An intelligent identification and emergency disposal system for offshore oil spill

Abstract

The application provides an offshore oil spill intelligent identification and emergency disposal system, and relates to the technical field of marine pollution prevention.The system comprises: an intelligent identification module that acquires multi-source images, extracts features by using a segmentation network model and a color space segmentation algorithm, and obtains oil spill identification data; an intelligent decision module that obtains risk assessment data by analyzing the oil spill identification data, and determines an emergency disposal scheme and equipment dispatching control instructions; an emergency disposal module that dispatches disposal equipment, obtains operating state data to obtain a compensation coefficient, and adaptively adjusts working parameters in combination with the oil spill identification data; and a feedback optimization module that generates a review data set by using disposal effect data to iteratively update the segmentation network model and the color space segmentation algorithm.The application solves the technical defects of low identification accuracy, delayed response and poor coordination of the prior art, and greatly improves the oil spill recovery efficiency under complex sea conditions.

Description

Technical Field

[0001] This invention relates to the field of marine pollution prevention technology, and in particular to an intelligent oil spill identification and emergency response system. Background Technology

[0002] With the increasing frequency of global offshore oil and gas development and maritime oil transportation activities, marine oil spills occur frequently. Marine oil spills are characterized by rapid spread, wide impact, high difficulty in handling, and stringent technical requirements for emergency response. Achieving rapid and accurate identification and efficient emergency response to oil spills is an urgent need in the field of marine environmental protection.

[0003] However, existing marine oil spill monitoring and mitigation technologies still have the following technical shortcomings: Firstly, existing oil spill identification schemes mostly rely on single radar or spectral images for feature extraction, resulting in poor adaptability to changes in lighting, complex sea conditions, and weather conditions. This makes them prone to false and missed identifications, and makes it difficult to extract key characterization parameters such as oil spill area, oil film thickness, and diffusion evolution trend with high precision.

[0004] Secondly, the existing monitoring system and emergency response system are often independent of each other. Traditional satellite monitoring methods have a high data transmission delay, which makes it impossible for ships and aircraft at the front end of the on-site response to obtain real-time dynamic evolution data of the oil spill. This results in a serious delay in emergency response and makes it very easy to miss the best opportunity for physical containment and chemical treatment.

[0005] Third, existing emergency response equipment mostly uses one-way communication links, resulting in communication bottlenecks such as high command transmission latency and high data packet loss rate. Furthermore, the execution terminals lack a unified collaborative scheduling mechanism. In addition, the operating parameters of core response equipment are highly dependent on manual experience for adjustment, leading to poor scenario adaptability and extremely low oil recovery efficiency in complex conditions such as thin oil layers.

[0006] Fourth, existing pollution control systems are mostly limited to a single monitoring or single salvage stage, and have not yet built a closed-loop management network covering the entire process of identification, early warning, decision-making, disposal and status feedback. This makes it impossible for established emergency response plans to be evaluated in real time and optimized iteratively based on the actual hydrological conditions and oil film residue data on site.

[0007] Fifth, conventional oil booms and other basic physical containment equipment lack intelligent adjustment and control units. Under the influence of complex ocean currents and alternating wave loads, they are prone to structural deformation or overall displacement, making it difficult to effectively adapt to the dynamic diffusion requirements of oil spill containment. Summary of the Invention

[0008] To overcome the shortcomings of existing technologies, the purpose of this invention is to provide an intelligent identification and emergency response system for marine oil spills. This invention solves the technical problems of existing technologies, such as low accuracy of single-source image recognition under complex sea conditions, delayed response due to fragmented data transmission between monitoring and response stages, poor equipment coordination and adaptive adjustment capabilities of operating parameters, and lack of intelligent closed-loop management and optimization throughout the entire process.

[0009] To achieve the above objectives, the present invention provides the following solution: A marine oil spill intelligent identification and emergency response system includes: The data transmission module and the intelligent identification module, intelligent decision-making module, emergency response module, and feedback optimization module, all connected to the data transmission module; The data transmission module provides a communication link for data transmission to the intelligent identification module, the intelligent decision-making module, the emergency response module, and the feedback optimization module. The intelligent identification module acquires multi-source images of the monitoring area, extracts features from the multi-source images using a segmentation network model and a color space segmentation algorithm to obtain an oil spill parameter set, and determines corresponding oil spill identification data based on the oil spill parameter set. The intelligent decision-making module performs model analysis based on the oil spill identification data to obtain risk assessment data, and determines corresponding emergency response plans and equipment scheduling control instructions based on the risk assessment data. The emergency response module schedules corresponding response equipment according to the equipment scheduling control instructions, obtains compensation coefficients from the operating status data of the response equipment, and adaptively adjusts the operating parameters of the response equipment based on the compensation coefficients and the oil spill identification data. The feedback optimization module summarizes and analyzes the acquired oil spill area response effect data, the oil spill identification data, and the emergency response plan to obtain a review dataset, and iteratively updates the segmentation network model and the color space segmentation algorithm based on the review dataset.

[0010] A method for intelligent identification and emergency response to marine oil spills includes: Multi-source images of the monitoring area are acquired, and feature extraction is performed on the multi-source images using a segmentation network model and a color space segmentation algorithm to obtain an oil spill parameter set. The corresponding oil spill identification data is then determined based on the oil spill parameter set. Based on the oil spill identification data, model analysis is performed to obtain risk assessment data, and corresponding emergency response plans and equipment scheduling and control instructions are determined based on the risk assessment data. The corresponding disposal equipment is scheduled according to the equipment scheduling and control command, and the operating status data of the disposal equipment is obtained to obtain the compensation coefficient. The operating parameters of the disposal equipment are adaptively adjusted according to the compensation coefficient and the oil spill identification data. The acquired data on the effectiveness of the oil spill response, the oil spill identification data, and the emergency response plan are summarized and analyzed to obtain a retrospective dataset. Based on the retrospective dataset, the segmentation network model and the color space segmentation algorithm are iteratively updated.

[0011] The present invention discloses the following technical effects: This invention provides an intelligent oil spill identification and emergency response system. Firstly, it overcomes the poor environmental adaptability of single-source images by combining adaptive acquisition of multi-source images with a bilateral segmentation network and color space algorithms, significantly improving oil spill identification accuracy. Secondly, it breaks down the data gap between monitoring and response by utilizing LoRa and multi-mode encrypted communication, effectively shortening emergency response time. Thirdly, it intelligently generates scheduling schemes based on risk assessment and optimization algorithms, and adaptively adjusts the operating parameters of recovery equipment by combining ship roll compensation, solving the problems of low equipment coordination efficiency and poor adaptability of manual adjustments. Finally, it constructs a data review and feedback mechanism, iterating the identification model and decision-making algorithm in real time based on the response results, achieving closed-loop management of the entire oil spill early warning and response process and maximizing pollution prevention efficiency. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This is a schematic diagram of a marine oil spill intelligent identification and emergency response system provided in an embodiment of the present invention.

[0014] Figure label: 1-Data transmission module, 2-Intelligent identification module, 3-Intelligent decision-making module, 4-Emergency response module, 5-Feedback optimization module. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0017] like Figure 1 As shown, the present invention provides an intelligent oil spill identification and emergency response system, comprising: The data transmission module 1 and the intelligent identification module 2, intelligent decision-making module 3, emergency response module 4 and feedback optimization module 5, all of which are connected to the data transmission module 1; The data transmission module 1 provides a communication link for data transmission to the intelligent identification module 2, the intelligent decision-making module 3, the emergency response module 4, and the feedback optimization module 5. The intelligent identification module 2 acquires multi-source images of the monitoring area, extracts features from the multi-source images using a segmentation network model and a color space segmentation algorithm, obtains an oil spill parameter set, and determines corresponding oil spill identification data based on the oil spill parameter set. The intelligent decision-making module 3 performs model analysis based on the oil spill identification data to obtain risk assessment data, and determines corresponding emergency response plans and equipment scheduling control instructions based on the risk assessment data. The emergency response module 4 schedules corresponding response equipment according to the equipment scheduling control instructions, obtains compensation coefficients from the operating status data of the response equipment, and adaptively adjusts the operating parameters of the response equipment based on the compensation coefficients and the oil spill identification data. The feedback optimization module 5 summarizes and analyzes the acquired oil spill area response effect data, the oil spill identification data, and the emergency response plan to obtain a review dataset, and iteratively updates the segmentation network model and the color space segmentation algorithm based on the review dataset.

[0018] Specifically, this embodiment discloses the following system workflow: After the system starts up and enters real-time monitoring mode, the intelligent identification module 2 first acquires multi-source images of the monitoring area and activates its internal feature processing engine. Subsequently, the intelligent identification module 2 rigorously extracts features from the multi-source images using a segmentation network model and a color space segmentation algorithm, accurately separating the sea surface background and target oil slick features to obtain a set of oil spill parameters containing core physical characteristics. Based on this set of parameters, the corresponding oil spill identification data is determined. Throughout this process, the data transmission module 1 continuously provides the underlying communication link for the intelligent identification module 2 and other related modules, ensuring that the oil spill identification data can be transmitted securely and without delay.

[0019] After completing front-end identification and data transmission, the intelligent decision-making module 3 receives the oil spill identification data and performs situational assessment. Based on the oil spill identification data, the intelligent decision-making module 3 initiates model analysis, obtaining risk assessment data for the current sea conditions through comprehensive calculation of key indicators such as oil spill area, thickness, and diffusion trend. Subsequently, based on the risk assessment data, and combined with a preset knowledge base and path planning optimization mechanism, the intelligent decision-making module 3 generates an overall control strategy for the current oil spill incident, thereby determining the corresponding emergency response plan and equipment scheduling and control instructions for commanding the front-end hardware cluster.

[0020] Upon receiving the issued instructions, the front-end emergency response module 4 strictly follows the equipment scheduling and control instructions to dispatch the corresponding response equipment to the oil spill area to perform physical containment or recovery operations. During dynamic operations, the emergency response module 4 continuously acquires the operating status data of the response equipment and calculates compensation coefficients to offset environmental interference. Based on these compensation coefficients and the oil spill identification data, the module adaptively adjusts the operating parameters of the response equipment to ensure recovery efficiency. After the on-site response task is completed, the feedback optimization module 5 intervenes, performing a global summary analysis of the acquired oil spill area response effect data, the initial oil spill identification data, and the executed emergency response plan to obtain a review dataset. Finally, based on this review dataset, the underlying segmentation network model and color space segmentation algorithm are iteratively updated to achieve closed-loop self-evolution of the system's recognition accuracy and decision-making capabilities.

[0021] Furthermore, the specific working process of the intelligent recognition module 2 is as follows: For the reference feature acquisition unit, its core function is to collect the illumination feature information of the monitoring area and determine the matching image acquisition type accordingly. In the data processing of the ambient illuminance index, this embodiment uses direct illuminance as the basic variable, multiplying it by the negative exponent of the product of the sea surface scattering coefficient (based on the natural constant) and the equipment flight altitude, and then multiplying it by the cosine of the solar zenith angle to obtain the ambient illuminance index. The direct illuminance in the above data processing comes from the light sensor, and its value is usually between 0 and 1000 lux, which represents the absolute intensity of natural light sources; the sea surface scattering coefficient is a custom parameter that represents the degree of diffuse reflection attenuation of light from the sea surface, which comes from the historical mapping database. For example, its specific value can be set to 0.02 under micro-wave conditions; the equipment flight altitude is fed back in real time by the altimeter and is used to correct the atmospheric penetration thickness, with a typical value of 50 meters; the solar zenith angle comes from the ephemeris of the navigation and positioning system and is used to calculate the incident angle of light. In this embodiment, the calculated ambient light illuminance index is compared with the preset core illuminance threshold of 500 lux. If the index is greater than or equal to 500 lux, the image acquisition type is determined to be visible light acquisition mode; otherwise, it is determined to be infrared thermal imaging acquisition mode.

[0022] In this embodiment, the multi-source image acquisition unit dynamically acquires multi-source images of the monitoring area based on the aforementioned determined image acquisition type. When the visible light acquisition mode is determined, the multi-source image acquisition unit prioritizes activating the high-resolution visible light digital camera mounted on the cruise equipment; when the infrared thermal imaging mode is determined, it seamlessly switches and activates the infrared thermal imaging sensor matrix. During the acquisition process, this embodiment controls the cruise equipment to perform scanning flight along a preset zigzag path. To ensure the overlap rate and spatial resolution of the multi-source images, the flight altitude of the equipment is strictly controlled within a range of 1 meter above and below 50 meters, the coverage width of a single-sided scan is precisely set to 100 meters, and the sampling frequency of the image signal is kept constant at 1 Hz. Through the joint collaboration of airborne inspection equipment, space-based remote sensing equipment, and sea-based fixed monitoring stations, this embodiment can acquire high-quality multi-source images with comprehensive coverage and no limitations imposed by sudden environmental changes, providing a solid foundation of raw data for subsequent identification and analysis.

[0023] In this embodiment, the image preprocessing unit is responsible for performing contrast enhancement, denoising, and correction processing on the initial pixels in the multi-source images to generate a preprocessed image set. During the adaptive enhancement and denoising data processing, when calculating the preprocessed pixel value at a certain coordinate, this embodiment first divides the initial pixel value corresponding to that coordinate by the sum of the local window pixel mean and the minimum positive constant, and then performs an exponential operation on the quotient using the adaptive contrast adjustment factor to obtain an enhancement coefficient. Subsequently, the initial pixel value is multiplied by this enhancement coefficient to obtain an enhancement term. Next, the absolute value of the initial pixel gradient is divided by the edge preservation coefficient and squared, then the reciprocal of the result (after adding a constant 1) is taken as the preservation weight. This preservation weight is then multiplied by the initial pixel gradient and the spatial divergence is calculated to obtain a smoothing term. Finally, the enhancement term and the smoothing term are added to obtain the final preprocessed pixel value. The local window pixel mean is calculated from the arithmetic mean of a 3x3 pixel matrix centered on the target pixel, used to characterize the local brightness background; the minimum positive constant is set to 0.001, which prevents memory overflow due to a zero denominator during data processing; the adaptive contrast adjustment factor is a custom parameter that controls the degree of image contrast stretching, with a value ranging from 0.5 to 1.5. For example, for a highly reflective image with severe sea surface flares, it can be set to 1.2 to suppress highlights; the edge preservation coefficient controls the smoothness to prevent blurring of oil film boundaries, and can be set to 10 for example. The preprocessed pixels are rearranged and recombined, resulting in a preprocessed image set with sharp contrast and extremely low noise.

[0024] In this embodiment, after acquiring the preprocessed image set, the intelligent recognition unit converts it from the three-primary-color model to a color space constructed from hue, saturation, and brightness to extract target feature points and uses a color space segmentation algorithm to generate a continuous oil film region. During the data processing of the spatial cone distance measurement, this embodiment determines whether the target feature point belongs to the oil spill feature by calculating the color space distance. The specific calculation logic is as follows: the product of the cosine values ​​of the brightness, saturation, and hue components of the target feature point is subtracted from the product of the cosine values ​​of the reference brightness, reference saturation, and reference hue of the reference oil spill feature; the square of the difference yields the first spatial difference. Similarly, the product of the sine values ​​of the brightness, saturation, and hue of the target feature point is subtracted from the product of the corresponding reference feature values; the square of this subtraction yields the second spatial difference. The square of the difference between the brightness component of the target feature point and the reference brightness yields the third spatial difference. Finally, the first, second, and third spatial differences are added together and the square root is taken to obtain the color space distance. The baseline hue, baseline saturation, and baseline brightness are derived from a pre-trained offline feature library of marine oil spills. For example, for heavy crude oil pollution, the baseline hue value could be 30, the baseline saturation value 0.4, and the baseline brightness value 0.5. This embodiment compares the calculated color space distance with a preset distance threshold of 15. If the distance is less than 15, it is determined to be an oil film feature point. Finally, all oil film feature points that pass the determination are used for connectivity repair using a morphological closing operation with a kernel size of 5x5, eliminating internal voids and isolated noise points, thereby generating a complete and continuous oil film region.

[0025] In this embodiment, the intelligent identification unit further utilizes a segmentation network model to extract targets from the generated continuous oil film area, obtaining an oil spill parameter set including the oil spill area and thickness, and ultimately establishing the oil spill identification data. In the parameter parsing of the oil spill area, this embodiment counts the total number of feature pixels contained in the closed boundary within the continuous oil film area and multiplies it by the spatial resolution base. This spatial resolution base is derived from the calibration mapping between the device's flight altitude and the camera's physical focal length. For example, at the current cruising altitude, this base is 0.1 square meters per pixel, thereby calculating the absolute oil spill area. In the extraction of the thickness parameter, the infrared radiation attenuation model is used to perform feature inversion calculation on the infrared absorption rate of the continuous oil film area to obtain the quantized thickness. Subsequently, this embodiment initiates a spatial coordinate mapping mechanism for data conversion, mapping the data in the oil spill parameter set based on the two-dimensional coordinate position of the image sensor to an absolute geographic coordinate system represented by the national geodetic coordinate system through a high-dimensional affine transformation matrix and combined with the latitude and longitude reference points pushed in real time by the inertial navigation system. Through the multi-dimensional mapping and parameter coupling from the pixel level to the geographic level described above, this embodiment outputs oil spill identification data that includes the absolute geographic location of the oil spill, the expanded area, and the real-time thickness. After actual testing and verification in a complex sea environment, the accuracy of oil spill feature boundary identification under this data processing logic reached 98%, providing highly reliable data support for subsequent intelligent decision-making and emergency response.

[0026] Specifically, the expression for the segmentation network model is: ; in, Output probability for segmentation mask; (.) represents the sigmoid activation function; (.) represents the spatial detail feature mapping function; Input matrix for continuous oil film region; For feature fusion operators; (.) represents the global semantic context feature mapping function; In this embodiment, to further clarify the data processing logic of the underlying algorithm, the intelligent recognition unit uses a two-sided structure to perform deep feature extraction on the boundary details and global semantics of the oil film. The specific data processing process of the segmentation network model is as follows: The intelligent recognition unit first inputs the continuous oil film region input matrix in parallel to two feature mapping calculation branches. In the first calculation branch, the spatial detail feature mapping function is used to perform a convolution operation on the continuous oil film region input matrix to obtain a spatial feature map. In the second calculation branch, the global semantic context feature mapping function is used to perform a convolution operation on the same continuous oil film region input matrix to obtain a semantic feature map. Subsequently, the spatial feature map and the semantic feature map are superimposed using a feature fusion operator to obtain a fused feature tensor. Finally, the sigmoid activation function is used to perform a nonlinear mapping calculation on the fused feature tensor to obtain the final segmentation mask output probability.

[0027] Among them, the segmentation mask output probability is used to characterize the confidence level of each pixel in the input matrix belonging to the final oil spill target; the sigmoid activation function is a computational logic that uses the exponential operation logic of the natural constant base to nonlinearly compress and map the real number tensor to the zero-to-one standard probability distribution space; the spatial detail feature mapping function is a computational logic with a shallow high-frequency two-dimensional convolution kernel configured inside, used to extract boundary contours and local texture gradient features; the continuous oil film region input matrix is ​​a two-dimensional pixel tensor input source as the underlying visual feature; the feature fusion operator is a computational logic that achieves lossless feature superposition through element-wise matrix addition; the global semantic context feature mapping function is a computational logic that uses a deep convolution branch with a dilation rate to obtain a large-scale receptive field, used to extract global environmental semantic features spanning long-distance pixels; Furthermore, the specific working process of the intelligent decision-making module 3 is as follows: In this embodiment, the intelligent decision-making module 3, acting as the system's central hub, is responsible for coordinating comprehensive monitoring data and intelligently generating emergency response strategies. Specifically, the data integration unit receives and integrates the oil spill identification data transmitted by the preceding modules. In practice, the data integration unit receives not only basic identification information such as the location, area, thickness, total amount, and type of the oil spill, but also synchronously accesses fluid dynamics-related environmental parameters (such as ocean currents, wind speed, tides, and illumination) and equipment status data from various front-end hardware (the aforementioned environmental parameters and equipment status data are packaged and transmitted together with the basic identification information in the system's underlying logic and used to assist decision-making, i.e., the oil spill identification data). During data processing, this embodiment uses a built-in data cleaning algorithm to remove outliers and perform spatiotemporal registration on the received multidimensional concurrent data, thereby structuring it to obtain a unified data matrix, providing a standardized and clean data foundation for subsequent risk assessment.

[0028] In this embodiment, the risk assessment unit is used to conduct in-depth disaster situation analysis based on the aforementioned generated data matrix. During execution, this unit constructs a risk scoring model based on the data matrix. This model combines statistical methods and machine learning algorithms to predict oil spill diffusion trends. This embodiment extrapolates the oil slick drift trajectory and assesses the spatial threat level of the calculated risk level (set to low, medium, and high levels within the system) against environmentally sensitive resources identified around the accident site (such as aquaculture areas, marine protected areas, ports, etc.). Through the above data processing, the system generates the risk assessment data, including specific early warning signal triggering mechanisms and threat quantification indicators, within an extremely low latency of one minute.

[0029] In this embodiment, the solution generation unit is used to formulate targeted multi-dimensional response strategies based on the risk assessment data. This unit is equipped with a comprehensive oil spill emergency response knowledge base, covering response methods and parameter settings for different oil spill types and environmental conditions. In the data processing stage of solution generation, this embodiment uses the risk assessment data and an improved Dijkstra algorithm with a specific cost function (i.e., the Dijkstra algorithm described in the claims) to perform optimal path planning. Specifically, when optimizing the path, the algorithm incorporates multi-dimensional environmental interference factors into its cost function. Through the precise calculation of this algorithm, the solution generation unit can automatically plan the optimal path that ensures the fastest possible arrival of response equipment at the site, thereby determining the emergency response plan, which includes dimensions such as oil boom deployment, equipment scheduling, and dispersant spraying.

[0030] In this embodiment, the instruction issuing unit serves as the core output port of the decision-making results, used to convert the generated emergency response plan into standard control instructions. In the instruction issuing implementation logic, this unit is internally configured with an instruction compilation and conversion engine, which decomposes the macro-level emergency response plan into underlying machine action codes that can be directly identified and parsed by different types of front-end hardware (such as unmanned deployment vessels, oil spill recovery vessels, automatic spraying devices, etc.), determining the corresponding equipment scheduling and control instructions. Subsequently, these equipment scheduling and control instructions are transmitted in real-time and accurately to each designated hardware node of the emergency response module 4 via the encrypted link of the data transmission module 1, and simultaneously mapped to the remote monitoring terminal of the emergency command center. This achieves seamless connection and efficient control throughout the entire process from data perception and algorithmic decision-making to physical execution, while ensuring data security.

[0031] Specifically, the expression for the risk scoring model is: ; in, These are risk assessment indicator values; Vulnerability index for environmentally sensitive areas; This represents the cumulative total of the oil spill. The sea surface diffusion damping coefficient; This represents the initial oil spill diffusion area; The magnitude of the composite drift velocity vector; To predict the time step; Furthermore, the specific working process of the data transmission module 1 is as follows: In this embodiment, the data transmission module 1 acts as the neural network for information interaction across the entire system, responsible for achieving low-latency, high-reliability data relay among distributed heterogeneous nodes. The wireless transmission unit is used to establish robust, seamless connections in complex maritime environments. In specific implementation, this unit deeply integrates a low-power local area network wireless standard (LoRa) communication link (e.g., using an E32-900t hardware module) with 5G networks and satellite communication technology (wireless communication technology) to construct a multi-mode communication network spanning short-range and ocean-going areas. The LoRa communication link primarily handles high-frequency, short-range data transmission between near-field terminals such as UAV inspection equipment, response vessels, and surrounding equipment, with a communication cycle strictly set to 200 milliseconds to ensure instantaneous response to front-end collaborative commands. Meanwhile, the 5G network and satellite communication technology overcome the problem of insufficient communication coverage in remote maritime areas, transmitting large-scale, long-distance images and sensing data back across the sea to the rear command center. Through gateway routing and seamless protocol conversion of the aforementioned heterogeneous communication protocols, the wireless transmission unit successfully constructs the all-weather interconnected wireless transmission link.

[0032] In this embodiment, for core fixed nodes with extremely high anti-interference and confidentiality requirements, the wired transmission unit is used to construct a wired network for key areas. When deploying data transmission hardware, for infrastructure such as emergency command centers and offshore fixed monitoring stations located in key sea areas (such as ports and around oil and gas drilling platforms) (i.e., the key areas in the claims), this embodiment lays high-bandwidth wired communication media to establish physical hard connections. By introducing a wired network channel, this unit not only effectively avoids the spatial attenuation and electromagnetic interference effects of extreme weather conditions at sea on radio frequency signals, greatly improving the security threshold and network throughput rate of core control signaling and situational awareness data interaction between core nodes, but also obtains the wired transmission link with extremely strong physical resilience.

[0033] In this embodiment, given that marine oil spill disaster data involves sensitive information about the marine environment and shipping safety, the data encryption unit is used to perform low-level security hardening on data packets across the entire link. In terms of data processing logic, this unit uses the AES encryption algorithm (i.e., the Advanced Encryption Standard algorithm in the claims) under a symmetric key cryptography system to dynamically encrypt various heterogeneous data such as real-time transmitted image data streams, sensor message sequences, and low-level operation commands. During this encryption process, complex matrix transformation operations such as multiple rounds of byte substitution, row shifting, column obfuscation, and round key addition are used to prevent data from being maliciously tampered with or stolen in public networks or open wireless frequency bands, thereby generating high-fidelity, eavesdropping-proof encrypted communication data. Finally, the underlying communication middleware of the system works collaboratively, transparently mapping and dynamically mounting the encrypted communication data onto the aforementioned established wireless and wired transmission links based on the payload characteristics of the encrypted communication data, integrating it into a multi-mode security network with automatic fault self-healing and multi-link redundancy switching capabilities, thus serving as the highly available communication link for the system. Based on this multi-mode security mechanism, the system successfully completed the stable and extremely low-latency data transmission task between the intelligent identification module 2, the intelligent decision-making module 3, the emergency response module 4, and the feedback optimization module 5, thus constructing an unbreakable data flow closed loop.

[0034] Furthermore, the specific working process of the emergency response module 4 is as follows: In this embodiment, the emergency response module 4, acting as the execution hub of the front-end hardware, is responsible for translating back-end decision-making instructions into physical response actions. Specifically, the oil spill containment control unit is used to control the unmanned deployment vessel to quickly deploy an intelligent oil spill containment boom equipped with omnidirectional balls and mesh composite materials to the designated oil spill area according to the received equipment scheduling and control instructions. During the dynamic monitoring of oil spill containment, this embodiment uses an ultrasonic monitoring unit to measure the time difference between sound wave transmission and reception to obtain containment status data. In the data processing of the distance calculation, this embodiment multiplies the sound wave transmission time difference obtained by the ultrasonic sensor hardware timer by a preset air speed calibration value, and then divides the product by a constant 2 to accurately calculate the absolute distance from the sensor probe to the actual oil surface. The air speed calibration value is an environmental constant parameter, and its typical value is set to 340 meters per second. To ensure the real-time performance and penetration of the monitoring, this embodiment sets the ultrasonic transmission frequency to 40 kHz and the distance calculation refresh period to 100 milliseconds. Based on the containment status data including the real-time absolute oil surface distance obtained from the above calculation, this embodiment automatically drives the unmanned deployment vessel to fine-tune its course, so that the intelligent oil boom with a total length of 500 meters always maintains the best interception posture, thereby achieving efficient oil spill containment based on the containment status data.

[0035] In this embodiment, the chemical treatment unit is used to control an automatic spraying device to spray environmentally friendly oil spill dispersants for specific sea conditions such as small areas or thin oil films, according to the equipment scheduling and control instructions. During the data processing of the spraying control, this embodiment calculates the optimal oil spill dispersant spray flow rate and diffusion radius parameters based on the oil spill area and oil film thickness values ​​transmitted from the preceding module, using a built-in chemical reaction ratio mapping table. This flow rate parameter is then converted into a pulse width modulation signal for the spray pump valve opening, thus obtaining spraying status data including the actual spray flow rate and pipeline pressure. To avoid secondary marine pollution caused by over-spraying, this embodiment strictly limits the maximum spray pressure threshold to 0.5 MPa and precisely controls the oil spill dispersant atomization particle size to 50 micrometers to ensure sufficient contact and reaction between the chemical agent and the oil film. Based on the real-time feedback spraying status data, this embodiment performs closed-loop fine-tuning of the nozzle working matrix, thereby safely and environmentally supporting the decomposition of oil spills.

[0036] In this embodiment, the oil spill recovery unit serves as the core pollution removal node, used to schedule the oil spill recovery vessel and its onboard skimming device, acting as the disposal equipment, for oil spill recovery according to the equipment scheduling control command. During the recovery operation, this embodiment acquires real-time oil film thickness data through an infrared oil film thickness sensor and adaptively adjusts the basic suction plate speed of the skimming device in segments accordingly. In the data processing of the basic speed mapping, when the infrared sensor detects an oil film thickness less than or equal to 2 mm, this embodiment maintains a constant output of 200 rpm for the basic suction plate speed; when the oil film thickness is greater than 2 mm but less than or equal to 5 mm, this embodiment calculates the difference between the oil film thickness and the constant 2, multiplies this difference by the speed increment coefficient, and adds the product to the basic value of 200 rpm, using this calculated value as the current basic suction plate speed; when the oil film thickness is greater than 5 mm, this embodiment locks the basic suction plate speed at a full-load 350 rpm. The rotational speed increment coefficient is a custom parameter that characterizes the weight of the thickness on the rotational speed, with a fixed value of 50 revolutions per minute per millimeter. The aforementioned oil film thickness is obtained from an infrared oil film thickness sensor with a sampling frequency of 1000 Hz, whose measurement range covers 0 to 10 millimeters and whose measurement accuracy reaches ±0.5 millimeters, providing a high-frequency data source for precise adjustment of rotational speed.

[0037] After determining the base rotational speed, to overcome the interference of complex sea conditions on recovery efficiency, this embodiment further obtains the compensation coefficient by acquiring the operating status data of the disposal equipment and performs dynamic correction. In the data processing of roll compensation, this embodiment acquires real-time operating status data including the ship's roll angle using a ship attitude sensor, smooths and filters noise from the roll angle time series using a Kalman filter algorithm, and obtains the absolute roll angle value. Subsequently, this embodiment multiplies the absolute roll angle value by a first angle compensation factor and multiplies the aforementioned oil film thickness data by a second thickness compensation factor. Finally, these two products are added to a constant 1 to calculate the final roll compensation coefficient. In this calculation logic, the ship's roll angle is derived from a high-precision attitude sensor with a measurement range of ±15 degrees and an accuracy of ±0.2 degrees; the first angle compensation factor is a custom environmental disturbance immunity parameter used to characterize the degree of weakening of oil absorption efficiency by wave swaying, and in this embodiment, it is fixed at 0.05; the second thickness compensation factor is used to characterize the compensation gain of the oil layer's own viscous resistance, and its value is 0.02. Finally, in this embodiment, the operating parameters of the treatment equipment are adaptively adjusted by multiplying the compensation coefficient by the rotational speed of the basic oil suction plate, ensuring that the rotational speed adjustment error is always less than or equal to 5% when the ship's roll angle is within ±10 degrees. Simultaneously, this embodiment coordinates the operation of asphalt powder spraying equipment to spray asphalt powder with a particle size of less than 10 mm, causing the spilled oil to solidify into easily salvaged solid blocks.

[0038] In this embodiment, the equipment management unit is used to monitor the underlying hardware status of the disposal equipment in real time throughout the entire emergency response lifecycle. Specifically, this unit continuously collects underlying status characteristics such as operating current, hydraulic network pressure, and communication handshake heartbeat packets from the aforementioned unmanned deployment vessel, oil spill recovery vessel, and automatic spraying device via an industrial fieldbus with a polling cycle of 500 milliseconds. During the data processing for fault determination and command generation, this embodiment compares the real-time collected operating current and hydraulic network pressure data with the upper and lower limits of the rated safety threshold specified on the equipment nameplate using a sliding window. When the difference exceeds 15% of the rated safety threshold for three consecutive polling cycles, this embodiment immediately determines that the hardware node is in an abnormal state, thereby obtaining fault warning information including the equipment physical number and fault code. Based on the fault warning information, this embodiment automatically searches for idle devices of the same type in the background standby resource pool, and uses a preset instruction compilation template to generate corresponding backup scheduling instructions, ensuring that the backup disposal equipment completes hot start and task takeover within an extremely short delay of 3 seconds, thereby ensuring high reliability throughout the entire oil spill emergency disposal process and reducing equipment failure rate by more than 25%.

[0039] Specifically, the oil containment and source control unit, serving as the first line of defense against pollution spread, comprises an unmanned deployment vessel, an ultrasonic monitoring unit, and a smart oil boom. In practice, the unmanned deployment vessel, based on the equipment scheduling and control commands issued by the upstream intelligent decision-making module 3, quickly enters the target sea area and rapidly deploys the smart oil boom. To suit harsh maritime conditions, the smart oil boom is made of corrosion-resistant and wave-resistant mesh composite material. Its skirt has small holes and is covered with multiple layers of mesh material. The bottom is equipped with a load-bearing component to stabilize the center of gravity, and omnidirectional balls are fitted at the nodes to flexibly adjust the containment range. After deployment, the system obtains the initial containment state. During the dynamic containment data processing, the ultrasonic monitoring unit continuously emits 40 kHz detection waves based on the initial containment state, measuring the time difference between sound wave transmission and reception. In the specific calculation logic of distance resolution, this embodiment multiplies the time difference by the air speed constant and then divides the product by the reference denominator to accurately calculate the absolute oil surface distance from the probe to the actual oil surface. The air speed constant, as an environmental constant parameter, is set to 340 meters per second, serving as a reference for the propagation speed of sound waves in the air medium; the reference denominator is a round-trip distance conversion constant, fixed at 2. Based on the aforementioned high-frequency refreshed oil surface distance data, this embodiment determines the dynamic adjustment position command and then, by controlling the movement of the omnidirectional ball, precisely adjusts the intelligent oil boom, with a total length extended to 500 meters, to the target position, completing the oil spill containment through physical isolation.

[0040] In this embodiment, after initial containment is completed, the oil spill recovery unit intervenes as the core execution matrix for pollution cleanup, integrating an oil spill recovery vessel, oil suction device, oil skimming device, and asphalt powder spraying equipment. In terms of physical space scheduling, the oil spill recovery vessel is used to reach the oil spill area at full speed along the optimal path defined by the aforementioned dynamic planning, serving as the main carrier of the complete recovery hardware. To ensure the operational stability of the recovery device under adverse sea conditions, this embodiment uses a high-precision attitude sensor mounted on the center of gravity of the oil spill recovery vessel to acquire the ship's roll angle in real time at a sampling rate of 100 Hz as the operational status data. During the roll compensation data processing, this embodiment combines a Kalman filter to perform state estimation and optimal filtering smoothing on the ship's roll angle time series containing high-frequency wave interference noise, extracting the absolute roll angle value. Subsequently, this embodiment multiplies the absolute roll angle value by a roll attenuation gain factor, and then sums the product with a reference constant to rigorously obtain the compensation coefficient. Among them, the sway attenuation gain factor is a custom parameter that characterizes the torsional stiffness of the mechanical structure. Its value is 0.08. Its function is to suppress drastic angle changes and avoid oscillations in the control system. The reference constant is fixed at 1. Its function is to ensure that the compensation coefficient remains neutral and does not decay unnecessarily under absolutely stable conditions without wind and waves.

[0041] In this embodiment, based on the obtained anti-interference compensation parameters, the skimming device is used in conjunction with the suction device to perform negative pressure stripping-type oil-water mixture recovery. During the data processing implementation of the core parameter adaptive adjustment, this embodiment uses the oil spill identification data (specifically, the absolute oil film thickness value retrieved by the microwave or infrared module) transmitted in the preceding steps and the calculated compensation coefficient as the working parameter for adaptive adjustment. The specific calculation logic is as follows: This embodiment first extracts the absolute oil film thickness value from the oil spill identification data, multiplies this absolute oil film thickness value by the basic speed mapping coefficient, and adds the product to the idle speed constant to calculate the theoretical suction plate speed; subsequently, this embodiment divides the theoretical suction plate speed by the compensation coefficient obtained in the aforementioned steps to calculate the final output actual suction plate speed. The basic speed mapping coefficient is derived from the historical test database calibration and is set to 50 revolutions per minute per millimeter, its function being to linearly map the physical thickness of the on-site oil film to the speed power requirement; the idle speed constant is the lower limit threshold to ensure the minimum hydraulic balance of the equipment, fixed at 100 revolutions per minute. Through this data processing mechanism that deeply couples macroscopic environmental data with microscopic equipment conditions, this embodiment can ensure that the oil suction plate speed and the oil layer distribution at the actual contact interface are dynamically matched under complex operating conditions where the ship experiences severe rolling of ±15 degrees, and the speed adjustment error is always kept within 5%.

[0042] In this embodiment, for some heavy crude oil or extremely high-viscosity oil slicks that are severely emulsified and difficult to directly pump out, the oil spill recovery unit is also equipped with auxiliary physical coagulation methods. Specifically, the asphalt powder spraying equipment is used to uniformly spray specially made asphalt powder onto the oil spill area using high-pressure airflow, based on the identified heavily polluted area. In this embodiment, the average particle size of the sprayed asphalt powder is strictly screened and controlled within the range of 5 mm to 10 mm, and the spraying coverage density is precisely set to 0.5 kg per square meter. After the asphalt powder comes into contact with the oil spill on the sea surface, it can rapidly undergo a high-intensity adsorption and capture effect at the physical level, obtaining solid blocks with stable internal structure that are not easily dispersed by alternating wind and waves within a very short time of 3 minutes. Subsequently, in this embodiment, in conjunction with the mechanical grab and conveyor belt matrix configured at the stern of the oil spill recovery vessel, the solid blocks floating on the sea surface are directly and efficiently mechanically salvaged and recovered. This auxiliary salvage mechanism greatly expands the adaptability of this embodiment to extreme oil conditions, ensuring that the overall oil spill recovery rate of the entire system remains stable at over 95%, and the water content of the solid recovered materials is strictly reduced to below 10%, which greatly reduces the subsequent hazardous waste treatment costs.

[0043] Furthermore, the specific working process of the feedback optimization module 5 is as follows: In this embodiment, the feedback optimization module 5 serves as the final link in realizing closed-loop control and self-evolution of the entire system process. It includes an effect monitoring unit, a data review unit, and a model optimization unit. Specifically, the effect monitoring unit is used to continuously monitor the site using multi-source acquisition devices after the front-end physical treatment actions are completed, in order to collect oil spill residue and water quality parameters. In specific implementation, this embodiment dispatches a UAV formation and a matrix of unmanned surface vessels as the multi-source acquisition devices. Through the onboard laser fluorescence spectrometer and water quality probe, a gridded secondary scan of the target sea area after treatment is performed at a fixed polling cycle of 15 minutes. In the data processing of data extraction and fusion, this embodiment maps the reflectivity data fed back by the fluorescence spectrometer to an absolute oil spill residue value by combining it with the water background subtraction algorithm. At the same time, the dissolved oxygen content and oil concentration in the water returned by the water detection probe are extracted as the water quality parameters. This embodiment performs spatiotemporal alignment and structured packaging of the above-mentioned quantitative indicators to obtain the treatment effect data of the oil spill area, which serves as an objective physical benchmark for evaluating the compliance of this emergency response. The oil concentration parameters in the water are derived from a high-precision ultraviolet oil analyzer, which has a detection limit of 0.01 mg / L. In this embodiment, it is determined that when the oil concentration in the water drops below 0.05 mg / L for three consecutive monitoring cycles, the single local pollution prevention treatment meets the environmental protection acceptance standard.

[0044] In this embodiment, after obtaining real-time on-site feedback, the data review unit is used to perform in-depth cross-comparison and summary analysis of equipment operating efficiency and identification error by combining the aforementioned obtained disposal effect data, the oil spill identification data initially identified by the front end, and the emergency response plan issued by the intelligent decision-making module 3. In the data processing logic for error and efficiency analysis, this embodiment first performs a reverse deduction calculation of the identification error: adding the actual total recovered oil volume from the disposal effect data to the estimated residual oil volume obtained from system inversion, yields the true total oil spill distribution; then, the absolute difference between this true total oil spill distribution and the initial estimated oil spill volume in the oil spill identification data is calculated, and then this absolute difference is divided by the true total oil spill distribution to calculate the identification error rate. Simultaneously, this embodiment performs a quantitative calculation of equipment operating efficiency: extracting the expected disposal time and theoretical recovery volume from the emergency response plan, combining the actual consumption time and actual recovery volume of the operation nodes, dividing the actual recovery volume by the actual consumption time to obtain the actual recovery rate, and then dividing the actual recovery rate by the theoretical recovery rate to obtain the equipment operating efficiency coefficient. The actual recovery volume mentioned above is derived from the cumulative values ​​of the flow meters in the pipeline network of the oil spill recovery vessel, while the theoretical recovery rate is derived from the ideal calibration values ​​of the built-in expert knowledge base. In this embodiment, the abnormal data obtained above (e.g., the abnormal sample set with an identification error rate greater than 5%) is structurally correlated with the efficiency coefficient to extract the key feature vectors that deviate from the theoretical expectations, thereby obtaining the high-value retrospective dataset.

[0045] In this embodiment, based on the system shortcomings identified through in-depth analysis, the model optimization unit silently adjusts the model parameters in the background server cluster according to the retrospective dataset, and finally iteratively updates the segmentation network model and the color space segmentation algorithm. During the data processing implementation of model tuning, for algorithm-level iterations, this embodiment extracts high-recognition-error samples (i.e., complex sea state image slices that produce misjudgments or omissions) from the retrospective dataset as an incremental training set with real labels, inputting them into the underlying deep neural network. A backpropagation mechanism is used to recalculate the feature weights and bias nodes of each hidden layer of the network. During this calculation, this embodiment strictly limits the network learning rate decay parameter to an extremely small 0.01 to ensure that the fine-tuning process does not result in catastrophic forgetting and thus damage the model's existing generalization recognition ability. Simultaneously, for the iteration of the color space segmentation algorithm, this embodiment dynamically compensates for omissions caused by ambient lighting deviations by calculating the basic threshold. If the retrospective dataset shows omissions due to high reflectivity in the oil film, the preset baseline saturation lower limit threshold is lowered by a small step of 0.02, and the baseline brightness threshold range is widened by 3%. Through the online updating of the parameter matrix and the adaptive adjustment mechanism of the calculated threshold boundary, this embodiment has completed the seamless iteration of the underlying vision and decision-making algorithms without interrupting the real-time monitoring task at the front end. This enables the system to steadily increase the accuracy of oil spill feature identification to over 99% over the evolution cycle when facing specific sea areas and specific oil products, and completely establishes a closed-loop pollution prevention and control network from situational awareness to capability self-evolution.

[0046] Furthermore, this embodiment also discloses a specific application scenario of the system in a certain sea area: A medium-sized oil tanker ran aground in a certain sea area (20 nautical miles from the coastline, near an important marine aquaculture area), resulting in a heavy crude oil spill. The accident occurred at dusk, with gusts of wind reaching Force 5 and waves causing the ship to roll by ±12 degrees. The system was deployed at an emergency command center 30 nautical miles from the accident center, with three fixed-wing UAVs equipped with dual-mode visible light and infrared thermal imaging cameras and laser fluorescence spectrometers, two unmanned deployment vessels equipped with intelligent oil booms, and one oil spill recovery vessel equipped with skimming devices and asphalt powder spraying equipment.

[0047] After the system starts, the reference feature acquisition unit of the intelligent recognition module 2 collects real-time ambient light characteristic information through the UAV sensor. Calculations show that the current direct illuminance is only 800 lux. Combining the sea surface scattering coefficient and the UAV's flight altitude of 50 meters, the calculated ambient light illuminance index drops to 320 lux. Because this value is lower than the preset 500 lux threshold, the multi-source image acquisition unit adaptively switches to infrared thermal imaging acquisition for multi-source image acquisition. After the image preprocessing unit enhances the contrast and denoises the acquired infrared images, the intelligent recognition unit converts them to the HSV color space. During this process, the color space segmentation algorithm compares the target pixels with the preset heavy crude oil baseline features (baseline hue 30, baseline saturation 0.4) using a spatial cone distance, successfully extracting the continuous oil film area. Subsequently, the bilateral segmentation network model intervenes, outputting the oil spill identification data containing core parameters: the current oil spill center area thickness reaches 6 mm, the edge area thickness is 1.5 mm, and the total diffusion area reaches 25,000 square meters. The aforementioned data is transmitted back to the command center by the data transmission module 1 via a LoRa communication link equipped with AES encryption and a 5G network with an extremely low latency of 150 milliseconds.

[0048] After receiving the oil spill identification data, the data integration unit of the intelligent decision-making module 3 merges it with the on-site wind speed (level 5) and ocean current direction data to generate a unified data matrix. The risk assessment unit inputs this data matrix into the risk scoring model, predicting that the oil slick will drift with the ocean current to a high-value marine aquaculture area in 45 minutes, and then generates risk assessment data labeled as "high level". The solution generation unit uses an improved Dijkstra algorithm with a cost function, combined with ocean current resistance and oil slick concentration distribution, to plan an optimal "arc-shaped entry" path for the front-end equipment to avoid the reef area, and generates a corresponding emergency response plan. The command issuance unit compiles the plan into low-level machine code, determines the corresponding equipment scheduling and control commands, and immediately issues them.

[0049] The emergency response module 4 responded rapidly upon receiving the command. First, the oil spill containment unit controlled two unmanned deployment vessels to reach the oil spill front at a high speed of 25 knots and deploy a 1000-meter-long intelligent oil boom. The ultrasonic monitoring unit continuously emitted 40 kHz sound waves at a period of 100 milliseconds to accurately calculate the distance between the current oil boom skirt and the oil surface, and dynamically drove the omnidirectional ball to adjust its attitude, successfully completing the physical containment of the oil spill within 20 minutes of the accident. Next, the oil spill recovery unit controlled the oil spill recovery vessel to enter the heavily polluted area. Facing a rolling motion of ±12 degrees, the high-precision attitude sensor of the oil spill recovery vessel acquired the operational status data in real time, and calculated the roll compensation coefficient using a Kalman filter (after attenuation gain adjustment, this coefficient dynamically fluctuated between 0.92 and 1.08). For a core oil film with a thickness of 6 mm, the basic suction plate rotation speed of the skimming device is calibrated to 350 rpm at full load. The system uses this basic rotation speed in real time to adaptively adjust it in conjunction with the aforementioned compensation coefficient, ensuring that the suction plate can still stably adhere to the oil surface even under severe ship rolling. Simultaneously, for some heavily emulsified heavy crude oil, asphalt powder spraying equipment sprays 8 mm asphalt powder onto the sea surface, causing the crude oil to quickly solidify into solid blocks, which are then retrieved by a mechanical grab bucket.

[0050] After the oil spill was largely cleared, the effect monitoring unit of the feedback optimization module 5 dispatched a drone to conduct a gridded scan of the sea area using a high-precision ultraviolet oil analyzer. This confirmed that the oil concentration in the water had decreased to 0.04 mg / L, meeting environmental protection acceptance standards, thus obtaining the treatment effect data. The data review unit summarized and analyzed the entire process data, finding that the bilateral segmentation network model had a 3.5% reduction error in identifying the area of ​​the oil film edge (1.5 mm region) under low twilight and high wave clutter interference, thus obtaining the structured review dataset. Finally, the model optimization unit extracted edge error images as incremental samples based on this review dataset, silently fine-tuned the hidden layer weights of the segmentation network model with a learning rate of 0.0001, and widened the baseline brightness threshold range in the HSV color space by 2%, completing the iterative update of the recognition algorithm. This ensures that the edge recognition accuracy can be improved to over 99% when facing similar adverse conditions in the future.

[0051] This embodiment also provides a method for intelligent identification and emergency response to marine oil spills, including: Multi-source images of the monitoring area are acquired, and feature extraction is performed on the multi-source images using a segmentation network model and a color space segmentation algorithm to obtain an oil spill parameter set. The corresponding oil spill identification data is then determined based on the oil spill parameter set. Based on the oil spill identification data, model analysis is performed to obtain risk assessment data, and corresponding emergency response plans and equipment scheduling and control instructions are determined based on the risk assessment data. The corresponding disposal equipment is scheduled according to the equipment scheduling and control command, and the operating status data of the disposal equipment is obtained to obtain the compensation coefficient. The operating parameters of the disposal equipment are adaptively adjusted according to the compensation coefficient and the oil spill identification data. The acquired data on the effectiveness of the oil spill response, the oil spill identification data, and the emergency response plan are summarized and analyzed to obtain a retrospective dataset. Based on the retrospective dataset, the segmentation network model and the color space segmentation algorithm are iteratively updated.

[0052] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0053] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A marine oil spill intelligent identification and emergency response system, characterized in that, include: The data transmission module and the intelligent identification module, intelligent decision-making module, emergency response module, and feedback optimization module, all connected to the data transmission module; The data transmission module provides a communication link for data transmission to the intelligent identification module, the intelligent decision-making module, the emergency response module, and the feedback optimization module. The intelligent identification module acquires multi-source images of the monitoring area, extracts features from the multi-source images using a segmentation network model and a color space segmentation algorithm to obtain an oil spill parameter set, and determines corresponding oil spill identification data based on the oil spill parameter set. The intelligent decision-making module performs model analysis based on the oil spill identification data to obtain risk assessment data, and determines corresponding emergency response plans and equipment scheduling control instructions based on the risk assessment data. The emergency response module schedules corresponding response equipment according to the equipment scheduling control instructions, obtains compensation coefficients from the operating status data of the response equipment, and adaptively adjusts the operating parameters of the response equipment based on the compensation coefficients and the oil spill identification data. The feedback optimization module summarizes and analyzes the acquired oil spill area response effect data, the oil spill identification data, and the emergency response plan to obtain a review dataset, and iteratively updates the segmentation network model and the color space segmentation algorithm based on the review dataset.

2. The intelligent oil spill identification and emergency response system according to claim 1, characterized in that, The intelligent recognition module includes: The system includes a reference feature acquisition unit, a multi-source image acquisition unit, an image preprocessing unit, and an intelligent recognition unit. The reference feature acquisition unit is used to collect illumination feature information of the monitoring area and determine the image acquisition type based on the illumination feature information. The multi-source image acquisition unit is used to acquire multi-source images of the monitoring area according to the image acquisition type. The image preprocessing unit is used to perform contrast enhancement, noise reduction, and correction processing on the multi-source images to obtain a preprocessed image set. The intelligent recognition unit is used to convert the preprocessed image set to the HSV color space, generate a continuous oil film region using a color space segmentation algorithm, and extract targets from the continuous oil film region using the segmentation network model to obtain the oil spill parameter set containing the oil spill area and thickness. The corresponding oil spill recognition data is determined based on the oil spill parameter set.

3. The intelligent oil spill identification and emergency response system according to claim 1, characterized in that, The intelligent decision-making module includes: Data integration unit, risk assessment unit, solution generation unit, and instruction issuance unit; The data integration unit is used to receive and integrate the oil spill identification data to obtain a unified data matrix. The risk assessment unit is used to construct a risk scoring model based on the data matrix to predict the oil spill spread trend and obtain the risk assessment data. The scheme generation unit is used to perform optimal path planning based on the risk assessment data and Dijkstra's algorithm to determine the corresponding emergency response scheme. The instruction issuing unit is used to convert the emergency response scheme into standard control instructions and determine the corresponding equipment scheduling control instructions.

4. The intelligent oil spill identification and emergency response system according to claim 1, characterized in that, The data transmission module includes: Wireless transmission unit, wired transmission unit, and data encryption unit; The wireless transmission unit is used to construct a multi-mode communication network using LoRa communication links and wireless communication technology to obtain a wireless transmission link. The wired transmission unit is used to construct a wired network in key areas to obtain a wired transmission link. The data encryption unit is used to encrypt the transmitted data using advanced encryption standard algorithms to obtain encrypted communication data. Based on the encrypted communication data, the wireless transmission link, and the wired transmission link, a multi-mode secure network is provided as the communication link for data transmission.

5. The intelligent oil spill identification and emergency response system according to claim 1, characterized in that, The emergency response module includes: Oil spill containment and source control unit, oil spill recovery unit, chemical treatment unit, and equipment management unit; The oil spill containment and source control unit is used to deploy oil booms to designated areas according to the equipment scheduling and control instructions, obtain containment status data, and perform oil spill containment based on the containment status data. The chemical treatment unit is used to control an automatic spraying device to spray oil dispersant according to the equipment scheduling and control instructions, obtain spraying status data, and perform auxiliary treatment based on the spraying status data. The oil spill recovery unit is used to schedule the treatment equipment to recover oil spills according to the equipment scheduling and control instructions, obtain the operating status data of the treatment equipment to obtain the compensation coefficient, and adaptively adjust the operating parameters of the treatment equipment based on the compensation coefficient and the oil spill identification data. The equipment management and control unit is used to monitor the treatment equipment, obtain fault warning information, and generate corresponding backup scheduling instructions based on the fault warning information.

6. The intelligent oil spill identification and emergency response system according to claim 5, characterized in that, The oil spill recovery unit includes: Oil spill recovery vessels, oil suction devices, oil skimming devices, and asphalt powder spraying equipment; The oil spill recovery vessel is used to reach the oil spill area along the optimal path as the main carrier. The vessel's roll angle is obtained through the attitude sensor as the operating status data. The roll angle is processed by a Kalman filter to obtain the compensation coefficient. The skimming device is used to coordinate with the suction device and, based on the compensation coefficient and the oil spill identification data, the speed of the suction disc of the skimming device is used as the working parameter for adaptive adjustment. The asphalt powder spraying device is used to spray asphalt powder into the oil spill area to obtain solid blocks, which are then mechanically salvaged and recovered.

7. The intelligent oil spill identification and emergency response system according to claim 5, characterized in that, The oil containment control unit includes: Unmanned deployment vessel, ultrasonic monitoring unit, and intelligent oil boom; The unmanned deployment vessel is used to quickly deploy the intelligent oil boom according to the equipment scheduling and control instructions to obtain the initial containment state. The ultrasonic monitoring unit is used to calculate the oil surface distance by measuring the time difference between sound wave transmission and reception according to the initial containment state, so as to determine the dynamic adjustment position instruction, and adjust the intelligent oil boom to the target position to complete the oil spill containment.

8. The intelligent oil spill identification and emergency response system according to claim 1, characterized in that, The feedback optimization module includes: The system comprises an effect monitoring unit, a data review unit, and a model optimization unit. The effect monitoring unit is used to collect oil spill residue and water quality parameters using multi-source acquisition equipment to obtain the treatment effect data of the oil spill area. The data review unit is used to summarize and analyze the treatment effect data, the oil spill identification data and the emergency response plan for equipment operation efficiency and identification error to obtain the review dataset. The model optimization unit is used to adjust the model parameters according to the review dataset to iteratively update the segmentation network model and the color space segmentation algorithm.

9. A method for intelligent identification and emergency response to marine oil spills, characterized in that, include: Multi-source images of the monitoring area are acquired, and feature extraction is performed on the multi-source images using a segmentation network model and a color space segmentation algorithm to obtain an oil spill parameter set. The corresponding oil spill identification data is then determined based on the oil spill parameter set. Based on the oil spill identification data, model analysis is performed to obtain risk assessment data, and corresponding emergency response plans and equipment scheduling and control instructions are determined based on the risk assessment data. The corresponding disposal equipment is scheduled according to the equipment scheduling and control command, and the operating status data of the disposal equipment is obtained to obtain the compensation coefficient. The operating parameters of the disposal equipment are adaptively adjusted according to the compensation coefficient and the oil spill identification data. The acquired data on the effectiveness of the oil spill response, the oil spill identification data, and the emergency response plan are summarized and analyzed to obtain a retrospective dataset. Based on the retrospective dataset, the segmentation network model and the color space segmentation algorithm are iteratively updated.

Images