Neutron quantitative inversion method and system for medium migration in organically contaminated soil and medium

Abstract

The application discloses a kind of organic contaminated soil medium neutron quantitative inversion method of migration, system and medium, it is related to soil pollutant data inversion technical field, the method includes based on the first neutron transmission image sequence of target test sample of preset time window acquisition;First neutron transmission image sequence in each picture is sequentially preprocessed, multi-scale feature extraction obtains corresponding first multidimensional feature matrix;Wherein, first multidimensional feature matrix includes one or more of color space component feature map, gray attenuation mean feature map and image texture feature map;First multidimensional feature matrix is input into preset based on ViT's physical-data double-driven general inversion model and obtains the predicted value of the absolute mass concentration of organic pollutant per pixel;Based on the preset time window and corresponding predicted value obtains spatiotemporal evolution atlas.The application can improve the data precision and data reliability of the data inversion of soil pollutant migration.

Description

Technical Field

[0001] This application relates to the field of soil pollutant data inversion technology, and in particular to a neutron quantitative inversion method, system and medium for media migration in organically contaminated soil. Background Technology

[0002] With the deepening of soil pollution prevention and control in my country, the accurate detection of the occurrence and migration process of pollutants in organically contaminated soil has become a core prerequisite for risk management and remediation effect evaluation of contaminated sites.

[0003] Neutron radiography (NR) is a non-destructive testing technique that uses the difference in intensity attenuation after a neutron beam penetrates an object to create an image. It has been successfully applied to the study of unsaturated seepage of pore water in soil and cement-based materials. By observing the attenuation signal of the neutron beam penetrating the medium, it enables non-destructive, continuous, and visual tracking of the transport process of hydrogen-rich fluids.

[0004] Existing NR technology cannot achieve precise spatiotemporal quantitative characterization of the entire process of pollutant migration, and cannot meet the actual engineering needs of in-situ dynamic monitoring and remediation effect evaluation of contaminated sites. Summary of the Invention

[0005] The purpose of this application is to provide a neutron quantitative inversion method, system, and medium for media migration in organically contaminated soil, which can improve the data accuracy and reliability of soil pollutant migration data inversion.

[0006] To achieve the above objectives, this application provides the following solution: Firstly, this application provides a neutron quantitative inversion method for media migration in organically contaminated soil, comprising: The first neutron transmission image sequence of the target test sample is obtained based on a preset time window; Each first neutron transmission image in the first neutron transmission image sequence is preprocessed to obtain a first preprocessed image sequence; For each first preprocessed image in the first preprocessed image sequence, multi-scale feature extraction is performed to obtain a corresponding first multi-dimensional feature matrix; wherein, the first multi-dimensional feature matrix includes one or more of the color space component feature map, gray-level attenuation mean feature map, and image texture feature map; The first multidimensional feature matrix is ​​input into a preset ViT-based physical-data dual-drive general inversion model to obtain the pixel-by-pixel predicted value of the absolute mass concentration of organic pollutants output by the preset ViT-based physical-data dual-drive general inversion model. The spatiotemporal evolution map is obtained based on the preset time window and the corresponding predicted value.

[0007] Optionally, the preset ViT-based physical-data dual-driven general inversion model is obtained according to the following method: The second neutron transmission image sequence of the test sample is obtained based on a preset time window; Each second neutron transmission image in the second neutron transmission image sequence is preprocessed to obtain a second preprocessed image sequence; For each preprocessed image in the second preprocessed image sequence, multi-scale feature extraction is performed to obtain the corresponding second multi-dimensional feature matrix; wherein, the second multi-dimensional feature matrix includes one or more of the color space component feature map, gray-level attenuation mean feature map, and image texture feature map; The core micro-region of pollutants is determined and obtained based on the spatial multidimensional feature gradient magnitude of the second multidimensional feature matrix. Obtain the transient gamma-ray characteristic energy spectrum of the core micro-region of the pollutant, and obtain the absolute mass concentration of organic pollutants in the core micro-region of the pollutant based on the transient gamma-ray characteristic energy spectrum; A training set is established based on the second multidimensional feature matrix, wherein the absolute mass concentration is used as the label of the second multidimensional feature matrix in the training set. The preset ViT inversion model is trained based on the training set to obtain the preset ViT-based physical-data dual-driven general inversion model. The preset ViT-based physical-data dual-driven general inversion model is used to predict the absolute mass concentration of organic pollutants in each pixel of the target test sample.

[0008] Optionally, the preset ViT inversion model includes an input layer, an image segmentation layer, a Transformer encoder, and an output layer connected in sequence. The preset ViT inversion model is trained and iterated using a joint loss function that integrates temperature and humidity monitoring parameters with the physical laws of neutron decay. The joint loss function that integrates temperature and humidity monitoring parameters with the physical laws of neutron decay is described by the following formula: Among them, data-driven items The physical constraint term is used to measure the error between the predicted value output by the model and the absolute mass concentration. To combine Beer-Lambert's law with real-time hydrological / thermodynamic parameter correction values ​​obtained from sensors; regularization term Used to prevent model overfitting. All are dynamically balanced weights; The data-driven item This can be described by the following formula: Where N represents the size of the training set. C represents the predicted value. PGNAA,i The error representing the absolute mass concentration; The physical constraints Described by the following formula: In the formula, M represents the total number of pixels in the first preprocessed image. For temperature T and moisture content The equivalent macroscopic cross section of the modulation. d The thickness of the target test sample. I j The grayscale value of the first preprocessed image. I 0 The grayscale value of the background image when there is no target test sample is specified.

[0009] Optionally, the image texture feature map includes the information entropy of the gray-level co-occurrence matrix, and the determination and acquisition of the core micro-region of pollutants based on the spatial multidimensional feature gradient magnitude of the multidimensional feature matrix includes: The gradient magnitude matrix is ​​obtained based on the grayscale attenuation average feature map and the information entropy feature map; Points with gradient magnitudes greater than the preset gradient threshold are considered as pixels with drastic gradient changes. Based on the pixels with drastic gradient changes, connected component analysis is performed to obtain candidate regions. Among the candidate regions, regions with an area greater than a preset area threshold are selected as the core micro-regions of the pollutants.

[0010] Optionally, obtaining the absolute mass concentration of organic pollutants in the core micro-region of the pollutant based on the transient gamma-ray characteristic energy spectrum includes: The characteristic energy spectrum of the instantaneous gamma rays is preprocessed to obtain a preprocessed energy spectrum signal; Based on the Gaussian-Lorentz mixture function, the energy peaks of the C target element and the H target element in the preprocessed energy spectrum signal are deconvolved and fitted to obtain the characteristic net peak area and absolute mass of the C target element and the H target element. The absolute mass of the pollutant is obtained based on the preset ratio of the C target element and the H target element, and the absolute mass concentration of organic pollutants in the core micro-region of the pollutant is obtained based on the absolute mass of the pollutant.

[0011] Optionally, the step of performing multi-scale feature extraction on each first preprocessed image in the first preprocessed image sequence to obtain the corresponding first multi-dimensional feature matrix includes: For each of the first preprocessed images, obtain the color space component feature map of the first preprocessed image in a preset color space; The first preprocessed image is processed pixel by pixel using a preset sliding window to obtain a feature map of average grayscale attenuation. Based on the preset sliding window, the first preprocessed image is processed to obtain the probability co-occurrence matrix of each pixel. Based on the probability co-occurrence matrix, the corresponding contrast information is obtained to obtain the contrast feature map. Based on the probability co-occurrence matrix, the corresponding information entropy is obtained to obtain the information entropy feature map. Based on the probability co-occurrence matrix, the corresponding correlation information is obtained to obtain the correlation feature map. The first multidimensional feature matrix is ​​obtained by concatenating the color space component feature map, the grayscale attenuation average feature map, the contrast feature map, the information entropy feature map, and the correlation feature map.

[0012] Optionally, the step of preprocessing each first neutron transmission image in the first neutron transmission image sequence to obtain a first preprocessed image sequence includes: performing background and dark field correction, nonlocal mean denoising, and neutron beam fluctuation compensation on each first neutron transmission image in sequence to obtain a first preprocessed image, and obtaining the first preprocessed image sequence based on all the first preprocessed images.

[0013] Secondly, this application provides a neutron quantitative inversion method system for media migration in organically contaminated soil, comprising: a data acquisition device and a data processing terminal; wherein The data acquisition device includes a displacement device, a sample support stage, a neutron generator, a beam-limiting aperture, a collimator, a scintillation screen, a reflector, and an imaging device, all housed within the radiation shielding experimental chamber. The sample support stage is used to fix the target test sample, and it is also equipped with a temperature sensor and a humidity sensor. The neutron generator is fixed in the radiation shielding experimental chamber. The collimator of the neutron generator is aligned with the surface of the sample support stage. A collimator and a beam-limiting aperture are also arranged sequentially between the neutron generator and the sample support stage. The scintillation screen, the reflector, and the imaging device are arranged on the back of the sample support stage. The light emitted by the neutron generator can pass through the collimator, the beam-limiting aperture, and the sample support stage in sequence, and then be reflected by the scintillation screen and the reflector to the imaging device. The imaging device acquires a transmission image. The data acquisition device is used to emit a first neutron beam to the sample support stage and acquire a first neutron transmission image sequence of the target test sample based on a preset time window; The data processing terminal is used for: The first neutron transmission image sequence of the target test sample is obtained based on a preset time window; Each first neutron transmission image in the first neutron transmission image sequence is preprocessed to obtain a first preprocessed image sequence; For each first preprocessed image in the first preprocessed image sequence, multi-scale feature extraction is performed to obtain a corresponding first multi-dimensional feature matrix; wherein, the first multi-dimensional feature matrix includes one or more of the color space component feature map, gray-level attenuation mean feature map, and image texture feature map; The first multidimensional feature matrix is ​​input into a preset ViT-based physical-data dual-drive general inversion model to obtain the pixel-by-pixel predicted value of the absolute mass concentration of organic pollutants output by the preset ViT-based physical-data dual-drive general inversion model. The spatiotemporal evolution map is obtained based on the preset time window and the corresponding predicted value.

[0014] Optionally, the data acquisition device further includes a high-purity germanium detector disposed inside the radiation shielding experimental chamber, the high-purity germanium detector being disposed on the side of the sample support stage; The data acquisition device is also used for: A second neutron beam is emitted toward the sample support stage, and a sequence of second neutron transmission images of the test sample is acquired based on a preset time window. The data processing terminal is also used for: Obtain the second neutron transmission image sequence; Each second neutron transmission image in the second neutron transmission image sequence is preprocessed to obtain a second preprocessed image sequence; For each preprocessed image in the second preprocessed image sequence, multi-scale feature extraction is performed to obtain the corresponding second multi-dimensional feature matrix; wherein, the second multi-dimensional feature matrix includes one or more of the color space component feature map, gray-level attenuation mean feature map, and image texture feature map; The core micro-region of pollutants is determined and obtained based on the spatial multidimensional feature gradient magnitude of the multidimensional feature matrix. The data acquisition device is also used for: The neutron generator is moved to be aligned with the core micro-region of the pollutant by a displacement device, and a third neutron beam is emitted into the core micro-region of the pollutant; at the same time, the transient gamma-ray characteristic energy spectrum of the core micro-region of the pollutant is collected by the high-purity germanium detector. The data processing terminal is also used for: Obtain the transient gamma-ray characteristic energy spectrum of the core micro-region of the pollutant, and obtain the absolute mass concentration of organic pollutants in the core micro-region of the pollutant based on the transient gamma-ray characteristic energy spectrum; A training set is established based on the second multidimensional feature matrix, wherein the absolute mass concentration is used as the label of the second multidimensional feature matrix in the training set. The preset ViT inversion model is trained based on the training set to obtain the preset ViT-based physical-data dual-driven general inversion model. The preset ViT-based physical-data dual-driven general inversion model is used to predict the absolute mass concentration of organic pollutants in each pixel of the target test sample.

[0015] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the neutron quantitative inversion method for media migration in organically contaminated soil as described above.

[0016] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the neutron quantitative inversion method for media migration in organically contaminated soil as described above.

[0017] Fifthly, this application provides a computer program product, including a computer program that, when executed by a data processing terminal, implements the steps of the neutron quantitative inversion method for media migration in organically contaminated soil as described above.

[0018] According to the specific embodiments provided in this application, the following technical effects are disclosed: This application provides a neutron quantitative inversion method, system, and medium for media migration in organically contaminated soil. The first multidimensional feature matrix is ​​used to characterize the strong correlation between the absolute concentration of organic pollutants in the neutron transmission image. The preset ViT-based physical-data dual-driven general inversion model is obtained by training by integrating physical mechanisms and observational data, which can improve the accuracy of predicting the absolute concentration of organic pollutants and the model's noise resistance. The spatiotemporal evolution map obtained based on the predicted value of the absolute mass concentration of organic pollutants realizes the four-dimensional dynamic visualization characterization of organic pollutants in the porous soil medium, providing an intuitive and continuous decision-making basis for pollution source tracing, diffusion trend prediction, and remediation scheme formulation. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application 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 this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1An application environment diagram of a neutron quantitative inversion method for media migration in organically contaminated soil provided in an embodiment of this application; Figure 2 This is a schematic diagram of a method for obtaining a preset ViT-based physical-data dual-driven general inversion model according to an embodiment of this application; Figure 3 This application provides a schematic diagram of the structure of a neutron quantitative inversion system for media migration in organically contaminated soil, according to one embodiment of the present application. Figure 4 A surface view of a sample support stage provided in an embodiment of this application; Figure 5 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application.

[0021] Explanation of reference numerals in the attached figures: 600 - Data acquisition device; 700 - Data processing terminal; 601 - Radiation shielding experimental chamber; 602 - Displacement and movement device; 603 - Neutron generator; 604 - Collimator; 605 - Beam limiting aperture; 606 - Neutron beam; 607 - Sample support stage; 608 - Gamma ray; 609 - High-purity germanium detector; 610 - Reflector; 611 - Imaging device; 612 - Scintillation screen. Detailed Implementation

[0022] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0023] It should be noted that the terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented, for example, in orders other than those illustrated or described herein.

[0024] It should be noted that "at the time of..." in the embodiments of this application can be either at the instant when a certain situation occurs, or for a period of time after the occurrence of a certain situation. The embodiments of this application do not make specific limitations on this.

[0025] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0026] In one exemplary embodiment, such as Figure 1 As shown, a neutron quantitative inversion method for media migration in organically contaminated soil is provided. This method is executed by computer equipment, specifically by a terminal or server alone, or by both a terminal and a server. In this embodiment, the method includes steps 101 to 105. Wherein: Step 101: Obtain the first neutron transmission image sequence of the target test sample based on a preset time window; Specifically, the first neutron transmission image sequence is acquired by the data acquisition device, and the first neutron transmission image sequence includes multiple first neutron transmission images corresponding to different preset times; As one example, typical clay minerals (such as kaolinite, montmorillonite, and illite) were selected, and samples were prepared by controlling the initial dry density and moisture content. These samples served as the target test samples. Typical organic pollutants (such as petroleum hydrocarbons and polycyclic aromatic hydrocarbons) were used as the detection targets. Point source / area source infiltration modes were set up, and dry-wet cycles and temperature gradient boundaries were applied to accurately reproduce the multiphase flow migration characteristics of organically contaminated soil under complex hydrological and thermodynamic conditions (such as thermal desorption processes).

[0027] Specifically, the data acquisition device includes a displacement device, a sample support stage, a neutron generator, a beam-limiting aperture, a collimator, a scintillation screen, a reflector, and an imaging device, all housed within the radiation shielding experimental chamber. The sample support stage is used to fix the target test sample and is also equipped with temperature and humidity sensors. The neutron generator is fixed in the radiation shielding experimental chamber. The collimator of the neutron generator is aligned with the surface of the sample stage. A collimator and a beam-limiting aperture are also arranged between the neutron generator and the sample stage. A scintillation screen, a reflector, and an imaging device are arranged on the back of the sample stage. The light emitted by the neutron generator passes through the collimator, the beam-limiting aperture, and the sample stage in sequence, and is reflected by the scintillation screen and the reflector to the imaging device. The imaging device acquires the transmitted image. The data acquisition device also includes a high-purity germanium detector arranged inside the radiation shielding experimental chamber and located on the side of the sample stage. Furthermore, by controlling the data acquisition device to emit a neutron beam to cover the target test sample, a first neutron transmission image sequence is obtained; Specifically, the data acquisition device enables complete temporal synchronization between neutron detection and pollutant migration processes, providing a guarantee for the completion of the method described in this application.

[0028] Step 102: Preprocess each first neutron transmission image in the first neutron transmission image sequence to obtain a first preprocessed image sequence; Specifically, for each first neutron transmission image, background and dark field correction, nonlocal mean denoising, and neutron beam fluctuation compensation are performed sequentially to obtain a first preprocessed image to eliminate neutron scattering and detector system errors. A first preprocessed image sequence is obtained based on all the first preprocessed images.

[0029] Step 103: Perform multi-scale feature extraction on each of the first preprocessed images in the first preprocessed image sequence to obtain the corresponding first multi-dimensional feature matrix; wherein, the first multi-dimensional feature matrix includes one or more of the color space component feature map, gray-level attenuation mean feature map and image texture feature map; Specifically, the color space component feature map, grayscale attenuation mean feature map, and image texture feature map are all extracted in this application, and are feature maps that can characterize the strong correlation between the absolute concentration of organic pollutants in neutron transmission images. These are used to improve the generalization ability of the preset ViT-based physics-data dual-driven general inversion model.

[0030] Step 104: Input the first multidimensional feature matrix into the preset ViT-based physical-data dual-drive general inversion model to obtain the pixel-by-pixel predicted value of the absolute mass concentration of organic pollutants output by the preset ViT-based physical-data dual-drive general inversion model. Specifically, this application targets organic pollutants (such as petroleum hydrocarbons and polycyclic aromatic hydrocarbons) for detection.

[0031] Specifically, the pre-defined ViT-based physical-data dual-driven general inversion model is trained by integrating physical mechanisms and observational data, which can improve the accuracy of predicting the absolute concentration of organic pollutants and the model's noise resistance.

[0032] Step 105: Obtain the spatiotemporal evolution map based on the preset time window and the corresponding predicted value.

[0033] Specifically, in this step, based on the preset time window and the corresponding predicted value, the absolute concentration distribution cloud map of organic pollutants corresponding to each time point is obtained. The spatiotemporal evolution map of organic pollutant transport in the porous soil medium is reconstructed by stacking the absolute concentration distribution cloud maps of organic pollutants at different time points.

[0034] By implementing steps 101 to 105 above, the first multidimensional feature matrix is ​​used to characterize the features in the neutron transmission image that are strongly correlated with the absolute concentration of organic pollutants. The pre-set ViT-based physical-data dual-driven general inversion model is obtained by training by integrating physical mechanisms and observational data, which can improve the accuracy of predicting the absolute concentration of organic pollutants and the model's noise resistance. The spatiotemporal evolution map obtained based on the predicted value of the absolute mass concentration of organic pollutants realizes the four-dimensional dynamic visualization of organic pollutants in the porous soil medium, providing an intuitive and continuous decision-making basis for pollution source tracing, diffusion trend prediction and remediation scheme formulation.

[0035] In another exemplary embodiment of this application, in order to characterize the strong correlation between the absolute concentration of organic pollutants in the neutron transmission image, the image texture feature map further includes a contrast feature map, an information entropy feature map, and a correlation feature map. The first multidimensional feature matrix (step 103) can be obtained through the following steps 201-204: Step 201: For each first preprocessed image, obtain the color space component feature map of the first preprocessed image in a preset color space; Specifically, the preset color space can be selected as either RGB or HSV color space components.

[0036] Step 202: Process the first preprocessed image pixel by pixel based on a preset sliding window to obtain the grayscale attenuation average value feature map; Specifically, the average grayscale decay of each pixel is obtained based on a preset sliding window using the following formula, and the average grayscale decay of all pixels constitutes the grayscale decay average feature map. ( x , y ): ; ( x , y ) is a pixel ( x , y The average grayscale attenuation, where ROI represents the preset sliding window. I ( x , y () represents the grayscale value of the preset sliding window. I 0( x , y ) represents the grayscale value of the preset sliding window when there is no sample; Step 203: Obtain the probability co-occurrence matrix of each pixel by processing the first preprocessed image based on a preset sliding window. P ( x , yBased on the probability co-occurrence matrix, the corresponding contrast information is obtained to obtain the contrast feature map; based on the probability co-occurrence matrix, the corresponding information entropy is obtained to obtain the information entropy feature map; based on the probability co-occurrence matrix, the corresponding correlation information is obtained to obtain the correlation feature map. Specifically, for each pixel, the contrast is obtained according to the following formula, and the contrast of all pixels is used to form a contrast feature map. T con ( x , y ): ; in, t con ( x , y Pixel ( x , y The contrast of ) P x,y ( i , j ) represents the preset sliding window corresponding to this pixel, and W represents the number of gray levels; Specifically, for each pixel, the information entropy is obtained according to the following formula, and the contrast of all pixels constitutes the information entropy feature map. T ent ( x , y ): ; in, t ent ( x , y Pixel ( x , y Information entropy σ x and σ y These are the standard deviations of the grayscale distribution of the pixel in the row and column directions, respectively.

[0037] Step 204: Concatenate the color space component feature map, grayscale attenuation average feature map, contrast feature map, information entropy feature map, and correlation feature map to obtain the first multidimensional feature matrix.

[0038] As one embodiment, the first multidimensional feature matrix X NR ( x , y It can be defined as: ; In the formula, C 1 、C2 and C 3 This represents the feature map components corresponding to the three independent channels of the pixel in the RGB or HSV color space.

[0039] In another exemplary embodiment of this application, the preset ViT-based physical-data dual-driven general inversion model is obtained by training by integrating physical mechanisms and observational data, such as... Figure 2 The training method includes the following steps 301 to 306: Step 301: Obtain the second neutron transmission image sequence of the test sample based on a preset time window; Furthermore, a second neutron transmission image sequence is obtained by controlling the data acquisition device to emit a neutron beam to cover the test sample; Step 302: Preprocess each second neutron transmission image in the second neutron transmission image sequence to obtain a second preprocessed image sequence; Step 303: Perform multi-scale feature extraction on each of the second preprocessed images in the second preprocessed image sequence to obtain the corresponding second multi-dimensional feature matrix; wherein, the second multi-dimensional feature matrix includes one or more of the color space component feature map, gray-level attenuation mean feature map and image texture feature map; Specifically, the feature extraction steps in steps 301 to 303 are similar to those in steps 101 to 103, and will not be described again here.

[0040] Step 304: Determine and obtain the core micro-region of pollutants based on the spatial multidimensional feature gradient magnitude of the second multidimensional feature matrix; Specifically, step 304 includes the following steps 401 to 403: Step 401: Obtain the gradient magnitude matrix based on the grayscale attenuation average feature map and the information entropy feature map; As one embodiment, the grayscale attenuation average feature map and the information entropy feature map are selected, and based on Sobel's algorithm, each pixel... x direction and y The spatial partial derivatives of the direction, taking the matrix composed of the spatial partial derivatives of all pixels as the gradient magnitude matrix, can be defined by the following formula: In the formula, G ( x , y Let be the spatial partial derivative of a given pixel. The weighting coefficients for balancing grayscale gradients and texture gradients are used to quantify the combined rate of change of the pixel in terms of physical decay and spatial topology.

[0041] Step 402: Obtain the gradient threshold by processing the gradient magnitude matrix based on the statistical threshold method; in the gradient magnitude matrix, the points with gradient magnitudes greater than the gradient threshold are regarded as pixels with drastic gradient changes. Furthermore, an adaptive statistical thresholding method is used to process the background image to obtain the gradient threshold, i.e., the threshold is determined to be... +3 ,in and These are the mean and standard deviation of the background image, respectively. If a certain pixel ( x,y The multidimensional feature gradient magnitude G() x,y )satisfy If a pixel is a local maximum in its local neighborhood, it is determined that the gradient at the location of the pixel has changed drastically, and it is marked as a candidate pixel in the pollutant migration front or enrichment region.

[0042] Step 403: Perform connected component analysis on pixels with drastic gradient changes to obtain candidate regions. Select regions with an area greater than a preset area threshold from the candidate regions as the core micro-regions of pollutants.

[0043] Specifically, based on pixels with drastic gradient changes, connected regions with an area greater than a set threshold are extracted as core micro-regions.

[0044] Step 305: Obtain the transient gamma-ray characteristic energy spectrum of the core micro-region of the pollutant, and obtain the absolute mass concentration of organic pollutants in the core micro-region of the pollutant based on the transient gamma-ray characteristic energy spectrum. Specifically, step 305 includes the following steps 501 to 503: Step 501: Perform nonlinear iterative calculations on the characteristic energy spectrum of the instantaneous gamma rays to obtain the preprocessed energy spectrum signal; Specifically, the γ-ray spectrum is nonlinearly iteratively calculated using the SNIP algorithm (Statistical Sensitive Nonlinear Iterative Peak-clipping) to extract and remove the continuous background baseline generated by environmental scattering, thus obtaining the net energy spectrum signal; or conventional nuclear energy spectrum processing methods such as polynomial fitting and step background subtraction are used for baseline correction. The goal is to make the characteristic peaks of C and H elements stand out from the background, resulting in a "preprocessed energy spectrum" with a clean background and smooth peaks.

[0045] Step 502: Deconvolve and fit the energy peaks of C target element and H target element in the preprocessed energy spectrum signal based on the Gaussian-Lorentz mixture function to obtain the characteristic net peak area and absolute mass of C target element and H target element. Specifically, on the preprocessed energy spectrum, the characteristic γ-energy peak positions corresponding to C and H elements are first located; then, a local spectral segment is extracted with the peak as the center, and the Gaussian-Lorentz mixture function is used to perform morphological fitting on each peak. By deconvolution calculation, overlapping or interfering components are removed to obtain the net area of ​​C and H peaks; finally, the net area is substituted into the pre-established standard curve to calculate the absolute mass of C and H target elements in the detection micro-region.

[0046] Step 503: Obtain the absolute mass of the pollutant based on the preset ratio of C and H target elements of the pollutant, and obtain the absolute mass concentration of organic pollutants in the core micro-region of the pollutant based on the absolute mass of the pollutant.

[0047] Specifically, a predetermined ratio (i.e., a fixed C and H composition ratio, such as CH2 and C8H) can be determined based on the molecular formula of the pollutant. 18 (etc.) The absolute mass of pollutants can be deduced from the absolute mass of C and H elements. The two calculation results should be verified against each other, and the average value is generally taken as the final value. Furthermore, the ratio of the absolute mass of pollutants to the volume of the core micro-region is the absolute mass concentration of that core micro-region.

[0048] Step 306: Establish a training set based on the second multidimensional feature matrix. The absolute mass concentration is used as the label of the second multidimensional feature matrix in the training set. Train the preset ViT inversion model based on the training set to obtain the preset ViT-based physical-data dual-driven general inversion model. The preset ViT-based physical-data dual-driven general inversion model is used to predict the absolute mass concentration of organic pollutants in each pixel of the target test sample.

[0049] Furthermore, while acquiring the second neutron transmission image sequence of the test sample based on a preset time window, temperature and water content are also acquired based on the same preset time window. The temperature and water content data are time-aligned with the second neutron transmission image sequence and used as parameters of the loss function to determine the physical constraint term. .

[0050] Furthermore, a temperature sensor and a humidity sensor are also installed on the sample support stage, and the temperature and moisture content are obtained based on the temperature sensor and the humidity sensor.

[0051] Specifically, the pre-defined ViT inversion model comprises an input layer, an image segmentation layer, a Transformer encoder, and an output layer connected sequentially. The pre-defined ViT inversion model is trained and iterated using a joint loss function that integrates temperature and humidity monitoring parameters with the physical laws of neutron decay. This joint loss function is described by the following formula: Among them, data-driven items The physical constraint term is used to measure the error between the predicted value and the absolute mass concentration in the model output. To combine Beer-Lambert's law with real-time hydrological / thermodynamic parameter correction values ​​obtained from sensors; regularization term Used to prevent model overfitting. All are dynamically balanced weights; Data-driven items This can be described by the following formula: Where N represents the size of the training set. C represents the predicted absolute mass concentration of organic pollutants. PGNAA,i This indicates the absolute mass concentration of organic pollutants; Physical constraints Described by the following formula: In the formula, M represents the total number of pixels in the first preprocessed image. For temperature T and moisture content The equivalent macroscopic cross section of the modulation. d For the target test sample thickness, I j The grayscale values ​​of the first preprocessed image. I 0 This represents the grayscale value of the background image when there are no target test samples. Physical constraint terms. To ensure that the theoretical transmittance calculated from the predicted concentration based on the output value is consistent with the neutron attenuation rate I / I0: This application provides a neutron quantitative inversion method for media migration in organically contaminated soil, which includes the following technical advantages: Through the integrated design of the neutron chamber and the precise spatial confinement of the boron carbide beam-limiting aperture, the full-field high-frequency transmission sequence images of NR (Neutron Radiography) and the characteristic energy spectra of transient gamma rays in the core micro-region of PGNAA (Prompt Gamma Neutron Activation Analysis) were simultaneously acquired within the same time window. Relative image features and local absolute concentrations with precisely matched spatial coordinates and completely consistent time nodes were obtained, successfully combining the continuous spatial visualization advantages of NR with the high-precision absolute quantification advantages of PGNAA for light elements.

[0052] A dual-driven inversion model based on Visual Transformer (ViT) was constructed, using the absolute concentration of micro-areas measured by PGNAA as the label for the training interface. A physical constraint term based on the Beer-Lambert law and real-time temperature and humidity data was introduced into the loss function. This effectively constrained the iteration direction of the neural network, avoiding the black-box defects of purely data-driven models, and enabling the model to maintain strong physical rigor and generalization ability even under drastically changing temperature and humidity gradients.

[0053] It solves the problem that existing NR technology can only perform qualitative or semi-quantitative visualization analysis through grayscale, fills the gap of lack of standardized fusion algorithms for multimodal neutron data, and realizes accurate inversion from dimensionless image features to pixel-by-pixel absolute concentration of organic pollutants across the entire domain.

[0054] The experimental platform coupled the complex hydrological and thermodynamic boundaries of the actual site and combined the full-cycle neutron inversion concentration field temporally to generate a non-invasive four-dimensional evolution map. This clearly characterizes the dynamic migration front, concentration gradient distribution, and spatiotemporal evolution of organic pollutants in porous media, achieving transparent characterization of target elements.

[0055] It breaks through the limitations of static or discrete spatial detection, solves the problem that existing technologies cannot continuously track the dynamic migration process of pollutants in situ, and provides direct and reliable quantitative data support for the refined risk management of contaminated sites and the evaluation of engineering effects such as thermal remediation.

[0056] Based on the same inventive concept, this application also provides a neutron quantitative inversion system for realizing the above-mentioned method for quantitative inversion of media migration in organically contaminated soil. The solution provided by this device is similar to the solution described in the above-described method; therefore, the specific limitations in one or more system embodiments provided below can be found in the limitations of the neutron quantitative inversion method for media migration in organically contaminated soil described above, and will not be repeated here.

[0057] In one exemplary embodiment, such as Figure 3 As shown, a neutron quantitative inversion system for media migration in organically contaminated soil is provided, including a data acquisition device 600 and a data processing terminal 700; wherein The data acquisition device 600 includes a displacement device 602, a sample support stage 607, a neutron generator 603, a beam-limiting aperture 605, a collimator 604, a scintillation screen 612, a reflector 610, and an imaging device 611, all housed within the radiation shielding experimental chamber 601. The sample support stage 607 is used to fix the target test sample, and it is also equipped with a temperature sensor and a humidity sensor. The neutron generator 603 is fixed inside the radiation shielding experimental chamber 601. The collimator of the neutron generator 603 is aligned with the surface of the sample support stage 607. A collimator 604 and a beam-limiting aperture 605 are also arranged between the neutron generator 603 and the sample support stage 607. The beam-limiting aperture 605 is fixed on the displacement device 602. The scintillation screen 612, the reflector 610 and the imaging device 611 are arranged on the back of the sample support stage 607. The light emitted by the neutron generator 603 can pass through the collimator 604, the beam-limiting aperture 605 and the sample support stage 607 in sequence, and then be reflected by the scintillation screen 612 and the reflector 610 to the imaging device 611. The imaging device 611 collects the transmission image. The data acquisition device 600 also includes a high-purity germanium detector 609 installed in the radiation shielding experimental chamber 601, and the high-purity germanium detector 609 is installed on the side of the sample support stage 607. Specifically, a pore water pressure sensor and a flow sensor are also connected to the sample support stage 607 to construct a multi-dimensional spatiotemporal boundary condition detection network.

[0058] Specifically, the data acquisition device 600 enables complete temporal synchronization between neutron detection and pollutant migration processes, providing a guarantee for the completion of the method described in this application.

[0059] Furthermore, the sample support stage 607 is a metal plate, and the high-purity germanium detector 609 is disposed on the side of the sample support stage 607. Its arrangement ensures that the rays detected by the high-purity germanium detector 609 are at a 90° angle to the photon beam, so as to receive the gamma rays 608. To accurately display the placement position of the sample support stage 607, such as... Figure 3 The sample support stage 607 shown is a side view, as follows: Figure 4 This is a surface view of the sample stage 607.

[0060] Furthermore, specifically, the displacement device 602 is a vision guidance system including a high-precision industrial robotic arm and a servo motor control system. The beam-limiting aperture 605 is set at the end of the robotic arm. The displacement device 602 can control the beam-limiting aperture 605 to move in a plane parallel to the sample, thereby changing the position of the beam-limiting aperture 605. By changing the parameters of the beam-limiting aperture, the size of the area on the sample hit by the beamline can be changed.

[0061] Furthermore, the neutron generator 603 is used to emit a neutron beam 606, the collimator 604 is used to filter the neutron beam 606 into parallel light, and the beam limiter 605 is used to determine the field of view and size of the illumination. The beam limiter 605 is a boron carbide-containing beam limiter.

[0062] Furthermore, the scintillation screen 612 is arranged parallel to the sample support stage 607. Neutrons are transmitted through the sample support stage 607 and strike the scintillation screen 612, where they are converted into visible light images. The incident angle of the visible light is 45° with the reflector. The image on the reflector is captured by the imaging device 611, thus preventing the imaging device 611 from being directly exposed to neutrons.

[0063] The data acquisition device 600 is used to emit a first neutron beam to the sample support stage 607 and acquire a first neutron transmission image sequence of the target test sample based on a preset time window. Data processing terminal 700, used for: The first neutron transmission image sequence of the target test sample is obtained based on a preset time window; Each first neutron transmission image in the first neutron transmission image sequence is preprocessed to obtain a first preprocessed image sequence; For each of the first preprocessed images in the first preprocessed image sequence, multi-scale feature extraction is performed to obtain the corresponding first multi-dimensional feature matrix; wherein, the first multi-dimensional feature matrix includes one or more of the color space component feature map, gray-level attenuation mean feature map, and image texture feature map; The first multidimensional feature matrix is ​​input into the preset ViT-based physical-data dual-drive general inversion model to obtain the pixel-by-pixel predicted value of the absolute mass concentration of organic pollutants output by the preset ViT-based physical-data dual-drive general inversion model. Spatiotemporal evolution maps are obtained based on preset time windows and corresponding predicted values.

[0064] As an optional implementation, the data acquisition device 600 is also used for: A second neutron beam is emitted toward the sample support stage 607, and a sequence of second neutron transmission images of the target test sample is acquired based on a preset time window. The data processing terminal 700 is also used for: Obtain the second neutron transmission image sequence; Each second neutron transmission image in the second neutron transmission image sequence is preprocessed to obtain a second preprocessed image sequence; For each preprocessed image in the second preprocessed image sequence, multi-scale feature extraction is performed to obtain the corresponding second multi-dimensional feature matrix; wherein, the second multi-dimensional feature matrix includes one or more of the color space component feature map, gray-level attenuation mean feature map, and image texture feature map; The core micro-region of pollutants is determined and obtained based on the spatial multidimensional feature gradient magnitude of the multidimensional feature matrix. The data acquisition device 600 is also used for: The neutron generator 603 is moved by the displacement device 602 to be aligned with the core micro-region of the pollutant, and a third neutron beam is emitted into the core micro-region of the pollutant; at the same time, the transient gamma-ray characteristic energy spectrum of the core micro-region of the pollutant is collected by the high-purity germanium detector 609. Specifically, the data processing terminal 700 sends the coordinates of the core micro-region of the pollutant to the displacement device 602 and the neutron generator 603. The displacement device 602 controls the neutron generator 603 to move to align with the core micro-region of the pollutant, so that the spatial center of the collimation slit of the neutron beam 606 is strictly aligned with the physical coordinates. The neutron generator 603 generates a third neutron beam in response to the instruction of the data processing terminal 700.

[0065] The data processing terminal 700 is also used for: Obtain the transient gamma-ray characteristic energy spectrum of the core micro-region of pollutants, and obtain the absolute mass concentration of organic pollutants in the core micro-region of pollutants based on the transient gamma-ray characteristic energy spectrum; A training set is established based on the second multidimensional feature matrix, with the absolute mass concentration used as the label of the second multidimensional feature matrix in the training set. The preset ViT inversion model is trained based on the training set to obtain a preset ViT-based physical-data dual-driven general inversion model. The preset ViT-based physical-data dual-driven general inversion model is used to predict the absolute mass concentration of organic pollutants in each pixel of the target test sample.

[0066] As an optional implementation, in the data processing terminal 700, the preset ViT inversion model includes an input layer, an image segmentation layer, a Transformer encoder, and an output layer connected in sequence. The preset ViT inversion model is trained and iterated using a joint loss function that integrates temperature and humidity monitoring parameters with the physical laws of neutron decay. The joint loss function that integrates temperature and humidity monitoring parameters with the physical laws of neutron decay is described by the following formula: Among them, data-driven items The physical constraint term is used to measure the error between the predicted value and the absolute mass concentration in the model output. To combine Beer-Lambert's law with real-time hydrological / thermodynamic parameter correction values ​​obtained from sensors; regularization term Used to prevent model overfitting. All are dynamically balanced weights; Data-driven items This can be described by the following formula: Where N represents the size of the training set. C represents the predicted value. PGNAA,i The error representing the absolute mass concentration; Physical constraints Described by the following formula: In the formula, M represents the total number of pixels in the first preprocessed image. For temperature T and moisture content The equivalent macroscopic cross section of the modulation. d For the target test sample thickness, I j The grayscale values ​​of the first preprocessed image. I 0 This represents the grayscale value of the background image when there are no target test samples.

[0067] As an optional implementation, the data processing terminal 700 is further used for: The gradient magnitude matrix is ​​obtained based on the grayscale attenuation average feature map and the information entropy feature map; Points with gradient magnitudes greater than a preset gradient threshold are considered as pixels with drastic gradient changes. Candidate regions are obtained by performing connected component analysis on pixels with drastic gradient changes. Regions with an area greater than a preset area threshold are selected from the candidate regions as the core micro-regions of pollutants.

[0068] As an optional implementation, the data processing terminal 700 is further used for: Preprocessing is performed on the characteristic energy spectrum of transient gamma rays to obtain the preprocessed energy spectrum signal; Based on the Gaussian-Lorentz mixture function, the energy peaks of C target element and H target element in the preprocessed energy spectrum signal are deconvolved and fitted to obtain the characteristic net peak area and absolute mass of C target element and H target element. The absolute mass of pollutants is obtained based on the preset ratio of C and H target elements, and the absolute mass concentration of organic pollutants in the core micro-region of the pollutants is obtained based on the absolute mass of the pollutants.

[0069] As an optional implementation, the data processing terminal 700 is further used for: For each first preprocessed image, obtain the color space component feature map of the first preprocessed image in a preset color space; The grayscale attenuation average feature map is obtained by processing the first preprocessed image pixel by pixel using a preset sliding window. The first preprocessed image is processed using a preset sliding window to obtain the probability co-occurrence matrix of each pixel. The corresponding contrast information is obtained based on the probability co-occurrence matrix to obtain a contrast feature map. The corresponding information entropy is obtained based on the probability co-occurrence matrix to obtain an information entropy feature map. The corresponding correlation information is obtained based on the probability co-occurrence matrix to obtain a correlation feature map. The first multidimensional feature matrix is ​​obtained by concatenating the color space component feature map, the grayscale attenuation average feature map, the contrast feature map, the information entropy feature map, and the correlation feature map.

[0070] As an optional implementation, the data processing terminal 700 is further used for: For each first neutron transmission image, background and dark field correction, nonlocal mean denoising and neutron beam fluctuation compensation are performed sequentially to obtain a first preprocessed image, and a first preprocessed image sequence is obtained based on all the first preprocessed images.

[0071] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 5 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs in the non-volatile storage media. The database stores data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a neutron quantitative inversion method for media migration in organically contaminated soil.

[0072] Those skilled in the art will understand that Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0073] In one exemplary embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0074] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0075] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0076] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0077] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0078] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0079] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. In summary, the content of this specification should not be construed as a limitation of this application.

Claims

1. A neutron quantitative inversion method for media migration in organically contaminated soil, characterized in that, The neutron quantitative inversion method for media migration in organically contaminated soil includes: The first neutron transmission image sequence of the target test sample is obtained based on a preset time window; Each first neutron transmission image in the first neutron transmission image sequence is preprocessed to obtain a first preprocessed image sequence; For each first preprocessed image in the first preprocessed image sequence, multi-scale feature extraction is performed to obtain a corresponding first multi-dimensional feature matrix; wherein, the first multi-dimensional feature matrix includes one or more of the color space component feature map, gray-level attenuation mean feature map, and image texture feature map; The first multidimensional feature matrix is ​​input into a preset ViT-based physical-data dual-drive general inversion model to obtain the pixel-by-pixel predicted value of the absolute mass concentration of organic pollutants output by the preset ViT-based physical-data dual-drive general inversion model. The spatiotemporal evolution map is obtained based on the preset time window and the corresponding predicted value.

2. The neutron quantitative inversion method for media migration in organically contaminated soil according to claim 1, characterized in that, The preset ViT-based physical-data dual-driven general inversion model is obtained according to the following method: The second neutron transmission image sequence of the test sample is obtained based on the preset time window; Each second neutron transmission image in the second neutron transmission image sequence is preprocessed to obtain a second preprocessed image sequence; For each preprocessed image in the second preprocessed image sequence, multi-scale feature extraction is performed to obtain the corresponding second multi-dimensional feature matrix; wherein, the second multi-dimensional feature matrix includes one or more of the color space component feature map, gray-level attenuation mean feature map, and image texture feature map; The core micro-region of pollutants is determined and obtained based on the spatial multidimensional feature gradient magnitude of the second multidimensional feature matrix. Obtain the transient gamma-ray characteristic energy spectrum of the core micro-region of the pollutant, and obtain the absolute mass concentration of organic pollutants in the core micro-region of the pollutant based on the transient gamma-ray characteristic energy spectrum; A training set is established based on the second multidimensional feature matrix, wherein the absolute mass concentration is used as the label of the second multidimensional feature matrix in the training set. The preset ViT inversion model is trained based on the training set to obtain the preset ViT-based physical-data dual-driven general inversion model. The preset ViT-based physical-data dual-driven general inversion model is used to predict the absolute mass concentration of organic pollutants in each pixel of the target test sample.

3. The neutron quantitative inversion method for media migration in organically contaminated soil according to claim 2, characterized in that, The pre-defined ViT inversion model comprises an input layer, an image segmentation layer, a Transformer encoder, and an output layer connected in sequence. The pre-defined ViT inversion model is trained and iterated using a joint loss function that integrates temperature and humidity monitoring parameters with the physical laws of neutron decay. This joint loss function is described by the following formula: Among them, data-driven items The physical constraint term is used to measure the error between the predicted value output by the model and the absolute mass concentration. To combine Beer-Lambert's law with real-time hydrological / thermodynamic parameter correction values ​​obtained from sensors; regularization term Used to prevent model overfitting. All are dynamically balanced weights; The data-driven item This can be described by the following formula: Where N represents the amount of data in the training set. C represents the predicted value. PGNAA,i This represents the error in the absolute mass concentration; The physical constraints Described by the following formula: In the formula, M represents the total number of pixels in the first preprocessed image. For temperature T and moisture content The equivalent macroscopic cross section of the modulation. d The thickness of the target test sample. I j The grayscale value of the first preprocessed image. I 0 The grayscale value of the background image when there is no target test sample is specified.

4. The neutron quantitative inversion method for media migration in organically contaminated soil according to claim 2, characterized in that, The image texture feature map includes the information entropy of the gray-level co-occurrence matrix. The determination and acquisition of the core micro-region of pollutants based on the spatial multidimensional feature gradient magnitude of the second multidimensional feature matrix includes: The gradient magnitude matrix is ​​obtained based on the grayscale attenuation average feature map and the information entropy feature map; Points with gradient magnitudes greater than a preset gradient threshold are considered as pixels with drastic gradient changes. Based on the pixels with drastic gradient changes, connected component analysis is performed to obtain candidate regions. Among the candidate regions, regions with an area greater than a preset area threshold are selected as the core micro-regions of the pollutants.

5. The neutron quantitative inversion method for media migration in organically contaminated soil according to claim 2, characterized in that, The step of obtaining the absolute mass concentration of organic pollutants in the core micro-region of the pollutant based on the characteristic energy spectrum of the transient gamma rays includes: The characteristic energy spectrum of the instantaneous gamma rays is preprocessed to obtain a preprocessed energy spectrum signal; Based on the Gaussian-Lorentz mixture function, the energy peaks of the C target element and the H target element in the preprocessed energy spectrum signal are deconvolved and fitted to obtain the characteristic net peak area and absolute mass of the C target element and the H target element. The absolute mass of the pollutant is obtained based on the preset ratio of the C target element and the H target element, and the absolute mass concentration of organic pollutants in the core micro-region of the pollutant is obtained based on the absolute mass of the pollutant.

6. The neutron quantitative inversion method for media migration in organically contaminated soil according to claim 1, characterized in that, The step of performing multi-scale feature extraction on each first preprocessed image in the first preprocessed image sequence to obtain the corresponding first multi-dimensional feature matrix includes: For each of the first preprocessed images, obtain the color space component feature map of the first preprocessed image in a preset color space; The first preprocessed image is processed pixel by pixel using a preset sliding window to obtain a feature map of average grayscale attenuation. Based on the preset sliding window, the first preprocessed image is processed to obtain the probability co-occurrence matrix of each pixel. Based on the probability co-occurrence matrix, the corresponding contrast information is obtained to obtain a contrast feature map. Based on the probability co-occurrence matrix, the corresponding information entropy is obtained to obtain an information entropy feature map. Based on the probability co-occurrence matrix, the corresponding correlation information is obtained to obtain a correlation feature map. The first multidimensional feature matrix is ​​obtained by concatenating the color space component feature map, the grayscale attenuation average feature map, the contrast feature map, the information entropy feature map, and the correlation feature map.

7. The neutron quantitative inversion method for media migration in organically contaminated soil according to claim 1, characterized in that, The step of preprocessing each of the first neutron transmission images in the first neutron transmission image sequence to obtain a first preprocessed image sequence includes: performing background and dark field correction, nonlocal mean denoising and neutron beam fluctuation compensation sequentially on each of the first neutron transmission images to obtain a first preprocessed image, and obtaining the first preprocessed image sequence based on all the first preprocessed images.

8. A neutron quantitative inversion system for media migration in organically contaminated soil, characterized in that, The system includes a data acquisition device and a data processing terminal; in The data acquisition device includes a displacement device, a sample support stage, a neutron generator, a beam-limiting aperture, a collimator, a scintillation screen, a reflector, and an imaging device, all housed within the radiation shielding experimental chamber. The sample support stage is used to fix the target test sample, and it is also equipped with a temperature sensor and a humidity sensor. The neutron generator is fixed in the radiation shielding experimental chamber. The collimator of the neutron generator is aligned with the surface of the sample support stage. A collimator and a beam-limiting aperture are also arranged sequentially between the neutron generator and the sample support stage. The scintillation screen, the reflector, and the imaging device are arranged on the back of the sample support stage. The light emitted by the neutron generator can pass through the collimator, the beam-limiting aperture, and the sample support stage in sequence, and then be reflected by the scintillation screen and the reflector to the imaging device. The imaging device acquires a transmission image. The data acquisition device is used to emit a first neutron beam to the sample support stage and acquire a first neutron transmission image sequence of the target test sample based on a preset time window; The data processing terminal is used for: The first neutron transmission image sequence of the target test sample is obtained based on a preset time window; Each first neutron transmission image in the first neutron transmission image sequence is preprocessed to obtain a first preprocessed image sequence; For each first preprocessed image in the first preprocessed image sequence, multi-scale feature extraction is performed to obtain a corresponding first multi-dimensional feature matrix; wherein, the first multi-dimensional feature matrix includes one or more of the color space component feature map, gray-level attenuation mean feature map, and image texture feature map; The first multidimensional feature matrix is ​​input into a preset ViT-based physical-data dual-drive general inversion model to obtain the pixel-by-pixel predicted value of the absolute mass concentration of organic pollutants output by the preset ViT-based physical-data dual-drive general inversion model. The spatiotemporal evolution map is obtained based on the preset time window and the corresponding predicted value.

9. The system according to claim 8, characterized in that, The data acquisition device also includes a high-purity germanium detector installed inside the radiation shielding experimental chamber, and the high-purity germanium detector is located on the side of the sample support stage. The data acquisition device is also used for: A second neutron beam is emitted toward the sample support stage, and a sequence of second neutron transmission images of the test sample is acquired based on a preset time window. The data processing terminal is also used for: Obtain the second neutron transmission image sequence; Each second neutron transmission image in the second neutron transmission image sequence is preprocessed to obtain a second preprocessed image sequence; For each preprocessed image in the second preprocessed image sequence, multi-scale feature extraction is performed to obtain the corresponding second multi-dimensional feature matrix; wherein, the second multi-dimensional feature matrix includes one or more of the color space component feature map, gray-level attenuation mean feature map, and image texture feature map; The core micro-region of pollutants is determined and obtained based on the spatial multidimensional feature gradient magnitude of the second multidimensional feature matrix. The data acquisition device is also used for: The neutron generator is moved to be aligned with the core micro-region of the pollutant by a displacement device, and a third neutron beam is emitted into the core micro-region of the pollutant; at the same time, the transient gamma-ray characteristic energy spectrum of the core micro-region of the pollutant is collected by the high-purity germanium detector. The data processing terminal is also used for: Obtain the transient gamma-ray characteristic energy spectrum of the core micro-region of the pollutant, and obtain the absolute mass concentration of organic pollutants in the core micro-region of the pollutant based on the transient gamma-ray characteristic energy spectrum; A training set is established based on the second multidimensional feature matrix, wherein the absolute mass concentration is used as the label of the second multidimensional feature matrix in the training set. The preset ViT inversion model is trained based on the training set to obtain the preset ViT-based physical-data dual-driven general inversion model. The preset ViT-based physical-data dual-driven general inversion model is used to predict the absolute mass concentration of organic pollutants in each pixel of the target test sample.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the data processing terminal, it implements the steps of the neutron quantitative inversion method for media migration in organically contaminated soil as described in any one of claims 1-7.

Images