Marine unmanned aerial vehicle multi-source data intelligent mining and early warning decision-making system

Abstract

The invention discloses a marine unmanned aerial vehicle multi-source data intelligent mining and early warning decision-making system, and relates to the technical field of unmanned aerial vehicles, and the system comprises an unmanned aerial vehicle control module which controls a marine unmanned aerial vehicle group to execute a monitoring task according to a preset inspection route; the multi-source data collection module is distributed on each marine unmanned aerial vehicle and is used for collecting various types of marine data in a target sea area; and the multi-source data integration module is used for integrating various types of ocean data obtained by the ocean unmanned aerial vehicle group and extracting valuable information. According to the invention, comprehensive monitoring and accurate analysis of the oil leakage condition of a target sea area can be realized through cooperation of the modules, potential risks can be found in time, and early warning can be generated; and effective decision support can be provided for countermeasures, and the efficiency and scientificity of petroleum leakage monitoring and processing are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicles, and more specifically, to an intelligent mining and early warning decision-making system for multi-source data of marine unmanned aerial vehicles. Background Art

[0002] A marine unmanned aerial vehicle is a high-tech device designed specifically for the marine environment, with environmental adaptability characteristics such as wind resistance, salt fog resistance, and corrosion resistance, and can achieve efficient operation through advanced technologies such as precise landing technology and integration of current measurement systems. It plays an important role in multiple fields such as marine monitoring, marine mapping, maritime law enforcement and rescue, and fishery management. It can carry a variety of sensors and quickly and accurately complete tasks such as marine environment monitoring, coastline mapping, maritime law enforcement evidence collection, and fishery resource investigation, providing strong support for marine scientific research, marine resource development, and marine ecological protection.

[0003] In the existing technical system, once an oil spill occurs in the ocean, it is a key problem to be solved urgently to identify and judge the spill situation in a timely and accurate manner, and then formulate an effective response plan. The marine environment is extremely complex, and factors such as seawater flow, tidal changes, and weather conditions are intertwined, greatly interfering with the accurate judgment of oil spills. Traditional monitoring methods, such as relying on ship patrols, not only have a limited monitoring range but also low efficiency, and it is difficult to detect spill signs in a timely manner in the vast sea area. Although satellite remote sensing can cover a large area, it is limited by cloud cover, resolution, etc., and it is difficult to accurately identify small or early spills. At the same time, there is a lack of efficient and accurate technical methods for judging key information such as the type, diffusion speed, and impact range of the spilled oil. The existing data collection methods are often relatively single, and it is impossible to comprehensively collect various data closely related to oil spills, such as ocean temperature, salinity, and sea current speed, resulting in an incomplete and in-depth analysis of the oil spill situation. These problems seriously restrict the timely response and effective handling of oil spill incidents, and new technologies and methods are urgently needed to break through the dilemma. Summary of the Invention

[0004] To solve the above problems, the present invention provides an intelligent mining and early warning decision-making system for multi-source data of marine unmanned aerial vehicles.

[0005] The present invention provides an intelligent mining and early warning decision-making system for multi-source data of marine unmanned aerial vehicles, which is applied to an oil spill network and includes: An unmanned aerial vehicle control module that controls a group of marine unmanned aerial vehicles to perform monitoring tasks according to a preset inspection route; A multi-source data collection module, distributed on each marine unmanned aerial vehicle, for collecting various types of marine data in the target sea area; A multi-source data integration module for integrating various types of marine data obtained by the group of marine unmanned aerial vehicles and extracting valuable information; A data mining unit for further analyzing the valuable information extracted by the multi-source data integration module, identifying abnormal data and performing risk assessment; An early warning decision-making module, including an early warning unit and a decision support unit. The early warning unit generates early warning information based on the risk assessment of the data mining module, and the decision support unit is used to provide decision support for the early warning information.

[0006] Preferably, it further includes an ocean oil spill identification module, and the specific working process is as follows: Use a hyperspectral camera carried by an ocean drone to photograph the target sea area, capture the light reflection information of different wavelengths, and generate image data containing rich spectral features; Based on the input of the image data into a trained convolutional neural network model, perform feature extraction and analysis on the image through convolutional layers, pooling layers, etc. in the model to distinguish oil spills from natural oil films and identify the type of oil product.

[0007] Preferably, it further includes an oil source tracing module, and the specific working method is as follows: When the ocean drone monitors the oil spill area, while recording the position information at the time of shooting, collect the sea current data in this area, including sea current speed and direction information; Starting from the current position of the oil spill, based on the sea current data and the time series image information taken by the ocean drone, reverse calculate the positions of the oil spill at different time points.

[0008] Preferably, after the ocean oil spill identification module distinguishes that it is an oil spill, it includes: Obtain the viscosity of the leaked oil , the density of seawater , the density of oil , the oil film thickness h, the current wind speed u in the sea area, the oil film diameter d, the time elapsed after the formation of the oil film , the current seawater temperature T, the current oil film area A, the initial area of the oil film , the current oil-water interface tension and the oil-water interface tension under standard conditions ; According to the formula , obtain the oil film stability; The early warning unit formulates an early warning decision based on the oil film stability S.

[0009] Preferably, after the ocean oil spill identification module distinguishes that it is an oil spill, it further includes: Obtain the lowest seawater temperature observed in the historical records of the current sea area ; Obtain the range of seawater temperature change observed in the historical records of the current sea area ; Obtain the current ocean current speed V; Obtain the weathering degree F of the oil; Based on the formula Obtain the oil diffusion assessment index, where is the wind direction factor, with a value range of 0 - 1, is the maximum ocean current speed of the historical distance in this sea area, used for normalizing the ocean current speed, is the maximum oil film thickness that is expected to appear in this type of oil spill scenario, used for normalizing the oil film thickness; The warning unit predicts the oil diffusion situation based on the oil diffusion assessment index

[0010] Preferably, the range of the observed seawater temperature change in the historical records of the current sea area is calculated as: Through Obtain the range of the observed seawater temperature change in the historical records of the current sea area , where is the highest seawater temperature observed in the historical records of the current sea area.

[0011] Preferably, the warning unit is also used to divide the warning levels, specifically: Regarding the large oil diffusion comprehensive index I and strong oil film stability S as the first warning level; Regarding the large oil diffusion comprehensive index I and poor oil film stability S as the second warning level; Regarding the small oil diffusion comprehensive index I and strong oil film stability S as the third warning level; Regarding the small oil diffusion comprehensive index I and poor oil film stability S as the fourth warning level; Among them, the first warning level > the second warning level > the third warning level > the fourth warning level.

[0012] Preferably, the specific steps for the multi-source data integration module to extract valuable information are: Obtain the data collected by the marine drone swarm from different angles and methods for the marine environment; Compare the collected data with the data in the pre-constructed database one by one; During the comparison process, according to the set difference criteria, screen out the data with obvious differences; Conduct in-depth analysis on the screened differential data, extract the differential content presented in the data, and determine it as valuable information.

[0013] Preferably, the specific method for the data mining unit to identify abnormal data and conduct risk assessment is:​ Preset a difference threshold; Identify data with differential content exceeding the difference threshold as abnormal data; Use the identified abnormal data as input to establish a risk assessment model; The warning unit generates warning information based on the risk assessment result.

[0014] Preferably, the decision support unit is used to provide decision support for the warning information, including but not limited to issuing navigation warnings, activating emergency plans, and adjusting operation plans.

[0015] Beneficial effects: Through the UAV control module, the marine UAV group can perform monitoring tasks according to the preset route. The multi-source data collection module is distributed on the UAVs to collect various marine data. The multi-source data integration module integrates the data and extracts valuable information. The data mining unit further analyzes the information to identify abnormal data and evaluate risks. The warning unit of the warning decision module generates warning information based on the risk assessment. The decision support unit provides decision support. In this way, a comprehensive monitoring, accurate analysis of the oil spill situation in the target sea area can be realized, potential risks can be discovered in time and warnings can be generated, and effective decision support can be provided for countermeasures, improving the efficiency and scientificity of oil spill monitoring and handling. Brief Description of the Drawings

[0016] Figure 1 It is a flowchart of the warning decision system of the present invention. Detailed Embodiment

[0017] As Figure 1 shown: The marine UAV multi-source data intelligent mining and warning decision system is applied to the oil spill network and includes: The UAV control module controls the marine UAV group to perform monitoring tasks according to the preset inspection route; It should be understood that the UAV control module issues precise instructions to the marine UAV group according to the pre-planned inspection route to ensure that each UAV can perform monitoring tasks according to the plan. When planning the inspection route, various factors such as the scope, shape, wind direction, and ocean current of the target sea area will be comprehensively considered to achieve efficient and comprehensive monitoring coverage.

[0018] The multi-source data collection module is distributed on each marine UAV and is used to collect various types of marine data in the target sea area; It should be understood that this module is equipped with a variety of advanced sensor devices, which can collect various types of ocean data in the target sea area. For example, temperature sensors are used to collect seawater temperature data, which is crucial for understanding the marine ecological environment and the impact of oil spills on local water temperature; salinity sensors are used to obtain seawater salinity information, and changes in salinity can reflect changes in seawater composition and have a potential association with oil spills; there are also water quality sensors that can detect the concentration of various pollutants in seawater, etc. These different types of data reflect the real-time situation of the target sea area from multiple dimensions, and so on.

[0019] It should be noted that in order to achieve the integrity of data collection, devices including but not limited to shipborne sensors and satellites are used for collection, and a high-precision clock synchronization system is synchronously equipped, such as the Global Positioning (GPS) clock, to ensure that all devices collect data under the same time reference. Moreover, the geographical location and attitude information of each device are clarified, and the data collected by different devices are unified into the same geographical coordinate system through coordinate transformation algorithms; then, the conversion relationship between different device coordinate systems is established to convert the data of drones, ships, and satellites into a unified geographical coordinate system; For data with different collection times, a time interpolation algorithm is used for processing to make the data have the same sampling frequency in the time dimension and fuse the data at the original data level by directly combining the data collected by different devices. For example, high-resolution images taken by drones are fused with satellite remote sensing images, and through technologies such as image registration and pixel-level fusion, more comprehensive and detailed image information is generated. First, the features of the data of each device are extracted, and then the extracted features are fused; On the basis of the independent processing and analysis of the data of each device, the obtained decision results are fused. For example, drones monitor the marine pollution area and make a preliminary judgment, and satellites conduct a macroscopic analysis of a large area of the sea area and give a pollution trend prediction, and then the results of the two are integrated to obtain a more accurate marine pollution assessment result; Data cleaning: The collected data is preprocessed to remove noise, outliers, and incorrect data. For example, noise points in drone images are removed through filtering algorithms, and outliers in ship sensor data are detected and corrected using statistical methods; Data calibration: The systematic errors and random errors of the devices are calibrated; for example, radiometric calibration and geometric correction are performed on satellite remote sensing data to improve the accuracy and reliability of the data; Visualization and analysis of the fused data - Visualization display: The fused data is visually displayed through tools such as Geographic Information System (GIS) to more intuitively observe and analyze the spatial distribution and temporal changes of the data. For example, the data of drones, ships, and satellites are superimposed on an electronic nautical chart to display the comprehensive situation of the marine environment.

[0020] Comprehensive analysis: Use technologies such as data mining and machine learning to deeply analyze the fused data and extract valuable information and knowledge; for example, by analyzing historical data and real-time data, predict the changing trends of the marine environment to provide support for marine management and decision-making.

[0021] Multi-source data integration module, used to integrate various types of marine data obtained by the marine drone swarm and extract valuable information; It should be understood that this module is responsible for integrating the massive and complex various types of marine data obtained by the marine drone swarm. First, it needs to sort out and unify the data in different formats and from different sources; for example, convert the temperature data recorded in a certain specific format by drone A and the salinity data in a different format by drone B into a common format that the system can recognize and process. Then, use specific algorithms and technologies to extract valuable information from the integrated data. These valuable information includes but is not limited to color changes, temperature changes, and texture changes in the mixture of seawater and oil.

[0022] Data mining unit, used to further analyze the valuable information extracted by the multi-source data integration module, identify abnormal data and conduct risk assessment; It should be understood that this module further analyzes the valuable information extracted by the multi-source data integration module. By using data mining algorithms such as clustering analysis and association rule mining, deeply explore the patterns and rules hidden behind the data; in this process, abnormal data can be accurately identified, for example, it is found that the concentration of seawater pollutants in a certain area far exceeds the normal range, or the water temperature shows an abnormal increase; at the same time, based on these abnormal data, combined with historical data and professional knowledge, conduct a risk assessment of the current marine situation, and judge the possible harm degree caused by oil spills, such as the threat to the living environment of marine organisms and the impact on the surrounding coastal line ecology.

[0023] Early warning decision-making module, including an early warning unit and a decision support unit. The early warning unit generates early warning information based on the risk assessment of the data mining module, and the decision support unit is used to provide decision support for the early warning information.

[0024] It should be understood that the early warning unit quickly generates early warning information based on the risk assessment results of the data mining module. If the risk assessment shows that an oil spill may pose a serious threat to an important marine ecological protection area, the early warning unit will immediately issue an early warning of the corresponding level, such as a red warning, to alert relevant departments and personnel. The decision-making support unit provides comprehensive decision-making support for the early warning information. It comprehensively considers various factors, such as the early warning level, the current status of the marine environment, available response resources, etc. Through data analysis and simulation, it provides decision-makers with multiple response plans and predictions of their possible effects, helping decision-makers make scientific and reasonable decisions, such as whether to immediately initiate a large-scale oil cleaning operation and where to allocate oil cleaning resources.

[0025] As a further embodiment, it also includes an ocean oil spill identification module. The specific working process is as follows: The target sea area is photographed by a hyperspectral camera carried by an ocean drone, capturing the light reflection information of different wavelengths and generating image data containing rich spectral features. Based on the input of the image data into a trained convolutional neural network model, the convolutional layer, pooling layer, etc. in the model are used to extract and analyze the features of the image, distinguish oil spills from natural oil films, and identify the type of oil product.

[0026] It should be understood that the hyperspectral camera carried by the drone can capture the light reflection information of different wavelengths in the target sea area. This information is extremely rich, covering multiple bands from visible light to near-infrared. During the shooting process, it shoots the target sea area with extremely high resolution, generating image data containing rich spectral features. For example, when shooting a suspected oil spill area, the hyperspectral camera can clearly record the differences in the reflection of different wavelengths of light in different areas. These differences are important bases for identifying oil spills. Subsequently, based on the collected image data, it is input into a convolutional neural network model trained with a large number of samples. The convolutional neural network model contains multiple convolutional layers and pooling layers inside. The convolutional layer slides different convolutional kernels on the image to extract local features of the image, such as edges, textures, etc. The pooling layer then performs dimensionality reduction processing on the features extracted by the convolutional layer, reducing the amount of data while retaining key features. When processing oil spill images, the model can effectively distinguish oil spills from natural oil films through the analysis of these features because there are subtle differences in the spectral features between oil spills and natural oil films, and the trained model can accurately identify these differences. At the same time, the model can further identify the type of oil product based on the spectral features. Different types of oil products have unique spectral absorption and reflection characteristics. The model learns these characteristics to achieve accurate judgment of the type of oil product, providing an important reference for formulating subsequent response measures. Different oil products have different treatment methods and degrees of harm.

[0027] As a further embodiment, it further includes an oil source tracing module, and the specific working method is as follows: When the marine drone monitors the oil spill area, while recording the position information at the time of shooting, it collects the sea current data in this area, including sea current speed and direction information; It should be understood that when the marine drone monitors the oil spill area, it undertakes a dual data collection task. On the one hand, through a high-precision positioning device, it accurately records its own position information at the time of shooting, accurate to multiple decimal places of longitude and latitude, to ensure the accuracy of the position information. On the other hand, it uses a special sea current monitoring device to collect the sea current data in this area, including sea current speed and direction information. For example, the sea current speed is measured by a Doppler current meter, and the sea current direction is determined by devices such as a compass; these data can reflect the flow situation of the seawater in the oil spill area, facilitating the tracing of the oil source to formulate different emergency plans.

[0028] Taking the current position of the oil spill as the starting point, based on the sea current data and the time series image information captured by the marine drone, reverse calculate the positions of the oil spill at different time points.

[0029] It should be understood that taking the current position of the oil spill as the starting point, based on the collected sea current data and the time series image information captured by the marine drone, use mathematical models and algorithms for reverse calculation; the time series images record the positions and morphological changes of the oil spill at different time points. Combining with the sea current data, it can simulate the drift trajectory of the oil spill in seawater; for example, assume that at a certain moment the oil spill is at point A. According to the sea current speed and direction at that time, and the position of the oil spill in the image captured at the previous moment, it can be inferred through calculation that the position of the oil spill at the previous moment is point B, and so on, gradually reverse calculating the positions of the oil spill at different time points; in this way, the source of the oil leakage can be traced, providing a strong basis for measures such as plugging the leakage source.

[0030] As a further embodiment, after the marine oil spill identification module distinguishes that it is an oil spill, it includes; Obtain the viscosity of the leaked oil , the density of seawater , the density of oil , the thickness h of the oil film, the current sea area wind speed u, the diameter d of the oil film, the time elapsed after the formation of the oil film , the current seawater temperature T, the current oil film area A, the initial area of the oil film , the current oil-water interfacial tension and the oil-water interfacial tension under standard conditions ; According to the formula , obtain the oil film stability; it should be understood that the density difference between oil and seawater It has an important impact on the stability of the oil film. When the density of the oil is close to that of seawater, the buoyancy and gravity acting on the oil film in seawater are relatively balanced, and the oil film is more likely to maintain a relatively stable position and shape on the sea surface; if the density difference is too large, the oil film may quickly float or sink, resulting in a decrease in its stability. For example, light oil with a smaller density is more likely to spread on the sea surface to form a thinner oil film, but if the density is too small, it may be quickly dispersed on the water surface under the action of waves, etc., affecting the integrity and stability of the oil film; The greater the wind speed, the greater the shear force acting on the surface of the oil film. Strong winds will generate a large frictional force on the surface of the oil film, causing the molecules on the surface of the oil film to be subjected to an outward pulling force, which is likely to cause the surface of the oil film to deform and break, thereby reducing the stability S of the oil film.

[0031] Make early warning decisions based on the stability S of the oil film.

[0032] Specifically, collect oil samples from the leakage site and send them to the laboratory as soon as possible. Use a rotational viscometer. Place the sample in the measuring cup of the viscometer, and the rotor rotates in the sample. Calculate the viscosity of the oil by measuring the resistance suffered by the rotor during rotation. A capillary viscometer can also be used. According to Poiseuille's law, measure the time required for the oil to pass through the capillary under a certain pressure, and then calculate the viscosity. Use a high-precision densitometer, such as a vibrating cylinder densitometer, to obtain the seawater density etc. The oil film thickness h can be obtained by using an optical sensor carried by a satellite or an unmanned aerial vehicle. By analyzing the reflection and absorption characteristics of the oil film for light of a specific wavelength and combining with a radiative transfer model to invert the oil film thickness, or using a sounding rod or a special oil film thickness measuring instrument to conduct on-site measurements at multiple representative points in the oil film area, etc.; the wind speed v in the current sea area can be obtained by querying the real-time wind speed data of the ground meteorological station or ocean meteorological buoy closest to the oil leakage area. These stations are equipped with high-precision wind speed measuring instruments, such as a three-cup anemometer or an ultrasonic anemometer, which can provide relatively accurate wind speed information; with the help of satellite remote sensing images, visible light or infrared images taken by unmanned aerial vehicles, through image processing software, use an edge detection algorithm to identify the boundary of the oil film, and then calculate the oil film diameter d according to the shape of the oil film (assuming it is circular or through an equivalent diameter calculation method); if the shape of the oil film is irregular, the method of equivalent area can be adopted, that is, according to the oil film area A, through the formula Calculate the equivalent diameter, etc.; The time elapsed since the formation of the oil film , starting from the moment when the oil spill is discovered, by accurately recording the time point of the spill and the current measurement time point, the time difference between the two is the time elapsed after the oil film is formed, etc.; the current seawater temperature T is measured using a high-precision thermometer, such as a platinum resistance thermometer or a thermistor thermometer, at multiple depths in different areas of the oil spill to obtain the vertical distribution and average value of the seawater temperature in this area, etc.; the current oil film area A and the initial area of the oil film , similar to the image analysis method in the way of obtaining the oil film diameter d, the oil film boundary is identified through satellite or drone images, and the image processing software is used to calculate the oil film area; for the initial area , it is necessary to obtain the images in the initial stage of the spill for analysis and calculation; the current area A is obtained from the real-time monitoring images, the current oil-water interfacial tension and the oil-water interfacial tension under standard conditions can be calculated according to information such as the salinity of the seawater and the chemical composition of the oil using theoretical models (such as the extended Langmuir - Schaefer model), and other methods are not listed one by one.

[0033] By obtaining the oil film stability S to formulate different early warning levels and plans, the higher the oil film stability S, the higher the oil film stability, and the oil film is difficult to decompose naturally. Judging the oil film stability S helps to predict the long-term behavior of oil in the marine environment; on the one hand, the higher the oil film stability S indicates that the oil is not easy to disperse, and the stable oil film may float on the sea surface for a long time, blocking the penetration of sunlight and affecting the photosynthesis of marine phytoplankton, thereby destroying the basis of the entire marine food chain; through continuous monitoring and analysis of the oil film stability, the long-term changes in the structure and function of the marine ecosystem caused by the oil spill can be predicted, providing forward-looking guidance for ecological restoration and protection.

[0034] As a further embodiment, after the marine oil spill identification module distinguishes the oil spill, it further includes: Obtaining the lowest seawater temperature observed in the historical records of the current sea area ; it should be understood that it is used to eliminate the lower limit influence of the temperature data to ensure that the temperature value is positive and can reflect the relative change.

[0035] Obtaining the range of seawater temperature change observed in the historical records of the current sea area ; it should be understood that it is used to normalize the temperature data to a relative proportion range.

[0036] Obtaining the current sea current velocity V; it should be understood that the sea current, as a physical driving force, directly affects the diffusion speed and direction of the oil.

[0037] Obtain the weathering degree F of the oil; it should be understood that 0 represents no weathering and 1 represents complete weathering. Weathering will change the physical and chemical properties of the oil. For example, a decrease in viscosity will make the oil easier to spread. Therefore, (1 + F) is used to reflect the promoting effect of weathering on diffusion.

[0038] Based on the formula Obtain the oil diffusion evaluation index, where is the wind direction factor, and its value range is 0 - 1. is the maximum sea current speed of the historical distance in this sea area, which is used to normalize the sea current speed. is the maximum oil film thickness that is expected to appear in this type of oil spill scenario, which is used to normalize the oil film thickness; specifically, a higher temperature T can accelerate the movement of oil molecules and promote diffusion. The thicker the oil film thickness d, it means that more oil quantity participates in the diffusion process. It reflects the relationship between the wind direction and the sensitive areas (such as the coastline, marine ecological protection areas, etc.) in the direction of oil diffusion. If the wind direction is towards the sensitive area, this value is close to 1; if the wind direction is away from the sensitive area, this value is close to 0; its specific calculation can be determined based on the angle between the wind direction and the direction of the sensitive area through trigonometric functions, etc.

[0039] Based on the oil diffusion evaluation index Predict the oil diffusion situation.

[0040] It should be understood that by normalizing each parameter and comprehensively considering their interactions, a quantitative comprehensive index is provided to help understand and evaluate the diffusion situation of oil in the marine environment. It converts multiple qualitative or quantitative parameters into a comprehensive index I, realizing the quantification of the potential impact degree of oil diffusion; this quantification method is convenient for comparing the oil diffusion situations in different scenarios. Whether it is oil spills in different regions or the evaluation of the diffusion states at different times in the same region, it can be intuitively shown through this value, providing a clear judgment basis for decision-makers, such as assisting in decision-making in aspects such as resource allocation and response strategy selection.

[0041] It should be understood that the comprehensive oil diffusion index can integrate various factors affecting oil diffusion, comprehensively reflect the diffusion range, speed and direction of oil in the ocean. Combined with the oil film stability, it can more accurately judge which marine ecological regions will be threatened and the degree of threat; for example, if an oil film with high stability spreads towards the coral reef area, because it can maintain a relatively large area and a relatively concentrated state for a long time, the damage to the coral reef ecosystem may be more serious. Through accurate assessment, protective measures can be taken in advance for key ecological regions, such as setting up oil booms to reduce the damage of oil to the ecosystem. The comprehensive index of oil spill diffusion and the assessment of oil film stability can help decision-makers accurately judge the severity of oil spills and their possible development trends. For oil spills with high stability and great diffusion potential, more resources can be concentrated for response, such as deploying more oil cleaning vessels, oil booms and other equipment to key areas, avoiding wasting resources in areas with low diffusion possibility, and improving the cost-effectiveness of oil cleaning operations. Based on the comprehensive index of oil spill diffusion and the stability of the oil film, the emergency command department can formulate more targeted and scientific response strategies. For oil films with high stability, due to their great diffusion hazards, physical methods (such as mechanical recovery) can be preferentially used for treatment. For oil films with relatively low stability and easy to disperse, chemical oil dispersants may be more suitable. This strategy formulation based on scientific assessment can improve the efficiency of oil cleaning and accelerate the emergency response process.

[0042] As a further embodiment, the range of seawater temperature variations observed in the historical records of the current sea area is calculated as follows: By obtaining the range of seawater temperature variations observed in the historical records of the current sea area , where is the highest seawater temperature observed in the historical records of the current sea area.

[0043] As a further embodiment, the warning unit is also used to divide the warning levels, specifically as follows: Regarding the comprehensive index of oil spill diffusion I being large and the stability S of the oil film being strong as the first warning level. Specifically, a large comprehensive index indicates strong oil spill diffusion ability, which will quickly and widely affect a large area of the sea area; strong stability means that the oil film remains intact for a long time, continuously blocking sunlight from reaching marine phytoplankton, seriously damaging the basic productivity of the marine ecosystem, and causing long-term and serious impacts on the entire marine food chain. On the one hand, its rapid diffusion and long-term stable existence pose a huge threat to a large number of marine resources, coastal economic regions and human activities, such as affecting industries like fisheries and tourism, and threatening the health of coastal residents. On the other hand, large-scale and long-term resource investment is required for interception, cleaning and ecological restoration during response, involving complex coordination work with extremely high difficulty. If not handled in a timely manner, the scope and degree of harm will rapidly expand.

[0044] Take the large comprehensive index I of oil spill and the poor stability S of the oil film as the second warning level; it should be understood that although a large comprehensive index will also lead to rapid and extensive spread, the poor stability makes the oil film easy to break and disperse, and the impact on the ecosystem is relatively dispersed and will not continuously affect the same area in the long term. From the perspective of the long-term and overall ecological damage degree, it is relatively less serious than the situation with strong stability. Although the fast spread increases the difficulty of response, due to its instability, in emergency treatment, its characteristics can be utilized to relatively quickly disperse and treat it by using chemical oil dispersants, etc., and the long-term impact on a single area is small, and the overall response difficulty and potential risk are slightly lower than the situation with strong stability.

[0045] Take the small comprehensive index I of oil spill and the strong stability S of the oil film as the third warning level; it should be understood that the small comprehensive index limits the spread range, but the strong stability will cause continuous pressure on the local marine ecosystem, and the impact is relatively limited to the local area. Compared with the situation with a large spread range, the overall impact degree on the ecological environment is lower. The small comprehensive index limits the spread range, but the strong stability will cause continuous pressure on the local marine ecosystem, and the impact is relatively limited to the local area. Compared with the situation with a large spread range, the overall impact degree on the ecological environment is lower; the main risk is concentrated in the long-term impact on the local area. Although long-term monitoring and local intervention are required, compared with the situation with a large spread range, the response scale and complexity are smaller, and the potential risk is also relatively lower.

[0046] Take the small comprehensive index I of oil spill and the poor stability S of the oil film as the fourth warning level; it should be understood that the small comprehensive index determines that the spread range and speed are limited, and the poor stability makes the oil film easy to dissipate, mainly causing a certain impact on the local sea area near the leakage point in a short time, and the degree of damage to the overall marine ecological environment is relatively the smallest. Due to the small spread range and instability, it is relatively easy to clean up the pollution. Through simple and timely cleaning and monitoring, the situation can be better controlled, and the potential risk and response difficulty are the lowest.

[0047] Among them, the first warning level > the second warning level > the third warning level > the fourth warning level.

[0048] It should be understood that for the first warning level, a large-scale emergency response mechanism is immediately activated to quickly allocate a large number of resources such as oil cleaning vessels and aircraft, and utilize their mobility to track dispersed oil droplets in the vast sea area. Oil-absorbing vessels, oil-absorbing felts and other equipment are used to adsorb and recover the oil droplets on the sea surface. At the same time, for the small and dispersed oil droplets that are difficult to recover, chemical oil dispersants are reasonably sprayed to accelerate their emulsification and decomposition. However, the dosage and scope of the oil dispersants should be strictly controlled to reduce the secondary pollution to the marine environment. Multiple means such as satellite remote sensing, unmanned aerial vehicle monitoring and sea surface monitoring buoys are used to track and monitor the oil diffusion range and oil droplet concentration in real time; the diffusion model is updated in a timely manner to predict the oil diffusion path and speed, providing accurate decision-making basis for oil cleaning operations and ecological protection, and issuing warning information to the surrounding sea areas to remind passing ships and relevant departments to pay attention to prevention; For the second warning level, large-scale and high-strength oil booms are set up on the oil film diffusion path, especially in the direction close to sensitive areas (such as coasts, important marine protected areas), to prevent the further diffusion of the oil film. At the same time, a large number of oil cleaning vessels are allocated, and equipment such as skimmers is used to continuously recover the oil film within the oil boom, minimizing the coverage area of the oil film on the sea surface as much as possible; a long-term ecological restoration plan is formulated. Considering the long-term impact of the oil film on the marine ecosystem, a large amount of resources may be required for ecological restoration work, such as measures like planting marine plants and releasing fry, to gradually restore the damaged marine ecological environment; at the same time, long-term supervision of this area is strengthened to ensure the effective implementation of ecological restoration work; For the third warning level, long-term monitoring stations are set up in the local area where the oil film is located to continuously monitor the dynamic changes of the oil film, the water quality of the surrounding sea water, the biological health status, etc., closely observing whether there are signs of oil film diffusion and its chronic impact on the surrounding ecological environment, providing long-term data accumulation for subsequent decision-making, conducting a detailed risk assessment of this area, analyzing the possible diffusion risks and ecological impacts of the oil film under different marine environmental conditions; according to the assessment results, the emergency plan is adjusted in a timely manner to ensure that possible expanded oil pollution can be quickly and effectively responded to in case of emergencies; For the fourth warning level, after the discovery of an oil spill, a small oil cleaning team and equipment, such as small oil booms, oil-absorbing felts, etc., are quickly organized to quickly block and adsorb and clean the oil film near the spill point. Since the oil film has poor stability and a limited diffusion range, timely cleaning in the initial stage can effectively control the pollution range and reduce the impact on the surrounding sea area. Although the current diffusion trend is not obvious, continuous observation of this area is still required, paying attention to changes in marine environmental conditions (such as wind direction, sea current), to prevent the expansion of the oil film diffusion range due to sudden factors; at the same time, the oil spill risk assessment and preventive measures for similar areas are strengthened to avoid the recurrence of similar incidents.

[0049] As a further embodiment, the specific steps for the multi-source data integration module to extract valuable information are as follows: Obtain the data collected by the marine drone swarm from different angles and in different ways on the marine environment; Compare the collected data with the data in the pre-constructed database one by one; During the comparison process, filter out the data with obvious differences according to the set difference criteria; Deeply analyze the filtered differential data, extract the differential content presented in the data, and determine it as valuable information.

[0050] It should be understood that the marine drone swarm collects data on the marine environment from different angles and in different ways. Different drones may be equipped with sensors with different focuses. Some focus on water quality monitoring, some on marine biology monitoring, and some on physical environment monitoring; they collect various types of data in the target sea area according to their respective patrol routes and task requirements; for example, some drones take close-up photos of the texture and color changes on the sea surface at low altitude, and some drones use radar and other equipment to detect the sea depth and seabed topography at high altitude; Compare the collected data with the data in the pre-constructed database one by one. This database is constructed based on long-term marine monitoring data, historical oil spill case data, and marine environment standard data, etc.; during the comparison process, use an efficient data matching algorithm to quickly find the similarities and differences between the collected data and the data in the database; for example, compare the collected sea water temperature data with the historical temperature data of the same period in this sea area in the database to check if there are any abnormal deviations; During the comparison process, filter out the data with obvious differences according to the set difference criteria; the setting of the difference criteria is based on in-depth research on the laws of marine environmental changes and the prediction of data changes that may be caused by oil spills; for example, set that the concentration of sea water pollutants exceeding the normal standard by 10% is considered significantly different. When the pollutant concentration in the collected data reaches this standard, this data will be filtered out; Deeply analyze the filtered differential data; use data analysis tools and algorithms to extract the differential content presented in the data and determine it as valuable information; for example, through further analysis of the filtered abnormal sea water temperature data, discover the distribution law of its temperature difference from the surrounding area, and take this difference law and the corresponding temperature data as valuable information to provide a basis for subsequent judgment of the impact range and degree of oil spills on the marine environment.

[0051] As a further embodiment, the specific way for the data mining unit to identify abnormal data and conduct risk assessment is: Preset a difference threshold; the setting of the difference threshold is not arbitrary, but based on in-depth analysis of a large amount of historical ocean data and professional domain knowledge; in the scenario of oil spill monitoring, different types of data have different normal fluctuation ranges. For example, for seawater temperature data, by analyzing the historical temperature data of this sea area over the past years and combining factors such as seasonal changes and day-night temperature differences, determine its normal temperature fluctuation range under normal circumstances, and so on.

[0052] Identify the data with differential content exceeding the difference threshold as abnormal data; it should be understood that after the multi-source data integration module extracts the differential data, the data mining unit compares this differential content with the preset difference threshold. Taking seawater salinity data as an example, if the seawater salinity data collected at a certain moment shows 35.5‰, and the normal salinity range of this sea area at the current season and current geographical location is determined through analysis to be 34‰ to 35‰, and the difference threshold is set at ±0.5‰, then the salinity data of 35.5‰ exceeds the difference threshold and will be identified as abnormal data. For various types of data, the data mining unit will use efficient data processing algorithms to quickly and accurately compare them one by one, screen out all the data exceeding the difference threshold, and determine them as abnormal data.

[0053] Use the identified abnormal data as input to establish a risk assessment model; The warning unit generates a warning message according to the risk assessment result.

[0054] It should be understood that using the identified abnormal data as input, the data mining unit starts to build a risk assessment model. When building the model, a variety of data analysis techniques and algorithms will be comprehensively used. For example, use the regression analysis algorithm to analyze the correlation between abnormal data and the risk of oil spill; assume that the abnormal increase in seawater temperature is related to the oil film formed after the oil spill hindering the heat dissipation of seawater. Through regression analysis, this relationship can be quantified, that is, for every 1℃ increase in seawater temperature, what is the probability of an increase in the risk of oil spill; At the same time, combine classification algorithms in machine learning, such as the support vector machine (SVM) algorithm, to classify abnormal data into different risk levels; according to historical cases and expert experience, divide the risk levels into three levels: low, medium, and high. For example, when the concentration of seawater pollutants increases abnormally and exceeds a certain multiple of the difference threshold, the model may determine it as a high-risk level; if it only slightly exceeds the difference threshold, it is determined as a low-risk level; by continuously optimizing the model parameters, it can more accurately evaluate the risk brought by oil spill according to abnormal data.

[0055] As a further embodiment, the decision support unit is used to provide decision support for the warning message, including but not limited to issuing navigation warnings, activating emergency plans, and adjusting operation plans.

[0056] It should be understood that after the early warning unit generates an early warning message based on the risk assessment result, the decision support unit actually issues an early warning according to the situation, for example, issues a navigation warning, decides whether to activate the emergency plan, and decides whether to adjust the operation plan during offshore oil exploration, maritime transportation and other operation activities, etc.

[0057] The above are only the preferred embodiments of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions within the idea of the present invention belong to the protection scope of the present invention. It should be noted that for those of ordinary skill in the art, several improvements and refinements made without departing from the principle of the present invention should also be regarded as within the protection scope of the present invention.

Claims

1. An intelligent mining and early warning decision-making system for multi-source data of marine drones, which is applied to the oil spill network, is characterized in that, It includes: A drone control module that controls the marine drone swarm to perform monitoring tasks according to a preset inspection route; A multi-source data collection module, which is distributed on each marine drone and is used to collect various types of marine data in the target sea area; A multi-source data integration module that integrates various types of marine data obtained by the marine drone swarm and extracts valuable information; A data mining unit that further analyzes the valuable information extracted by the multi-source data integration module, identifies abnormal data and conducts risk assessment; An early warning decision-making module, which includes an early warning unit and a decision support unit. The early warning unit generates early warning information based on the risk assessment of the data mining module, and the decision support unit is used to provide decision support for the early warning information.

2. The intelligent mining and early warning decision-making system for multi-source data of marine drones according to claim 1, characterized in that It also includes an oil spill identification module for marine areas. The specific work process is as follows: The target sea area is photographed by a hyperspectral camera carried by a marine drone, capturing the light reflection information of different wavelengths and generating image data containing rich spectral features; Based on the input of the image data into a trained convolutional neural network model, the convolutional layer, pooling layer, etc. in the model are used to extract and analyze the features of the image to distinguish oil spills from natural oil films and identify the type of oil product.

3. The intelligent mining and early warning decision-making system for multi-source data of marine drones according to claim 2, characterized in that, It also includes an oil source tracing module for petroleum. The specific working method is as follows: When the marine drone monitors the oil spill area, while recording the position information at the time of shooting, it collects the sea current data in this area, including sea current speed and direction information; Starting from the current position of the oil spill, based on the sea current data and the time series image information taken by the marine drone, the positions of the oil spill at different time points are calculated backward.

4. The intelligent mining and early warning decision-making system for multi-source data of marine drones according to claim 3, characterized in that, After the oil spill identification module for marine areas distinguishes that it is an oil spill, it includes: Obtain the viscosity of the leaked oil , seawater density , oil density , oil film thickness h, the wind speed u in the current sea area, oil film diameter d, the time elapsed after the oil film is formed , the current seawater temperature T, the current oil film area A, the initial oil film area , the current oil-water interfacial tension and the oil-water interfacial tension under standard conditions ; According to the formula , the oil film stability is obtained; The early warning unit makes an early warning decision based on the oil film stability S.

5. The intelligent mining and early warning decision-making system for multi-source data of marine drones according to claim 4, characterized in that, After the oil spill identification module for marine areas distinguishes that it is an oil spill, it also includes: Obtain the lowest sea water temperature observed in the historical records of the current sea area ; Obtain the range of seawater temperature changes observed in the historical records of the current sea area ; Obtaining the current sea current speed V; Obtaining the weathering degree F of the petroleum; Based on the formula the oil spill diffusion assessment index is obtained, where is the wind direction factor, with a value range of 0 - 1, is the maximum sea current speed of the historical distance in this sea area, which is used to normalize the sea current speed, is the maximum oil film thickness that is expected to occur in this type of oil spill scenario, which is used to normalize the oil film thickness; The early warning unit predicts the oil spill situation based on the oil spill assessment indicators ​ 6. The intelligent mining and early warning decision-making system for multi-source data of marine drones according to claim 5, characterized in that, The range of seawater temperature changes observed in the historical records of the current sea area is calculated as follows: By obtain the range of seawater temperature changes observed in the historical records of the current sea area , where is the highest seawater temperature observed in the historical records of the current sea area.

7. The intelligent mining and early warning decision-making system for multi-source data of marine drones according to claim 6, characterized in that The early warning unit is also used to divide the early warning levels, specifically: Taking the large petroleum diffusion comprehensive index I and the strong stability S of the oil film as the first early warning level; Taking the large petroleum diffusion comprehensive index I and the poor stability S of the oil film as the second early warning level; Taking the small petroleum diffusion comprehensive index I and the strong stability S of the oil film as the third early warning level; Taking the small petroleum diffusion comprehensive index I and the poor stability S of the oil film as the fourth early warning level; Among them, the first early warning level > the second early warning level > the third early warning level > the fourth early warning level.

8. The intelligent mining and early warning decision-making system for multi-source data of marine drones according to claim 7, characterized in that, The specific steps for the multi-source data integration module to extract valuable information are as follows: Obtaining the data collected by the marine drone swarm from different angles and methods for the marine environment; Comparing the collected data with the data in the pre-constructed database one by one; During the comparison process, according to the set difference criteria, screening out the data with obvious differences; Deeply analyzing the screened data with differences, extracting the differential content presented in the data, and determining it as valuable information.

9. The intelligent mining and early warning decision-making system for multi-source data of marine drones according to claim 8, wherein The specific method for the data mining unit to identify abnormal data and conduct risk assessment is as follows: Presetting a difference threshold; Identifying the data with differential content exceeding the difference threshold as abnormal data; Taking the identified abnormal data as input to establish a risk assessment model; The early warning unit generates early warning information based on the risk assessment results.

10. The intelligent mining and early warning decision-making system for multi-source data of marine drones according to claim 9, characterized in that, The decision support unit is used to provide decision support for the early warning information, including but not limited to issuing navigation warnings, activating emergency response plans, and adjusting operation plans.

Images