A water pollution event situation analysis inversion method

Abstract

This application discloses a method for analyzing and retrieving the situation of water pollution incidents, relating to the fields of remote sensing technology and environmental modeling technology. This method fuses features from acquired infrared and ultraviolet fluorescence images, establishes an oil film thickness calculation model based on the fused comprehensive features, uses this model to obtain the observed concentration in the detection area, establishes a convection-diffusion model for the target water area, and introduces time- and velocity-related loss terms. The convection-diffusion model is then used to estimate the predicted concentration in the detection area. Based on the observed and predicted concentrations, an inversion model is established, and the leak location and time in the target water area are inferred through optimization iterations. The technical solution in this application fuses features from infrared and ultraviolet fluorescence images, enabling accurate calculation of the oil concentration in the detection area. The introduction of time- and velocity-related loss terms into the convection-diffusion model improves the accuracy of retrieving the leak location and time.

Description

Technical Field

[0001] This application relates to the fields of remote sensing technology and environmental modeling technology, and more specifically, to a method for analyzing and retrieving the situation of water pollution incidents. Background Technology

[0002] Due to limitations in transportation costs and carrying capacity, waterway transportation has become the primary form of oil transport. With the prosperity and development of the social economy, the demand for oil transportation is booming, leading to frequent oil spills and water pollution incidents. Oil waterway transportation uses tankers or barges as transport vehicles, covering oceans, rivers, and lakes. The vast majority of my country's imported crude oil is transported by tankers, and major oil refineries are located along the coast or around navigable rivers and lakes. Most refined oil products are also transported via coastal and inland waterways. Therefore, the frequent occurrence of water pollution incidents has become a focus of public concern. By developing and applying advanced methods for analyzing and retrieving water pollution incidents, it is possible to effectively identify and track pollution sources, predict pollutant diffusion paths, provide strong support for environmental management and emergency response, and ultimately protect the aquatic ecological environment and public health.

[0003] Existing technologies typically employ a single detection method or in-situ measurements to measure oil film thickness in selected target water areas. These methods are easily affected by external background light or other waves, such as infrared remote sensing and ultrasonic detection. The accuracy of oil film thickness measurements obtained using a single method is low and cannot meet the requirements of subsequent inversion calculations. In-situ measurements are cumbersome and difficult to implement, with long detection cycles, making it impossible to acquire data within a single timeframe. Furthermore, in the subsequent inversion analysis, existing technologies often ignore oil loss due to external factors in the model construction, leading to inaccurate oil spill concentration calculations. Alternatively, they may use linear or nonlinear reaction terms to represent the lost oil volume, but the coefficients of these reaction terms are unrelated to the river's flow velocity and the duration of the event, making them unsuitable for different water areas and resulting in low calculation accuracy. Summary of the Invention

[0004] The purpose of this application is to solve the problem in the prior art that the oil film thickness detection is not accurate enough and the description of the amount of oil lost during water flow is not accurate enough, which leads to the inability to accurately infer the oil leak point and leak time in the later inversion calculation.

[0005] The technical solution of this application is: to provide a method for analyzing and retrieving the situation of water pollution incidents, the method comprising:

[0006] Step 1: Acquire the infrared image and ultraviolet fluorescence image of the detection area, and perform image preprocessing on each image. Then, register the preprocessed infrared image and ultraviolet fluorescence image to align the pixel positions of the two images.

[0007] Step 2: Extract the features and contrast information of corresponding pixels in the infrared image and the ultraviolet fluorescence image, assign weights according to the contrast information of corresponding pixels in the two images, and perform feature fusion on corresponding pixels in the two images based on the weights.

[0008] Step 3: Establish an oil film thickness calculation model based on the fused features, use the oil film thickness calculation model to calculate the oil film thickness corresponding to each pixel in the fused image, and calculate the oil spill concentration corresponding to each pixel based on the oil film thickness, as the observed concentration;

[0009] Step 4: Establish a convection-diffusion model for the target water area. Introduce a loss term related to time and flow velocity into the convection-diffusion model. Preset the initial oil spill location and initial oil spill time. Divide the target water area based on the spatial coordinate system. Discretize the convection-diffusion model and calculate the predicted concentration at each detection point in the spatial coordinate system using the discretized model.

[0010] Step 5: Based on the observed concentration and the predicted concentration calculated by the convection-diffusion model, establish an inversion model, and iteratively optimize the parameters of the inversion model to obtain the optimal values ​​of the initial oil spill location and initial time.

[0011] Furthermore, step 2 specifically includes:

[0012] Extracting the radiation temperature T of each pixel in the infrared image IR (i,j), texture features F IR (i,j) and thermal contrast C IR (i,j), extract the fluorescence intensity I of each pixel in the ultraviolet fluorescence image. UV (i,j), texture features F UV (i,j) and fluorescence contrast C UV (i,j), based on the extracted thermal contrast C IR (i,j) and fluorescence contrast C UV (i,j) assigns weights to two corresponding pixels in the infrared and ultraviolet fluorescence images:

[0013]

[0014] In the formula, ω IR (i,j) represents the weight of a pixel in the infrared image, ω UV (i,j) represents the weight of a pixel in the ultraviolet fluorescence image, (i,j) represents the coordinate label of the pixel, and ω IR(i,j) and ω UV The sum of (i,j) is 1, and the registered infrared image and ultraviolet fluorescence image have the same reference coordinate system;

[0015] Based on the weight allocation of different pixel positions, feature fusion is performed on the corresponding pixels in the two images sequentially to obtain a fused image. Each pixel in the fused image corresponds to a comprehensive feature vector P. fusion (i,j):

[0016] P fusion (i,j)=[ω IR ·T IR (i,j),ω UV ·I UV (i,j),ω IR ·F IR (i,j),ω UV ·F UV (i,j)] T

[0017] In the formula, [·] T This represents the transpose of a vector.

[0018] Furthermore, step 3, which establishes an oil film thickness calculation model based on the fused features, specifically includes:

[0019] Based on the comprehensive feature vector P fusion (i,j) Establish an oil film thickness calculation model:

[0020]

[0021] In the formula, h is the oil film thickness, β0 is the intercept term, and β n P is the regression coefficient. fusion,n Let be the nth component of the integrated feature vector, ∈ be the error term, n be the label of the pixel in the fused image, and N be the total number of pixels in the fused image;

[0022] Optimize the oil film thickness calculation model and calculate parameters β0 and β n The optimal value is obtained, and the oil film thickness calculation model is transformed into:

[0023] H=Xβ+∈

[0024] In the formula, H is the actual observed value matrix of oil film thickness, X is the input feature matrix, and β is the matrix including the intercept term β0 and the regression coefficient β. n Given the parameter matrix, assuming a total of M samples were measured, H is an M×1 observation vector, X is an M×(N+1) input feature matrix, and β is an (N+1)×1 parameter vector. The specific details of matrices H, X, and β are as follows:

[0025]

[0026] Minimize the sum of squares of the error terms:

[0027]

[0028] In the formula, m is the label of the sample, ∈ T For the transpose of ∈, min the sum of squares of the error terms. β Find the partial derivative of the parameter β, and set the derivative to 0, and we get:

[0029]

[0030] Solving the equation yields:

[0031] β=(X T X) -1 X T h

[0032] Finally, the intercept term β0 and regression coefficients β1, β2, ..., β0 were calculated. n ; Combine the feature vector P of different pixels fusion (i,j) and regression coefficient β n By substituting the data into the oil film thickness calculation model, the oil film thickness h at different pixels is calculated. n .

[0033] Furthermore, step 3, which calculates the oil spill concentration for each pixel based on the oil film thickness, specifically includes:

[0034] Based on the area S of the detection region corresponding to the fused image and the total number N of pixels in the fused image, the specific area s corresponding to each pixel in the detection region is calculated, s = S / N. The volume of water corresponding to each pixel is then calculated based on the area s. Based on the volume of water corresponding to each pixel With oil film thickness value h n Calculate the oil spill concentration C corresponding to each pixel. obs,n :

[0035]

[0036] In the formula, C obs,n This is the observed concentration corresponding to the nth pixel in the fused image.

[0037] Furthermore, in step 4, a convection-diffusion model for the target water area is established, and a time- and velocity-related loss term is introduced into the convection-diffusion model, specifically including:

[0038] Establish a convection-diffusion model and introduce a loss term R(C):

[0039]

[0040] In the formula, C represents the oil spill concentration. Let D be the velocity vector of the water body, D be the diffusion coefficient, and t be time. For gradient operators, The rate of change of concentration C with respect to time t. The concentration C represents the velocity vector along the flow rate. Rate of change of direction The diffusion effect represents concentration C. It represents the sum of the second derivatives of the concentration in all directions;

[0041] The loss term R(C) is related to time t and the velocity vector. The relevant formula is as follows:

[0042]

[0043] In the formula, α is the loss proportion constant, and ε is the influence coefficient of flow velocity and time on the loss proportion. Let t0 be the initial time of the oil leak, and (t-t0) be the time difference between the current time and the initial time. Samples of oil and clean water are taken from the target water area, and simulated water flow velocity is used in the laboratory to represent the flow velocity. The measurement includes multiple sets of data, including concentration C, observation time difference (t-t0), and final loss value. These data are then substituted into the equation to calculate the proportionality constant α and coefficient ε.

[0044] Establish a spatial coordinate system for the target water area, with any point within the target water area as the origin, the direction of water flow as the x-axis, the direction from the center of the water body towards the shore as the y-axis, and the vertically upward direction as the z-axis. Measure the velocity vector within the target water area. Among them, u x u y and u z The flow velocities along the x, y, and z axes, as measured in practice, are represented by the convection-diffusion model as follows:

[0045]

[0046] In the formula, D x D y D z These are the diffusion coefficients along the x-axis, y-axis, and z-axis, respectively.

[0047] Furthermore, in step 4, the target water area is divided based on a spatial coordinate system, the convection-diffusion model is discretized, and the predicted concentration at each detection point in the spatial coordinate system is calculated using the discretized model. Specifically, this includes:

[0048] Let the initial oil spill location be (x0, y0, z0), the initial time be t0, the initial concentration at the current grid point be C0, and the oil slick float on the water surface. Neglecting the flow velocity along the z-axis, we get:

[0049] C(x,y,t0)=C0δ(x-x0)δ(y-y0)

[0050] In the formula, δ represents the instantaneous release of oil at point (x0, y0) at time t0, and the initial concentration of oil at that point is C0;

[0051] The discretized convection-diffusion model divides the space containing the water body into several grid points based on coordinates, and defines the discretized concentration at each grid point as... Where (x, y) are the coordinates of the grid points, τ represents the time step label, time is discretized with a time step size Δt, and space is discretized with spatial steps size Δx and Δy. The discretized convection-diffusion model is then transformed into:

[0052]

[0053] In the formula, Let (x, y) be the concentration at time step τ. Let (x, y) be the concentration at grid point (x, y) at time step τ+1. It is the discretization of the flow term. It is the discretization of the diffusion term. The loss term is discretized. Based on the initial oil spill location (x0, y0, z0) and the initial time t0, the coordinates of each detection point in the target water spatial coordinate system are substituted to progressively calculate the predicted concentration C of each detection point at different time steps. mod .

[0054] Furthermore, step 5 specifically includes:

[0055] The predicted concentration C at each detection point is calculated based on the convection-diffusion model. mod and the observed concentration C obtained in step 3. obs,n Lijian Inversion Model:

[0056]

[0057] In the formula, C obs To observe the concentration, C mod For the model to predict concentration, θ is the parameter that needs to be inverted and calculated, where θ∈{t0,x0,y0,z0}. * The optimized parameters, parameters k is the label of the detection point in the spatial coordinate system of the target water area, and K is the total number of detection points, where K = N. All detection points are distributed in the detection area, and there is a one-to-one correspondence between the detection points and the pixels in the fused image. The gradient number of Φ with respect to the parameters (t0,x0,y0,z0) is calculated using the gradient descent rule to update the parameters (t0,x0,y0,z0), as follows:

[0058]

[0059] in:

[0060]

[0061] In the formula, η is the learning rate. Iteration stops when the inversion model reaches the predetermined number of iterations, and the initial oil spill location is output after the iteration is completed. and initial time

[0062] Furthermore, step 1 specifically includes:

[0063] Using a drone equipped with an infrared sensor, an ultraviolet light source, and a fluorescence imaging device, infrared and ultraviolet fluorescence images of the detection area are collected at night or under conditions without significant sunlight interference. The infrared sensor is used to collect infrared images, capturing and recording the infrared radiation intensity of each pixel to generate an infrared image. The ultraviolet light source and fluorescence imaging device are used to collect ultraviolet fluorescence images. The ultraviolet light source is aimed at the detection area, and the oil film in the detection area generates a fluorescence signal under ultraviolet light excitation. The fluorescence imaging device captures and records the fluorescence signal to generate an ultraviolet fluorescence image.

[0064] The acquired infrared and ultraviolet fluorescence images are denoised separately. The denoised infrared and ultraviolet fluorescence images are then registered to align the pixels in the infrared and ultraviolet fluorescence images in space. After registration, the two images have the same spatial reference coordinate system.

[0065] The beneficial effects of this application are:

[0066] The technical solution in this application acquires infrared and ultraviolet fluorescence images of the target water area within the same time period, and performs feature fusion on the infrared and ultraviolet fluorescence images. The combined fused features are then used to calculate the oil film thickness in the detection area. This multi-dimensional data acquisition and fusion method allows for the extraction of more information from different images, thereby improving the accuracy and robustness of the oil film thickness calculation. Compared to the single-dimensional oil film thickness method in previous technologies, the technical solution in this application is more accurate, which is beneficial for subsequent inversion analysis and reduces errors.

[0067] The technical solution in this application introduces a loss term related to time and flow velocity into the convection-diffusion model of water bodies. This can more accurately describe the amount of oil lost due to the passage of time and reactions such as evaporation, degradation, sedimentation, and dissolution caused by water movement. This improves the accuracy of the convection-diffusion model, further reduces the error of the inversion analysis, and makes the parameters of the oil leak point and leak time more accurate. Attached Figure Description

[0068] The advantages of the above and / or additional aspects of this application will become apparent and readily understood in the description of the embodiments in conjunction with the following drawings, wherein:

[0069] Figure 1 This is a flowchart illustrating a method for analyzing and retrieving the situation of a water pollution incident according to an embodiment of this application.

[0070] Figure 2 This is a schematic block diagram of image fusion according to an embodiment of this application. Detailed Implementation

[0071] To better understand the above-mentioned objectives, features, and advantages of this application, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in the embodiments of this application can be combined with each other.

[0072] In the following description, many specific details are set forth in order to provide a full understanding of this application. However, this application may also be implemented in other ways different from those described herein. Therefore, the scope of protection of this application is not limited to the specific embodiments disclosed below.

[0073] like Figure 1 As shown in this embodiment, a method for analyzing and retrieving the situation of water pollution incidents is provided. This method includes:

[0074] Step 1: Acquire the infrared image and ultraviolet fluorescence image of the detection area, and perform image preprocessing on each image. Then, register the preprocessed infrared image and ultraviolet fluorescence image to align the pixel positions of the two images.

[0075] A drone equipped with an infrared sensor, an ultraviolet light source, and a fluorescence imaging device is used to collect infrared and ultraviolet fluorescence images of a detection area (the area selected for measurement experiments within the target water body) in nighttime or under conditions without significant sunlight interference. The infrared sensor captures and records the infrared radiation intensity of each pixel to generate the infrared image. The ultraviolet light source and fluorescence imaging device acquire ultraviolet fluorescence images. The ultraviolet light source is aimed at the detection area, and the oil film in the detection area generates a fluorescence signal under ultraviolet light excitation. The fluorescence imaging device captures and records the fluorescence signal to generate an ultraviolet fluorescence image.

[0076] In this embodiment, images are acquired at night or under conditions without significant sunlight interference, which reduces interference and results in clearer images. For infrared sensors, the background radiation is lower and the ambient temperature is more uniform, reducing interference from sunlight and other light sources. This typically leads to clearer infrared images. For fluorescence imaging devices, it is easier to distinguish fluorescence signals excited by ultraviolet light sources, ensuring the accuracy and reliability of detection.

[0077] The acquired infrared and ultraviolet fluorescence images are denoised separately to enhance recognition accuracy. Infrared images are denoised using methods such as Gaussian filtering and wavelet denoising, while ultraviolet fluorescence images are denoised using histogram equalization and beamforming algorithms. The denoised infrared and ultraviolet fluorescence images are then registered to ensure spatial alignment. After registration, the two images have the same spatial reference coordinate system. Registration ensures that the pixel positions of the infrared and ultraviolet fluorescence images are aligned, meaning that the coordinates of the pixels in the two images correspond one-to-one, which facilitates subsequent fusion.

[0078] In this embodiment, feature point matching or other image registration algorithms can be used. For example, feature point matching can be used to extract feature points from two images, the nearest neighbor algorithm can be used to match feature point pairs between the two images, and the RANSAC algorithm can be used to remove mismatches to obtain accurate feature point pairs. The affine transformation matrix between the two images can be calculated using the least squares method or other optimization methods. The affine transformation matrix is ​​then applied to one image and transformed to the coordinate system of the other image to complete the registration.

[0079] Step 2: Extract the features and contrast information of corresponding pixels in the infrared image and the ultraviolet fluorescence image, assign weights according to the contrast information of corresponding pixels in the two images, and perform feature fusion on corresponding pixels in the two images based on the weights.

[0080] Extracting the radiation temperature T of each pixel in the infrared image IR (i,j), texture features FIR (i,j) and thermal contrast C IR (i,j); Extract the fluorescence intensity I of each pixel in the ultraviolet fluorescence image. UV (i,j), texture features F UV (i,j) and fluorescence contrast C UV (i,j); In this embodiment, these features and contrast information can be extracted using existing image processing algorithms, such as Python (programming language) and the OpenCV library (an open-source computer vision library that provides a wealth of tools and algorithms for processing images and videos).

[0081] Based on the extracted thermal contrast C IR (i,j) and fluorescence contrast C UV (i,j) represents the weights assigned to two corresponding pixels in the infrared and ultraviolet fluorescence images, respectively. The weight of a pixel in the infrared image is ω. IR (i,j), the weight of each pixel in the ultraviolet fluorescence image is ω. UV (i,j), where, for each pixel, ω IR (i,j) and ω UV The sum of (i,j) is 1, and the weight is calculated using the following formula:

[0082]

[0083] In the formula, (i,j) are the coordinate labels of the pixel. Since the registered infrared image and the ultraviolet fluorescence image have the same reference coordinate system and the pixels in the two images correspond one-to-one, the coordinates (i,j) can correspond to the coordinates of a certain pixel in both images at the same time. The subscript IR represents the infrared image and the subscript UV represents the ultraviolet fluorescence image.

[0084] Based on the weight allocation of different pixel positions, feature fusion is performed on corresponding pixels in the two images sequentially to obtain a fused image, such as... Figure 2 As shown, the fused image still uses the original reference coordinate system (i.e., the coordinate system used by the infrared and ultraviolet fluorescence images), and each pixel in the fused image contains a comprehensive feature vector, P. fusion (i,j) is:

[0085] P fusion (i,j)=[ω IR ·T IR (i,j),ω UV ·I UV (i,j),ω IR ·F IR (i,j),ω UV ·F UV (i,j)]T

[0086] In the formula, [·] T The vector transpose is represented by the subscript fusion, which indicates the fused features. The fused image can comprehensively utilize the temperature and texture information of the infrared image and the fluorescence intensity and texture information of the ultraviolet fluorescence image to enhance the detection and recognition of the target.

[0087] Step 3: Establish an oil film thickness calculation model based on the fused features, use the oil film thickness calculation model to calculate the oil film thickness corresponding to each pixel in the fused image, and calculate the oil spill concentration corresponding to each pixel based on the oil film thickness.

[0088] Based on the comprehensive feature vector P fusion (i,j) Establish an oil film thickness calculation model:

[0089]

[0090] In the formula, h is the oil film thickness, β0 is the intercept term, and β n P is the regression coefficient. fusion,n Let be the nth component of the integrated feature vector, ∈ be the error term, n be the label of a pixel in the fused image, and N be the total number of pixels in the fused image. Here, n is the label of a pixel in the fused image, and n differs only in its corresponding coordinates (i,j).

[0091] In this embodiment, to improve the accuracy of the oil film thickness calculation model, it is necessary to optimize the model parameters β0 and β2. n When the model was initially established, the parameters β0 and β n The values ​​are the initial values ​​assigned by the system. These two parameters need to be continuously optimized using optimization methods to obtain the optimal values. Methods such as Ordinary Least Squares (OLS) and gradient descent can be used to optimize the model.

[0092] Optimize the oil film thickness calculation model and calculate parameters β0 and β n The optimal value is obtained by converting the above oil film thickness calculation model into:

[0093] H=Xβ+∈

[0094] In the formula, H is the actual observed value matrix of oil film thickness, X is the input feature matrix (including a column of 1s where the intercept term is all 1s), and β is the matrix including the intercept term β0 and the regression coefficient β. n Given the parameter matrix, assuming a total of M samples were measured, H is an M×1 observation vector, X is an M×(N+1) input feature matrix, and β is an (N+1)×1 parameter vector. The specific details of matrices H, X, and β are as follows:

[0095]

[0096] To calculate β and optimize the model, we need to minimize the sum of squares of the error terms:

[0097]

[0098] In the formula, m is the label of the sample, ∈ T For the transpose of ∈, min the sum of squares of the above error terms. β Find the partial derivative of the parameter β, and set the derivative to 0, and we get:

[0099]

[0100] Solving the equation yields:

[0101] β=(X T X) -1 X T H

[0102] Finally, the intercept term β0 and regression coefficients β1, β2, ..., β0 were calculated. n The value of the error term ∈ is equal to the difference between the actual observed value and the model's predicted value. The error term can be used to evaluate the model's fit and further optimize the model. In this embodiment, the above process uses the least squares method to directly obtain the closed-form solution through matrix operations. This is a more direct optimization method. This formula minimizes the sum of squares of the error term and uses matrix inversion to directly calculate the optimal parameter vector β.

[0103] Complete the intercept term β0 and the regression coefficients β1, β2, ..., β n After solving, the comprehensive feature vector P of different pixels is obtained. fusion (i,j) and regression coefficient β n After inputting the oil film thickness calculation model, the oil film thickness at different pixels can be calculated.

[0104] Based on the area S of the detection region corresponding to the fused image and the total number N of pixels in the fused image, the specific area s corresponding to each pixel in the detection region is calculated, s = S / N. The volume of water corresponding to each pixel is then calculated based on the area s. In this embodiment, for more accurate inversion calculations, the drone's shooting altitude needs to be adjusted so that the area s is less than or equal to 1m. 2 .

[0105] Based on the volume of water corresponding to each pixel The oil film thickness value h corresponding to each pixel n The oil spill concentration C corresponding to each pixel in the detection area is calculated. obs,n (Oil spill concentration refers to the amount of oil contained in a unit volume of water), as detailed below:

[0106]

[0107] In the formula, C obs,n The oil spill concentration corresponding to the nth pixel in the fused image is the observed concentration corresponding to the nth pixel. The coordinates of the pixel are (i,j), and each pixel in the fused image corresponds to a set of coordinates.

[0108] Step 4: Establish a convection-diffusion model for the target water area. Introduce a loss term related to time and flow velocity into the convection-diffusion model. Preset the initial oil spill location and initial oil spill time. Divide the target water area based on the spatial coordinate system. Discretize the convection-diffusion model and calculate the predicted concentration at each detection point in the spatial coordinate system using the discretized model.

[0109] Because the flowing water disperses oil spills, causing them to coalesce into clumps and float erratically on the surface (most common oils float on the surface), oil spill concentration is needed to represent the oil situation in the monitored area. Most oils have very limited solubility in water and typically form an oil film floating on the surface. In the target water area, some oil undergoes evaporation, degradation, sedimentation, and dissolution over time, further reducing its concentration. Considering these reactions, a loss term R(C) needs to be introduced into the oil convection and diffusion process. This loss term R(C) should be related to the elapsed time and the flow velocity of the target water area. The following convection-diffusion model is established:

[0110]

[0111] In the formula, C represents the oil spill concentration. Let D be the velocity vector of the water body, D be the diffusion coefficient, R(C) be the time- and velocity-dependent loss term used to describe reactions such as evaporation, degradation, sedimentation, and dissolution, and t be time. For gradient operators, The rate of change of concentration C with respect to time t. The concentration C represents the velocity vector along the flow rate. The rate of change of direction represents the change in concentration caused by water flow. The diffusion effect represents the concentration C, which is the concentration change caused by molecular diffusion. It is the Laplace operator, representing the sum of the second derivatives of the concentration in all directions.

[0112] The loss term R(C) is related to time t and the velocity vector. The specific formula is as follows:

[0113]

[0114] In the formula, α is the loss ratio constant, representing the natural loss ratio of oil in the absence of flow, and ε is the influence coefficient of flow velocity and time on the loss ratio. Let be the magnitude of the water flow velocity, t0 be the initial time of the oil leak, and (t-t0) be the time difference between the current time and the initial time. The loss proportionality constant α and the coefficient ε can be obtained experimentally by sampling oil and clean water in the target water area and using simulated water flow velocity in the laboratory. The measurement includes multiple sets of data, including concentration C, observation time difference (t-t0), and final loss value. These data are then substituted into the equation to calculate the proportionality constant α and coefficient ε. This process can also be obtained using other conventional methods, which will not be elaborated here.

[0115] Establish a spatial coordinate system for the target water area, with any point within the target water area as the origin, the direction of water flow as the x-axis, the direction from the center of the water body towards the shore as the y-axis, and the vertically upward direction as the z-axis. Obtain the flow velocity vector in the target water area through actual measurement. Among them, u x u y and u z These are the actual measured flow velocities along the x, y, and z axes, respectively. In this embodiment, instruments such as flow meters and ultrasonic Doppler flow meters can be used to measure the flow velocities along the x, y, and z axes in the spatial coordinate system of the target water area.

[0116] The above convection-diffusion model It can be represented as:

[0117]

[0118] In the formula, D x D y D z These are the diffusion coefficients along the x-axis, y-axis, and z-axis, respectively (which can be estimated through actual experiments).

[0119] Based on the aforementioned convection-diffusion model with introduced loss terms, given a preset initial oil spill location (x0, y0, z0) and initial time t0, the oil spill concentration at each grid point in the target water area's spatial coordinate system at different times can be calculated. In this embodiment, establishing the spatial coordinate system of the target water area can divide the target water area into several grid points according to their coordinate positions, with the spacing between any grid points being the same. Since this embodiment detects oil pollution and the observed concentration is the oil spill concentration at the water surface, the flow velocity in the z-axis direction can be ignored during calculation, and only the data in the x-axis and y-axis directions need to be calculated.

[0120] When calculating the concentration at different grid points using a convection-diffusion model with a loss term, assuming the initial oil spill location is (x0, y0, z0), the initial time is t0, and the initial concentration at the current grid point is C0, meaning the oil leak occurred at the location (x0, y0, z0) at time t0, we obtain:

[0121] C(x,y,t0)=C0δ(x-x0)δ(y-y0)

[0122] In the formula, δ is the Dirac delta function, which represents a function that has a value at point (x0, y0) and is zero at other points. That is, it represents that the oil spill is released instantaneously at point (x0, y0) at time t0, and the initial concentration of the pollutant at that point is C0. In this embodiment, the selection of the initial time and the initial oil spill location usually requires preliminary guessing based on prior knowledge or experimental data.

[0123] The convection-diffusion model is discretized, and the space containing the water body is divided into several grid points according to coordinates. The discretized concentration is defined at each grid point. Where (x, y) are the coordinates of the grid points, τ represents the time step label, time is discretized with a time step size Δt, and space is discretized with spatial steps size Δx and Δy. The discretized convection-diffusion model is then transformed into:

[0124]

[0125] In the formula, Let (x, y) be the concentration at time step τ. Let (x, y) be the concentration at grid point (x, y) at time step τ+1. It is the discretization of the flow term. It is the discretization of the diffusion term. This involves discretizing the loss term. Using the above formula and with the initial oil spill location as (x0, y0, z0) and the initial time as t0, the concentrations of other grid points at different time steps can be calculated step by step. After substituting the coordinates of each detection point in the target water space coordinate system, the predicted concentration C corresponding to each detection point can be calculated. mod .

[0126] Step 5: Based on the observed concentration and the predicted concentration calculated by the convection-diffusion model, establish an inversion model, and iteratively optimize the parameters of the inversion model to obtain the optimal values ​​of the initial oil spill location and initial time.

[0127] Based on the above diffusion model, the predicted concentration C at each detection point is calculated. mod and the observed concentration C obtained in step 3. obs,n Lijian Inversion Model:

[0128]

[0129] In the formula, C obs To observe the concentration, C mod For the model to predict concentration, θ is the parameter that needs to be inverted and calculated, where θ∈{t0,x0,y0,z0}. * The optimized parameters, parameters The initial time t0 and the estimated initial oil spill location (t0, x0, y0, z0) are set, k is the label of the detection point in the target water area spatial coordinate system, and K is the total number of detection points, where K = N. All detection points are distributed in the detection area, and each detection point corresponds one-to-one with a pixel in the fused image. The partial derivative of Φ with respect to the parameters (t0, x0, y0, z0), i.e., the gradient, is calculated using the gradient descent rule, and the parameters (t0, x0, y0, z0) are updated as follows:

[0130]

[0131] in:

[0132]

[0133] In the formula, η is the learning rate. Iteration stops when the inversion model reaches a predetermined number of iterations. The gradient descent method can effectively optimize the inversion model, thereby finding the optimal concentration C. mod As close as possible to the observed concentration C obs Optimal parameters Output the initial oil spill location after completing iterative optimization. and initial time In this embodiment, after obtaining the predicted concentration at the current location, an inversion model is established by calculating the error between the predicted and observed concentrations. The initial oil spill location and time are continuously updated through optimization and iteration, and adjustments are made repeatedly until the objective function inversion model Φ reaches a predetermined number of iterations, minimizing the error between the predicted and observed concentrations. Finally, the initial oil spill location is obtained through inversion. and initial time

[0134] This embodiment uses diesel fuel as the primary oil spill treatment. A simulated target water tank is used as the experimental tank, providing power to continuously flow the water within it. A fixed amount of diesel fuel is added at any time and location within the tank, and the time and location coordinates are used as the initial time and initial spill location. A section of the tank is selected as the detection area. Infrared and ultraviolet fluorescence images are captured using the methods described in steps 1 to 3 of this application at night or under conditions without significant sunlight interference to obtain the observed concentration at each detection point. The predicted concentration at each detection point is calculated using the methods described in steps 4 and 5. The initial time and initial spill location are inferred through an inversion model and compared with the recorded initial time and initial spill location, with an error of less than 5%.

[0135] In summary, the technical solution of this application has been described in detail above with reference to the accompanying drawings. This application proposes a method for analyzing and retrieving the situation of water pollution incidents, including: Step 1, acquiring infrared and ultraviolet fluorescence images of the detection area, and performing image preprocessing on each image, then registering the preprocessed infrared and ultraviolet fluorescence images to align the pixel positions of the two registered images; Step 2, extracting the feature and contrast information of corresponding pixels in the infrared and ultraviolet fluorescence images, assigning weights based on the contrast information of corresponding pixels in the two images, and performing feature fusion on corresponding pixels in the two images based on the weights; Step 3, establishing an oil film thickness calculation model based on the fused features, and using the oil film thickness... The computational model calculates the oil film thickness corresponding to each pixel in the fused image, and calculates the oil spill concentration corresponding to each pixel based on the oil film thickness, which is used as the observed concentration; Step 4, a convection diffusion model for the target water area is established, and a loss term related to time and flow velocity is introduced into the convection diffusion model. The initial oil spill location and initial oil spill time are preset, the target water area is divided based on the spatial coordinate system, the convection diffusion model is discretized, and the predicted concentration of each detection point in the spatial coordinate system is calculated through the discretized model; Step 5, an inversion model is established based on the observed concentration and the predicted concentration calculated by the convection diffusion model, and the parameters of the inversion model are iteratively optimized to obtain the optimal values ​​of the initial oil spill location and initial time.

[0136] The technical solution in this application fuses infrared and ultraviolet fluorescence images, introducing a time- and velocity-related loss term into the convection-diffusion model. This solves the problem in existing technologies where insufficient accuracy in oil film thickness detection or inaccurate description of oil loss during water flow leads to inaccurate inference of the oil leak point and time in subsequent inversion calculations, thus improving the accuracy of leak location and time inversion.

[0137] The steps in this application can be rearranged, combined, or deleted according to actual needs.

[0138] The units in the device of this application can be merged, divided, or deleted according to actual needs.

[0139] Although this application has been disclosed in detail with reference to the accompanying drawings, it should be understood that these descriptions are merely exemplary and not intended to limit the application of this application. The scope of protection of this application is defined by the appended claims and may include various variations, modifications, and equivalents of the invention without departing from the scope and spirit of this application.

Claims

1. A method for analyzing and retrieving the situation of water pollution incidents, characterized in that, The method for analyzing and retrieving the situation of water pollution incidents includes: Step 1: Acquire the infrared image and ultraviolet fluorescence image of the detection area, and perform image preprocessing on each image. Then, register the preprocessed infrared image and ultraviolet fluorescence image to align the pixel positions of the two images. Step 2: Extract the features and contrast information of corresponding pixels in the infrared image and the ultraviolet fluorescence image, assign weights according to the contrast information of corresponding pixels in the two images, and perform feature fusion on corresponding pixels in the two images based on the weights. Step 3: Establish an oil film thickness calculation model based on the fused features, use the oil film thickness calculation model to calculate the oil film thickness corresponding to each pixel in the fused image, and calculate the oil spill concentration corresponding to each pixel based on the oil film thickness, as the observed concentration; Step 4: Establish a convection diffusion model for the target water area, introduce a loss term related to time and flow velocity into the convection diffusion model, preset the initial oil spill location and initial oil spill time, divide the target water area based on the spatial coordinate system, discretize the convection diffusion model, and calculate the predicted concentration of each detection point in the spatial coordinate system through the discretized model. Step 5: Based on the observed concentration and the predicted concentration calculated by the convection-diffusion model, establish an inversion model, and iteratively optimize the parameters of the inversion model to obtain the optimal values ​​of the initial oil spill location and initial time.

2. The method for analyzing and retrieving the situation of water pollution incidents as described in claim 1, characterized in that, Step 2 specifically includes: Extract the radiation temperature T of each pixel in the infrared image. IR (i,j), texture features F IR (i,j) and thermal contrast C IR (i,j), extract the fluorescence intensity I of each pixel in the ultraviolet fluorescence image. UV (i,j), texture features F UV (i,j) and fluorescence contrast C UV (i,j), based on the extracted thermal contrast C IR (i,j) and fluorescence contrast C UV (i,j) assigns weights to two corresponding pixels in the infrared and ultraviolet fluorescence images: In the formula, ω IR (i,j) represents the weight of a pixel in the infrared image, ω UV (i,j) represents the weight of a pixel in the ultraviolet fluorescence image, (i,j) represents the coordinate label of the pixel, and ω IR (i,j) and ω UV The sum of (i,j) is 1, and the registered infrared image and ultraviolet fluorescence image have the same reference coordinate system; Based on the weight allocation of different pixel positions, feature fusion is performed on the corresponding pixels in the two images sequentially to obtain a fused image. Each pixel in the fused image corresponds to a comprehensive feature vector P. fusion (i,j): P fusion (i,j)=[ω IR ·T IR (i,j),ω UV ·I UV (i,j),ω IR ·F IR (i,j),ω UV ·F UV (i,j)] T In the formula, [·] T This represents the transpose of a vector.

3. The method for analyzing and retrieving the situation of water pollution incidents as described in claim 2, characterized in that, Step 3, which involves establishing an oil film thickness calculation model based on the fused features, specifically includes: Based on the comprehensive feature vector P fusion (i,j) Establish an oil film thickness calculation model: In the formula, h is the oil film thickness, β0 is the intercept term, and β n P is the regression coefficient. fusion,n ∈ is the nth component of the integrated feature vector, ∈ is the error term, n is the label of the pixel in the fused image, and N is the total number of pixels in the fused image; Optimize the oil film thickness calculation model and calculate parameters β0 and β n The optimal value is obtained, and the oil film thickness calculation model is transformed into: H=Xβ+∈ In the formula, H is the actual observed value matrix of oil film thickness, X is the input feature matrix, and β is the matrix including the intercept term β0 and the regression coefficient β. n Given the parameter matrix, assuming a total of M samples were measured, H is an M×1 observation vector, X is an M×(N+1) input feature matrix, and β is an (N+1)×1 parameter vector. The specific details of matrices H, X, and β are as follows: Minimize the sum of squares of the error terms: In the formula, m is the label of the sample, ∈ T For the transpose of ∈, min the sum of squares of the error terms. β Find the partial derivative of the parameter β, and set the derivative to 0, and we get: Solving the equation yields: β=(X T X) -1 X T H Finally, the intercept term β0 and regression coefficients β1, β2, ..., β0 were calculated. n The combined feature vector P of different pixels fusion (i,j) and regression coefficient β n By substituting the oil film thickness calculation model into the model, the oil film thickness h at different pixels is calculated. n .

4. The method for analyzing and retrieving the situation of water pollution incidents as described in claim 3, characterized in that, Step 3, which calculates the oil spill concentration for each pixel based on the oil film thickness, specifically includes: Based on the area S of the detection region corresponding to the fused image and the total number N of pixels in the fused image, the specific area s corresponding to each pixel in the detection region is calculated, s = S / N. The volume of water corresponding to each pixel is then calculated based on the area s. Based on the volume of water corresponding to each pixel With oil film thickness value h n Calculate the oil spill concentration C corresponding to each pixel. obs,n : In the formula, C obs,n This is the observed concentration corresponding to the nth pixel in the fused image.

5. The method for analyzing and retrieving the situation of water pollution incidents as described in claim 4, characterized in that, In step 4, a convection-diffusion model for the target water area is established. This model incorporates time- and velocity-related loss terms, specifically including: Establish a convection-diffusion model and introduce a loss term R(C): In the formula, C represents the oil spill concentration. Let D be the velocity vector of the water body, D be the diffusion coefficient, and t be time. For gradient operators, This represents the rate of change of concentration C with respect to time t. The concentration C represents the velocity vector along the flow rate. Rate of change of direction The diffusion effect represents concentration C. It represents the sum of the second derivatives of the concentration in all directions; The loss term R(C) is related to time t and the velocity vector. The relevant formula is as follows: In the formula, α is the loss proportion constant, and ε is the influence coefficient of flow velocity and time on the loss proportion. Let t0 be the initial time of the oil leak, and (t-t0) be the time difference between the current time and the initial time. Samples of oil and clean water are taken from the target water area, and simulated water flow velocity is used in the laboratory to represent the flow velocity. The measurement includes multiple sets of data, including concentration C, observation time difference (t-t0), and final loss value. These data are then substituted into the equation to calculate the proportionality constant α and coefficient ε. Establish a spatial coordinate system for the target water area, with any point within the target water area as the origin, the direction of water flow as the x-axis, the direction from the center of the water body towards the shore as the y-axis, and the vertically upward direction as the z-axis. Measure the velocity vector within the target water area. Among them, u x u y and u z The flow velocities along the x, y, and z axes, as measured in practice, are represented by the convection-diffusion model as follows: In the formula, D x D y D z These are the diffusion coefficients along the x-axis, y-axis, and z-axis, respectively.

6. The method for analyzing and retrieving the situation of water pollution incidents as described in claim 5, characterized in that, Step 4 involves dividing the target water area based on a spatial coordinate system, discretizing the convection-diffusion model, and calculating the predicted concentration at each detection point in the spatial coordinate system using the discretized model. Specifically, this includes: Let the initial oil spill location be (x0, y0, z0), the initial time be t0, the initial concentration at the current grid point be C0, and the oil slick float on the water surface. Neglecting the flow velocity along the z-axis, we get: C(x,y,t0)=C0δ(x-x0)δ(y-y0) In the formula, δ represents the instantaneous release of oil at point (x0, y0) at time t0, and the initial concentration of oil at that point is C0; The discretized convection-diffusion model divides the space containing the water body into several grid points based on coordinates, and defines the discretized concentration at each grid point as... Where (x, y) are the coordinates of the grid points, τ represents the time step label, time is discretized with a time step size Δt, and space is discretized with spatial steps size Δx and Δy. The discretized convection-diffusion model is then transformed into: In the formula, Let (x, y) be the concentration at time step τ. Let (x, y) be the concentration at grid point (x, y) at time step τ+1. It is the discretization of the flow term. It is the discretization of the diffusion term. The loss term is discretized. Based on the initial oil spill location (x0, y0, z0) and the initial time t0, the coordinates of each detection point in the target water spatial coordinate system are substituted to progressively calculate the predicted concentration C of each detection point at different time steps. mod .

7. The method for analyzing and retrieving the situation of water pollution incidents as described in claim 6, characterized in that, Step 5 specifically includes: The predicted concentration C at each detection point is calculated based on the convection-diffusion model. mod and the observed concentration C obtained in step 3. obs,n Lijian Inversion Model: In the formula, C obs To observe the concentration, C mod For the model to predict concentration, θ is the parameter that needs to be inverted and calculated, where θ∈{t0,x0,y0,z0}. * The optimized parameters, parameters k is the label of the detection point in the spatial coordinate system of the target water area, and K is the total number of detection points, where K = N. All detection points are distributed in the detection area, and there is a one-to-one correspondence between the detection points and the pixels in the fused image. The gradient number of Φ with respect to the parameters (t0,x0,y0,z0) is calculated using the gradient descent rule to update the parameters (t0,x0,y0,z0), as follows: in: In the formula, η is the learning rate. Iteration stops when the inversion model reaches the predetermined number of iterations, and the initial oil spill location is output after the iteration is completed. and initial time 8. The method for analyzing and retrieving the situation of water pollution incidents as described in claim 1, characterized in that, Step 1 specifically includes: Using a drone equipped with an infrared sensor, an ultraviolet light source, and a fluorescence imaging device, infrared and ultraviolet fluorescence images of the detection area are collected at night or under conditions without significant sunlight interference. The infrared sensor is used to collect infrared images, capturing and recording the infrared radiation intensity of each pixel to generate an infrared image. The ultraviolet light source and fluorescence imaging device are used to collect ultraviolet fluorescence images. The ultraviolet light source is aimed at the detection area, and the oil film in the detection area generates a fluorescence signal under ultraviolet light excitation. The fluorescence imaging device captures and records the fluorescence signal to generate an ultraviolet fluorescence image. The acquired infrared and ultraviolet fluorescence images are denoised separately. The denoised infrared and ultraviolet fluorescence images are then registered to align the pixels in the infrared and ultraviolet fluorescence images in space. After registration, the two images have the same spatial reference coordinate system.

Images