dynamics data mainly include current speed, direction, water temperature, salinity, etc. These data are crucial for simulating and predicting the movement trajectory of oil spills in the ocean. For example, the direction and speed of the current will directly affect the drift direction of the oil spill, while the water temperature may affect the solubility and biodegradation rate of the oil spill.

[0067] Ship trajectory data records information such as the navigation path and speed of ships at sea. This data is of great significance for determining the source of oil spills and possible spread paths. By analyzing ship trajectories, the location and time of possible oil spills can be inferred, thereby more accurately predicting the spread trend of oil spills.

[0068] Historical oil spill case data contains detailed information about oil spills that occurred in the past, such as the amount of oil spilled, the type of oil spilled, the speed of spread, the scope of impact, etc. These data provide valuable reference and basis for predicting future oil spills. By comparing and analyzing historical cases, we can discover the laws and trends of oil spill spread, and thus formulate more effective response strategies.

[0069] The physical and chemical properties of oil spills are one of the key factors in predicting their diffusion behavior. These properties include the density, viscosity, volatility, solubility, etc. of the oil spills. Different types of oil spills have different physical and chemical properties, which directly affect the behavior of the oil spills in the ocean. For example, oil spills with lower density are more likely to float on the water surface, while oil spills with higher viscosity may form a thicker oil film, affecting the diffusion rate.

[0070] The data cleaning module is used to clean the collected multi-source data to remove duplicate values ​​and process missing values.

[0071] The feature engineering module is used to reduce the dimensionality of the multi-source data after data cleaning using a nonlinear dimensionality reduction algorithm, and to establish a multi-source data set based on the physical mechanism of oil spill diffusion.

[0072] In this optional embodiment, the feature engineering module includes: a parameter setting submodule, a data dimension reduction submodule, a feature calculation submodule and a data set establishment submodule;

[0073] The parameter setting submodule is used to set the relevant parameters of the nonlinear dimensionality reduction algorithm based on the multi-source data after data cleaning.

[0074] The data dimension reduction submodule is used to use a nonlinear dimension reduction algorithm to perform dimension reduction processing on the multi-source data after data cleaning, output the multi-source data after dimension reduction, and establish an initial data set.

[0075] The feature calculation submodule is used to analyze and calculate the kinetic potential energy and pollutant drift index of the oil spill point according to the physical mechanism of oil spill diffusion.

[0076] It should be noted that the dynamic potential energy is to calculate the potential energy of the oil spill point by taking into account the influence of wind speed, water flow speed, wind direction and other factors on the spread of oil spill.

[0077] Pollutant drift index: Calculates the degree and direction of pollutant drift in the water based on water flow velocity, wind direction, oil spill area, etc.

[0078] The dataset establishment submodule is used to construct new feature variables based on the calculation results, and add the new feature variables to the initial dataset to establish a multi-source dataset.

[0079] It should be noted that, in the specific embodiment, it is assumed that the cleaned multi-source data contains 1000 samples, each sample has 20 features (including location, time, wind speed, etc.), as follows:

[0080] Step 1: Parameter setting

[0081] Select the t-SNE algorithm, set the perplexity to 30, the learning rate to 200, and the number of iterations to 1000.

[0082] Step 2: Dimensionality reduction

[0083] The t-SNE algorithm is used to reduce the 20-dimensional data to 2D; the reduced-dimensional data is used as the initial data set, containing 1000 samples, each with 2 features (coordinates after dimensionality reduction).

[0084] Step 3: Feature calculation

[0085] For each sample, calculate its kinetic potential energy = (wind speed 2) * 0.5.

[0086] Calculate the pollutant drift index = (water flow velocity * cosine value of wind direction + wind speed * cosine value of water flow direction) * oil spill area.

[0087] Step 4: Multi-source dataset creation

[0088] Initial data set (data after dimensionality reduction): 1000 samples, each with 2 features.

[0089] New feature variables: Kinetic potential (1000 values), Pollutant drift index (1000 values).

[0090] Multi-source dataset: 1000 samples, 4 features per sample (2 features after dimensionality reduction + kinetic potential + pollutant drift index).

[0091] The model building and path simulation unit 2 is used to build an offshore oil spill transport model based on the ocean numerical model, and simulate the path of oil spill diffusion in combination with the pre-processed multi-source data.

[0092] In this optional embodiment, the model building and path simulation unit 2 includes: a parameter setting module, a model building module and a diffusion path simulation module;

[0093] The parameter setting module is used to set the ocean area grid based on the ocean numerical model, discretize the control equations, and configure boundary conditions to simulate the dynamic process of the ocean area.

[0094] The model building module is used to combine the physical and chemical characteristics of oil spills to build an offshore oil spill transport model.

[0095] In this optional embodiment, the model construction module includes: a characteristic analysis submodule, a horizontal transport model construction submodule, an initial model construction submodule and a transport model construction submodule;

[0096] The characteristic analysis submodule is used to analyze the pre-processed multi-source data to obtain the physical and chemical characteristics of the oil spill during transportation.

[0097] The horizontal transport model construction submodule is used to establish an oil spill horizontal transport model based on the two-dimensional diffusion equation and combined with the physical and chemical characteristics of the oil spill during the transport process.

[0098] In this optional embodiment, the formula of the two-dimensional diffusion equation is:

[0099]

[0100] In the formula, represents partial differential; C represents oil film concentration; t represents time; λ represents the component of flow velocity in the x direction; x represents the spatial coordinate axis along the east-west direction; v represents the component of flow velocity in the y direction; y represents the spatial coordinate axis along the north-south direction; K represents the diffusion coefficient; represents the Laplace operator.

[0101] The initial model construction submodule is used to combine the oil spill horizontal transport model with the three-dimensional diffusion equation to establish the initial offshore oil spill transport model.

[0102] In this optional embodiment, the formula of the three-dimensional diffusion equation is:

[0103]

[0104] In the formula, represents partial differential; C represents oil film concentration; t represents time; represents the divergence operator; γ represents the three-dimensional velocity vector; N represents the diffusion tensor; S represents the source term.

[0105] The transport model construction submodule is used to perform spatial discretization processing on the initial offshore oil spill transport model using the finite difference method, and to construct the offshore oil spill transport model based on the spatial discretization processing results.

[0106] The diffusion path simulation module is used to simulate the motion trajectory of oil spill particles based on the constructed offshore oil spill transport model and adopt the Lagrangian particle tracking algorithm, and generate the oil spill diffusion path by combining the pre-processed multi-source data.

[0107] It should be noted that based on the constructed offshore oil spill transport model, the Lagrangian particle tracking algorithm is used to simulate the trajectory of oil spill particles, and combined with the pre-processed multi-source data, the oil spill diffusion path is generated, including:

[0108] Step 1: Parameter determination

[0109] Based on the constructed offshore oil spill transport model, the basic framework of the model is established, including the physical and chemical processes of oil spill particles such as convection, diffusion, evaporation, dissolution, and emulsification; model parameters are set, such as wind speed, water flow speed, direction, oil film thickness, physical properties of oil, etc.

[0110] 2. Lagrangian Particle Tracking Algorithm

[0111] (1) The oil spill film is discretized into a large number of oil particles, each of which represents a certain amount of oil; and the initial position, velocity, concentration and other attributes are set for each oil particle.

[0112] (2) The Lagrangian particle tracking algorithm is used to simulate the motion trajectory of each oil particle based on the preprocessed multi-source data (such as wind speed, water flow velocity, etc.); the convective transport and turbulent diffusion processes of oil particles, as well as the influence of chemical processes such as volatilization, dissolution, and emulsification of oil particles on the motion trajectory are considered.

[0113] (3) In each time step, the position information of the oil particles is updated according to their movement speed and direction; the interaction and collision between the oil particles, as well as the interaction between the oil particles and the marine environment (such as oil particles adhering to the shore, etc.) are taken into account.

[0114] 3. Generate oil spill diffusion path

[0115] (1) Record the movement trajectory of each oil particle, including position, time, speed and other information; analyze the trajectory data to extract the key paths and areas of oil spill diffusion.

[0116] (2) Use GIS or visualization software to present the oil spill diffusion path in a graphical manner. If necessary, marine environmental data (such as ocean currents, wind speed, etc.) can be superimposed to more intuitively understand the relationship between oil spill diffusion and environmental factors.

[0117] The path prediction and anomaly detection unit 3 is used to predict the oil spill diffusion path based on the simulation results of the oil spill diffusion path and historical multi-source data using a causal convolutional neural network model, and to detect anomalies in the multi-source data in real time.

[0118] In this optional embodiment, the path prediction and anomaly detection unit 3 includes: a training set building module, a model building and training module, a path prediction module and an outlier detection module;

[0119] The training set establishment module is used to collect and preprocess historical multi-source data and establish a training set in combination with the simulation results of the oil spill diffusion path;

[0120] The model building and training module is used to build a causal convolutional neural network model based on the convolutional neural network algorithm, and use the training set to train the built causal convolutional neural network model;

[0121] The path prediction module is used to collect real-time multi-source data and input the real-time multi-source data into the trained causal convolutional neural network model to predict the oil spill diffusion path;

[0122] The outlier detection module is used to detect outliers on the input real-time multi-source data stream in combination with the isolation forest algorithm.

[0123] The model parameter optimization unit 4 is used to feed back the detected abnormal values ​​to the causal convolutional neural network model, and optimize the path prediction result of the oil spill diffusion by adjusting the parameters of the causal convolutional neural network model.

[0124] It should be noted that the detected outliers are fed back into the causal convolutional neural network model. By adjusting the parameters of the causal convolutional neural network model, the prediction results of the oil spill diffusion path are optimized, including:

[0125] Step 1: Anomaly detection and feedback

[0126] (1) Mark and record the location and characteristics of outliers, and analyze whether the outliers are caused by model prediction errors, data noise, or the particularity of the actual oil spill; distinguish between real anomalies (such as changes in oil spill diffusion caused by sudden environmental events) and model errors.

[0127] (2) For outliers caused by model errors, they are used to adjust model parameters; for real anomalies, they can be considered to be included in the training set to increase the generalization ability of the model.

[0128] Step 2: Model parameter adjustment and optimization

[0129] According to the results of outlier analysis, adjust the model's hyperparameters, such as convolution kernel size, number, learning rate, batch size, etc.; you can use grid search, random search, or Bayesian optimization to find the optimal parameter combination.

[0130] The visualization display unit 5 is used to visualize the optimized oil spill diffusion path prediction results by using graph neural network technology.

[0131] It should be noted that the use of graph neural network technology can visualize the optimized oil spill diffusion path prediction results, including:

[0132] Step 1: Build a graph neural network model

[0133] (1) Design the graph neural network structure based on the physical mechanism and causal relationship of oil spill diffusion; select appropriate graph neural network models, such as Graph Convolutional Networks (GCNs), Graph Attention Networks (GATs), etc.

[0134] (2) Use the preprocessed data to train the graph neural network model.

[0135] Set appropriate loss functions and optimizers, such as cross entropy loss function and Adam optimizer.

[0136] Through iterative training, model parameters are optimized and prediction accuracy is improved.

[0137] Step 2: Prediction and optimization of oil spill diffusion paths

[0138] (1) Use the trained graph neural network model to predict the oil spill diffusion path based on real-time or historical data, and draw an oil spill diffusion path map based on the prediction results.

[0139] (2) Verify and evaluate the prediction results, such as using the validation set data to calculate indicators such as prediction accuracy and recall rate; based on the evaluation results, adjust the model parameters or structure to optimize the prediction performance.

[0140] Step 3: Visualization

[0141] (1) Select appropriate visualization tools, such as TensorBoard, Matplotlib, Seaborn, etc., and select appropriate chart types according to visualization requirements, such as heat maps, line graphs, scatter plots, etc.

[0142] (2) Draw an oil spill diffusion path map based on the prediction results; mark key locations and time points in the map, such as the oil spill source, diffusion range, diffusion speed, etc., and use different colors or marks to distinguish the diffusion situation in different time periods.

[0143] (3) Export the drawn oil spill diffusion path map into image or video format; display the visualization results on the emergency response system or decision support platform to provide intuitive information support for relevant personnel.

[0144] like Figure 2 According to another embodiment of the present invention, a method for early warning and prediction of oil spill pollution spread at sea is also provided, the method comprising:

[0145] S1. Collect multi-source data in the marine environment and pre-process the collected multi-source data;

[0146] S2. Based on the ocean numerical model, a marine oil spill transport model is constructed, and the oil spill diffusion path is simulated by combining the pre-processed multi-source data;

[0147] S3, based on the simulation results of the oil spill diffusion path and historical multi-source data, the causal convolutional neural network model is used to predict the oil spill diffusion path and detect outliers in the multi-source data in real time;

[0148] S4, feeding back the detected abnormal values ​​to the causal convolutional neural network model, and optimizing the oil spill diffusion path prediction results by adjusting the causal convolutional neural network model parameters;

[0149] S5. Use graph neural network technology to visualize the optimized oil spill diffusion path prediction results.

[0150] In summary, with the help of the above technical solution of the present invention, by collecting and preprocessing multi-source data in the marine environment, the monitoring scope of marine oil spill incidents is broadened, ensuring the comprehensiveness and timeliness of the data; at the same time, by constructing new characteristic variables, not only the accuracy of model simulation is enhanced, but also the prediction results are more in line with the actual situation, thereby greatly improving the accuracy and reliability of the prediction. By utilizing the causal convolutional neural network model, the oil spill diffusion path can be predicted quickly and accurately, thereby making full use of historical data and simulation results, capturing the dynamic characteristics of oil spill diffusion, and then realizing accurate prediction of the oil spill diffusion trend. By introducing graph neural network technology to visualize the optimized oil spill diffusion path prediction results, the prediction results are more intuitive and easy to understand, which helps decision makers quickly understand the trend and scope of oil spill pollution diffusion, so as to take timely countermeasures to reduce environmental pollution and economic losses.

[0151] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A marine oil spill pollution spread warning prediction system, characterized in that: The system includes: data collection and preprocessing unit, model building and path simulation unit, path prediction and anomaly detection unit, model parameter optimization unit and visualization display unit; The data acquisition and preprocessing unit is used to acquire multi-source data in the marine environment and preprocess the acquired multi-source data; The model building and path simulation unit is used to build an offshore oil spill transport model based on an ocean numerical model, and simulate the path of oil spill diffusion in combination with pre-processed multi-source data; The path prediction and anomaly detection unit is used to predict the oil spill diffusion path based on the simulation results of the oil spill diffusion path and historical multi-source data using a causal convolutional neural network model, and detect abnormal values ​​in the multi-source data in real time; The model parameter optimization unit is used to feed back the detected abnormal values ​​to the causal convolutional neural network model, and optimize the oil spill diffusion path prediction result by adjusting the causal convolutional neural network model parameters; The visualization display unit is used to visualize the optimized oil spill diffusion path prediction results using graph neural network technology.

2. The marine oil spill pollution diffusion early warning prediction system according to claim 1 is characterized in that: The data acquisition and preprocessing unit includes: a data acquisition module, a data cleaning module and a feature engineering module; The data acquisition module is used to collect multi-source data related to oil spill diffusion in the marine environment from a number of data sources; The data cleaning module is used to clean the collected multi-source data to remove duplicate values ​​and process missing values; The feature engineering module is used to perform dimensionality reduction processing on the multi-source data after data cleaning using a nonlinear dimensionality reduction algorithm, and to establish a multi-source data set based on the physical mechanism of oil spill diffusion.

3. The marine oil spill pollution diffusion early warning prediction system according to claim 2 is characterized in that: The multi-source data include: meteorological data, ocean dynamics data, ship trajectories, historical oil spill case data and physical and chemical properties of oil spills.

4. The marine oil spill pollution diffusion early warning and prediction system according to claim 3 is characterized in that: The feature engineering module includes: a parameter setting submodule, a data dimension reduction submodule, a feature calculation submodule and a data set establishment submodule; The parameter setting submodule is used to set relevant parameters of the nonlinear dimensionality reduction algorithm based on the multi-source data after data cleaning; The data dimension reduction submodule is used to perform dimension reduction processing on the multi-source data after data cleaning using a nonlinear dimension reduction algorithm, output the multi-source data after dimension reduction, and establish an initial data set; The characteristic calculation submodule is used to analyze and calculate the kinetic potential energy and pollutant drift index of the oil spill point according to the physical mechanism of oil spill diffusion; The data set establishment submodule is used to construct new feature variables based on the calculation results, and add the new feature variables to the initial data set to establish a multi-source data set.

5. The marine oil spill pollution diffusion early warning prediction system according to claim 1 is characterized in that: The model building and path simulation unit includes: a parameter setting module, a model building module and a diffusion path simulation module; The parameter setting module is used to set the ocean area grid based on the ocean numerical model, discretize the control equations, and configure boundary conditions to simulate the dynamic process of the ocean area; The model building module is used to build an offshore oil spill transport model by combining the physical and chemical properties of the oil spill; The diffusion path simulation module is used to simulate the movement trajectory of oil spill particles based on the constructed offshore oil spill transport model and adopt the Lagrangian particle tracking algorithm, and generate the oil spill diffusion path in combination with the pre-processed multi-source data.

6. The marine oil spill pollution diffusion early warning prediction system according to claim 5 is characterized in that: The model building module includes: a characteristic analysis submodule, a horizontal transport model building submodule, an initial model building submodule and a transport model building submodule; The characteristic analysis submodule is used to analyze the pre-processed multi-source data to obtain the physical and chemical characteristics of the oil spill during transportation; The horizontal transport model construction submodule is used to establish an oil spill horizontal transport model based on a two-dimensional diffusion equation and in combination with the physical and chemical characteristics of the oil spill during the transport process; The initial model building submodule is used to combine the oil spill horizontal transport model with the three-dimensional diffusion equation to establish an initial offshore oil spill transport model; The transport model construction submodule is used to perform spatial discretization processing on the initial offshore oil spill transport model using the finite difference method, and construct the offshore oil spill transport model based on the spatial discretization processing result.

7. The marine oil spill pollution diffusion early warning and prediction system according to claim 6 is characterized in that: The formula of the two-dimensional diffusion equation is: In the formula, represents partial differential; C represents oil film concentration; t represents time; λ represents the component of flow velocity in the x direction; x represents the spatial coordinate axis along the east-west direction; v represents the component of flow velocity in the y direction; y represents the spatial coordinate axis along the north-south direction; K represents the diffusion coefficient; represents the Laplace operator.

8. The marine oil spill pollution diffusion early warning and prediction system according to claim 7 is characterized in that: The formula of the three-dimensional diffusion equation is: In the formula, represents partial differential; C represents oil film concentration; t represents time; represents the divergence operator; γ represents the three-dimensional velocity vector; N represents the diffusion tensor; S represents the source term.

9. The marine oil spill pollution diffusion early warning and prediction system according to claim 1 is characterized in that: The path prediction and anomaly detection unit includes: a training set building module, a model building and training module, a path prediction module and an outlier detection module; The training set establishment module is used to collect and pre-process historical multi-source data, and establish a training set in combination with the simulation results of the oil spill diffusion path; The model building and training module is used to build a causal convolutional neural network model based on a convolutional neural network algorithm, and train the built causal convolutional neural network model using a training set; The path prediction module is used to collect real-time multi-source data and input the real-time multi-source data into the trained causal convolutional neural network model to predict the oil spill diffusion path; The outlier detection module is used to perform outlier detection on the input real-time multi-source data stream in combination with the isolation forest algorithm.

10. A method for early warning and prediction of oil spill pollution spread at sea, used to implement the operation of the early warning and prediction system for oil spill pollution spread at sea as claimed in any one of claims 1 to 9, characterized in that: The method includes: S1. Collect multi-source data in the marine environment and pre-process the collected multi-source data; S2. Based on the ocean numerical model, a marine oil spill transport model is constructed, and the oil spill diffusion path is simulated by combining the pre-processed multi-source data; S3, based on the simulation results of the oil spill diffusion path and historical multi-source data, the causal convolutional neural network model is used to predict the oil spill diffusion path and detect outliers in the multi-source data in real time; S4, feeding back the detected abnormal values ​​to the causal convolutional neural network model, and optimizing the oil spill diffusion path prediction results by adjusting the causal convolutional neural network model parameters; S5. Use graph neural network technology to visualize the optimized oil spill diffusion path prediction results.

Images