Method for predicting underground water pollution diffusion of chemical industry park based on artificial intelligence

Abstract

The invention relates to the technical field of data processing, in particular to a chemical industry park groundwater pollution diffusion prediction method based on artificial intelligence. The method comprises the following steps: acquiring pollutant concentration information data and geological exploration parameter data in underground water in a chemical industry park, and preprocessing to obtain preprocessed data; based on the preprocessed data, a self-shaping function embedding mechanism is introduced, a pollution action function is constructed, and a pollution potential energy field embedding tensor is generated; introducing a geological disturbance function generator, constructing a spatial disturbance tensor, and generating a spatial correction tensor of pollution transmission in combination with the pollution potential energy field embedded tensor; and based on the pollution potential energy field embedded tensor and the pollution transmission space correction tensor, pollution propagation modeling is carried out, and underground water pollution diffusion prediction is realized. The technical problem that the groundwater pollution diffusion prediction result is seriously deviated from the actual diffusion track due to the fact that a traditional groundwater pollution diffusion prediction method is inaccurate and incomplete in data processing and analysis is solved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a method for predicting the spread of groundwater pollution in a chemical park based on artificial intelligence. Background Art

[0002] Due to factors such as complex process flows and challenging emission management, chemical industrial parks are high-risk areas for groundwater contamination. Once a pollution incident occurs, pollutants often enter the groundwater system through a variety of pathways, including leakage, surface infiltration, and infiltration from storage sites. These pollutants can spread in unpredictable ways within heterogeneous porous media, contaminating water resources and potentially causing regional ecological damage and drinking water safety risks. Therefore, accurately predicting the spatiotemporal evolution of pollution in its early stages of spread has become a key enabling technology for pollution prevention and control, emergency response, and scientific governance.

[0003] Traditional methods for predicting groundwater pollution diffusion are primarily based on traditional physical models and numerical simulation techniques, such as the finite difference method (FDM), finite volume method (FVM), and finite element method (FEM) based on the Advance-Dispersion equation. These methods emphasize the simulation of the conservation process of pollutant migration under hydrodynamic conditions, requiring the acquisition of high-precision hydrogeological parameters and the accurate expression of the convection, diffusion, and attenuation behavior of pollutants in the governing equations. However, due to the complex geological conditions, drastic parameter variations, and diverse distribution of pollution sources within chemical parks, the actual collected data is often incomplete, discontinuous, or even missing, making it difficult to fully model the actual diffusion behavior through accurate numerical simulations based on the governing equations. Furthermore, the pollution diffusion process is often driven by external disturbances such as rainfall, groundwater pumping, and artificial water injection. Traditional static models are unable to respond to the periodic disturbance effects and propagation path reconstruction problems in the pollution dynamics process.

[0004] In summary, the traditional groundwater pollution diffusion prediction method still has technical problems such as inaccurate and incomplete data processing and analysis, which leads to serious deviations between the groundwater pollution diffusion prediction results and the actual diffusion trajectory. Summary of the Invention

[0005] The present invention provides an artificial intelligence-based prediction method for groundwater pollution diffusion in chemical parks to solve the technical problem that traditional groundwater pollution diffusion prediction methods have inaccurate and incomplete data processing and analysis, resulting in serious deviation between the groundwater pollution diffusion prediction results and the actual diffusion trajectory.

[0006] The present invention provides an artificial intelligence-based method for predicting the spread of groundwater pollution in a chemical park, which specifically includes the following technical solutions: A method for predicting the spread of groundwater pollution in a chemical park based on artificial intelligence includes the following steps: S1. Obtain pollutant concentration data and geological survey parameter data from groundwater in the chemical park and preprocess them to obtain preprocessed data. Based on the preprocessed data, introduce a self-shaping function embedding mechanism to construct a pollution action function and generate a pollution potential energy field embedding tensor. S2. Introduce a geological disturbance function generator to construct a spatial disturbance tensor, and combine it with the pollution potential field embedding tensor to generate a spatial correction tensor for pollution transmission; based on the pollution potential field embedding tensor and the spatial correction tensor for pollution transmission, perform pollution propagation modeling to obtain a pollution diffusion prediction tensor to realize groundwater pollution diffusion prediction.

[0007] Preferably, the S1 specifically includes: The preprocessed data includes preprocessed pollutant concentration information data and preprocessed geological exploration parameter data; in the process of implementing the self-shaping function embedding mechanism, based on the soil permeability coefficient in the preprocessed pollutant concentration information data and the preprocessed geological exploration parameter data, and introducing a spatial attenuation modulation term, a pollution action function is constructed to quantify the diffusion behavior of pollutants in three-dimensional space.

[0008] Preferably, the S1 specifically includes: An attention weighting mechanism is introduced to perform weighted processing on the pollution action function and construct an embedding tensor of the pollution potential energy field.

[0009] Preferably, the S2 specifically includes: The geological disturbance function generator simulates nonlinear disturbance points at spatial positions by setting a disturbance kernel.

[0010] Preferably, the S2 specifically includes: Based on the preprocessed geological exploration parameter data, the disturbance intensity time weight, disturbance angle and space-time oscillation frequency factor are obtained; based on the disturbance angle and space-time oscillation frequency factor, the spatial attenuation function of the disturbance intensity and the space-time disturbance frequency modulation function are constructed, and combined with the disturbance intensity time weight, the spatial disturbance tensor is constructed.

[0011] Preferably, the S2 specifically includes: Based on the spatial disturbance tensor and the pollution potential field embedding tensor, an exponential function and a time modulation term are introduced to generate a spatial correction tensor of pollution transmission.

[0012] Preferably, the S2 specifically includes: The diffusion channel weight factor is introduced and combined with the diffusion frequency modulation factor, and path direction-aware convolution operation is performed on the pollution potential field embedding tensor and the spatial correction tensor of pollution transmission to obtain the pollution diffusion prediction tensor.

[0013] Preferably, the S2 specifically includes: The diffusion frequency modulation factor dynamically modulates the diffusion process by analyzing the time-frequency characteristics of the pre-processed pollutant concentration information data and combining it with the cosine function.

[0014] The beneficial effects of the technical solution of the present invention are: 1. Through the self-shaping function embedding mechanism, the pollutant concentration information data in groundwater is integrated with the geological exploration parameter data to construct a pollution action function. Through function mapping, a three-dimensional pollution potential field embedding tensor with geological responsiveness and source area positioning is formed. This not only converts heterogeneous data into a unified format, but also achieves continuous expression while retaining spatial heterogeneity, solving the problem that traditional pollution monitoring data can only be used for statistics but cannot generate evolution models, thereby improving the accuracy and expressiveness of the data.

[0015] 2. Design a geological disturbance function generator that integrates multiple factors such as the influence of geological faults, permeability discontinuities, and soil stratification adsorption. Combined with the spatial gradient of the pollution potential field embedding tensor, a spatial correction tensor for pollution transmission is constructed to achieve dynamic processing of the disturbance response in different regions along the pollution propagation path. This is different from the simplified modeling method of traditional groundwater pollution diffusion prediction methods that uniformly smooth the geological structure. It can more realistically reflect the deflection, slowdown, or aggregation behavior of pollutants in heterogeneous geological conditions.

[0016] 3. The introduction of diffusion channel weighting factors and path-direction-aware convolution operations allow for pollution propagation modeling. This concept of structurally coupled directional convolution allows for modeling of pollution diffusion in multiple dominant directions (such as along stratigraphic bedding, hydraulic mainstream, and slope), and forms a complete pollution migration forecast through tensor fusion. Compared to traditional holistic field modeling, this method can clearly reveal the combined causes of pollution flowing from different sources and paths into a target area, providing a solid data foundation for subsequent pollution source tracing and regional risk stratification. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a flow chart of a method for predicting groundwater pollution spread in a chemical park based on artificial intelligence according to the present invention. DETAILED DESCRIPTION

[0018] In order to further illustrate the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0019] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0020] The following describes in detail a specific scheme of a method for predicting groundwater pollution diffusion in a chemical park based on artificial intelligence provided by the present invention in conjunction with the accompanying drawings.

[0021] Refer to the attached Figure 1 , which shows a flow chart of a method for predicting groundwater pollution diffusion in a chemical park based on artificial intelligence provided by one embodiment of the present invention, the method comprising the following steps: S1. Obtain pollutant concentration data and geological survey parameter data from groundwater in the chemical park and preprocess them to obtain preprocessed data. Based on the preprocessed data, introduce a self-shaping function embedding mechanism to construct a pollution action function and generate a pollution potential energy field embedding tensor. Monitoring points are deployed underground in the chemical park according to expert experience, and data acquisition devices (such as hydrogeological sensors, etc.) are used to obtain pollutant concentration information data in groundwater at each monitoring point at different times, as well as geological exploration parameter data at the location of the monitoring point, such as soil permeability coefficient, hydraulic gradient, groundwater flow information (such as underground flow velocity changes, direction, runoff angle, etc.), underground structure disturbance spectrum, and sewage discharge activity records; the pollutant concentration information data and geological exploration parameter data are preprocessed to obtain preprocessed data; the preprocessed data includes preprocessed pollutant concentration information data and preprocessed geological exploration parameter data; the preprocessing process includes denoising, filling missing values, cleaning, standardization and normalization, etc., all of which use technical means well known to those skilled in the art and are not described in detail here; In order to map the pollution monitoring data with discontinuous time and non-uniform spatial distribution into a continuous space-time tensor structure, a self-shaping function embedding mechanism is introduced based on the preprocessed data. Based on the nonlinear water flow diffusion formula in seepage theory and the non-Euclidean disturbance field construction model, a pollution action function is constructed to quantify the diffusion behavior of pollutants at each monitoring point in three-dimensional space. The pollution monitoring data is the preprocessed data. The pollution action function represents the pollution impact projection of the monitoring point on the three-dimensional space at a certain point in time. The specific form is: , in, It is Monitoring points at time Point The output value of the pollution function; It is Monitoring points at time Pre-processed pollutant concentration information data; It is The soil permeability coefficient in the pre-processed geological exploration parameter data at the location of each monitoring point is an important geological parameter that represents the water migration capacity. The reference value range is ; It is Monitoring points at time The dynamic amplification coefficient of pollution impact is determined by the pollution growth rate and groundwater flow rate changes in the pre-processed geological exploration parameter data. The reference value range is ; It is The geological disturbance at the monitoring point The disturbance main frequency in the direction (used to simulate the irregular structure of the stratum) is obtained based on stratum analysis and hydrological experimental simulation. The reference value range is ; is the horizontal coordinate of the underground space; It is The geological disturbance at the monitoring point The phase frequency of the direction (used to control the superposition of fluctuations), the reference value range is , which is obtained based on the existing geological model and numerical fitting, is a technical means well known to those skilled in the art and will not be described in detail here; is the longitudinal coordinate of the underground space; It is The depth impedance factor of each monitoring point is used to control the adsorption and attenuation of pollution with depth. It is obtained by fitting the soil layer resistance experiment. The reference value range is , which are well known to those skilled in the art and will not be described in detail here; is the underground depth coordinate; It represents the net effective diffusion potential energy of the pollution source. It is the exponential scaling result of the coupling between the intensity of the pollution source itself (the pollutant concentration information data after pre-processing) and its geological release capacity (the soil permeability coefficient in the pre-processed geological exploration parameter data). It is the energy source term for the outward action of the pollution source. It is a spatial attenuation modulation term, which is used to simulate the nonlinear attenuation law of the pollution source's intensity in three-dimensional space as it changes with spatial position; The purpose of the above pollution action function is to transform the pre-processed pollutant concentration information data of the monitoring point into the influence function of the spatial deformation pollution field. Since the influence of each monitoring point is different, the attention weighting mechanism is introduced to assign weights to the pollution action function of each monitoring point. The attention weighting mechanism is a technical means well known to those skilled in the art and will not be described in detail here; Furthermore, the pollution action function is weighted based on the weight to construct the pollution potential field embedding tensor: , in, It is at the moment The pollution potential energy field is embedded in the tensor, which represents the dynamic pollution impact field with spatial coordinate dependence and is the basic input for path simulation and disturbance modeling. It is Monitoring points at time The weight of the pollution effect function; is the total number of monitoring points.

[0022] S2. Introduce a geological disturbance function generator to construct a spatial disturbance tensor, and combine it with the pollution potential field embedding tensor to generate a spatial correction tensor for pollution transmission; based on the pollution potential field embedding tensor and the spatial correction tensor for pollution transmission, perform pollution propagation modeling to obtain a pollution diffusion prediction tensor to realize groundwater pollution diffusion prediction.

[0023] The migration of groundwater in a medium does not follow the Euclidean law of continuity, but is instead influenced by the coupling of nonlinear factors such as stratum boundaries, porosity changes, and permeability. Therefore, to accurately simulate the nonlinear propagation path of pollution underground, a geological perturbation function generator is introduced to model the perturbation behavior of underground structures during pollution diffusion. This generator constructs a spatial perturbation tensor based on a multi-perturbation coupling mechanism by setting a series of perturbation kernels, each of which simulates a nonlinear perturbation point at a spatial location: , in, is the spatial perturbation tensor, representing the three-dimensional space and time the intensity of geological disturbance; is the disturbance kernel index, which is used to accumulate multiple disturbance centers in the underground structure; is the total number of disturbance cores, indicating the number of disturbance centers. It is set according to the specific scenario and is not limited here. is the disturbance intensity time weight, indicating the The perturbation kernel at time The disturbance intensity is estimated based on the groundwater velocity, pressure head change rate and other fitting in the pre-processed geological exploration parameter data. The reference value range is The methods used are all well known to those skilled in the art and will not be described in detail here. is the disturbance angle, which indicates the deviation angle of the disturbance direction relative to the mainstream direction. It is calculated based on the hydraulic gradient direction and runoff angle in the pre-processed geological exploration parameter data. The reference value range is The calculation method is a technical means well known to those skilled in the art and will not be described in detail here; is the lateral perturbation kernel function, which is in the form of , indicating that from perturbation core center The influence radius radiating to any point; is the longitudinal disturbance function, defined as , used to simulate the attenuation of the impact of disturbances in the geological depth direction, It is The vertical center depth coordinate of each disturbance core in the underground space; is the space-time oscillation frequency factor, which is used to control the amplitude of the disturbance space fluctuation. It is estimated based on the underground structure disturbance spectrum in the pre-processed geological exploration parameter data. The reference value range is , which are well known to those skilled in the art and will not be described in detail here; It is the spatial attenuation function of the perturbation intensity, which is used to simulate the attenuation of the influence of the perturbation kernel on the spatial position; It is a space-time disturbance frequency modulation function, which is used to simulate the propagation oscillation phenomenon caused by the pollution path in the geological periodic structure, reflecting the actual disturbances such as seepage stratification alternation, reflection waves, underground pumping cycles, etc. is the tangent of the disturbance angle, which amplifies the nonlinear disturbance behavior by the angle and is used to simulate the degree of deflection of the pollution path; Based on the control theory model of the disturbance response tensor field and the reversible disturbance physical field construction method of the diffusion-counter-diffusion system, the spatial disturbance tensor is coupled with the aforementioned pollution potential energy field embedding tensor to generate the spatial correction tensor of pollution transmission: , in, is the spatial correction tensor of the pollution transport, representing the position and time The disturbance intensity coefficient of the underground pollution path; It is at the moment The contamination potential field is embedded in the tensor The spatial gradient of represents the local rate of change of the pollution potential energy in three-dimensional space. It is an important reference indicator of the main direction of the pollution diffusion path and the pollution diffusion capacity. It is obtained by taking the spatial partial derivative of the pollution potential energy field embedded tensor. This is a technical means well known to those skilled in the art and will not be described in detail here. It is the disturbance amplification coefficient, which is used to control the amplification effect of time and simulate the situation of disturbance enhancement in certain special time periods (such as sewage peak period, infiltration during heavy rain, etc.). It is determined according to the specific scenario and the reference value range is ; It is a time modulation term that intervenes in the disturbance response by introducing a time periodic factor to simulate the physical phenomenon that the disturbance intensity fluctuates over time; It is an exponential amplification term of the local disturbance intensity, which is used to simulate the nonlinear amplification effect of disturbance on the pollution propagation ability. That is, the stronger the disturbance, the more significant the impact on the deflection or acceleration of the diffusion path. It is similar to the phenomenon that the disturbance flux in the physical field increases nonlinearly with energy. is the response adjustment factor, which represents the resistance factor under disturbance adjustment and is used to simulate the response difficulty after the diffusion path is disturbed; Based on the above-mentioned pollution potential field embedding tensor and the spatial correction tensor of pollution transmission, pollution propagation modeling is performed; the pollution potential field embedding tensor and the spatial correction tensor of pollution transmission are subjected to path direction-aware convolution operation to obtain the pollution diffusion prediction tensor : , in, Indicates at time The evolution and distribution of pollutants in the three-dimensional underground space of the entire chemical park are described in the form of a tensor, which is the pollution diffusion prediction tensor. is the index of the diffusion path / channel; is the total number of diffusion paths / channels, which is set according to specific scenarios based on existing propagation mechanisms and is not limited here; is the diffusion channel weight factor, indicating the pollutant The propagation proportion or energy weight on the diffusion path is obtained by regressing and fitting the historical pollution monitoring data from the existing database. The reference value range is , which are well known to those skilled in the art and will not be described in detail here; It is The pollution potential field on the channel is embedded in the tensor, indicating that at the current moment , pollution in the The potential energy impact formed by the superposition of multiple monitoring points on the channel; It is a directional three-dimensional path convolution operator, a tensor propagation operator with a spatially adjustable kernel, directional filtering and weight field mapping, which is equivalent to the finite volume method in geophysical simulation or the discrete propagator in FEM numerical diffusion modeling. In tensor calculations, represents the path direction aware convolution operation, which is used to propagate the potential contamination field to the surrounding grid nodes through the kernel weight in a certain direction; It is Spatially corrected tensor of contamination transport on the channel; It is a diffusion frequency modulation factor used to simulate the periodic fluctuations of pollution diffusion on a specific path, such as diurnal changes, tidal influences, and sewage discharge cycles. The diffusion frequency modulation factor uses Fourier transform to analyze the time-frequency characteristics of the pre-processed pollutant concentration information data and extracts the main frequency as the diffusion frequency adjustment parameter. , and finally the diffusion process is dynamically modulated by the cosine function; the diffusion frequency adjustment parameter The reference value range is ; The Fourier transform is a technical means well known to those skilled in the art and will not be described in detail here.

[0024] In summary, an artificial intelligence-based prediction method for groundwater pollution diffusion in chemical parks was completed.

[0025] The order in which the embodiments of the invention are presented is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0026] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

[0027] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A method for predicting the spread of groundwater pollution in a chemical park based on artificial intelligence, characterized in that: The following steps are involved: S1. Obtain pollutant concentration information data and geological survey parameter data in the chemical park groundwater, and preprocess them to obtain preprocessed data; Based on the preprocessed data, a self-shaping function embedding mechanism is introduced to construct the pollution action function and generate the pollution potential field embedding tensor; S2. Introduce a geological disturbance function generator to construct a spatial disturbance tensor, and combine it with the pollution potential field embedding tensor to generate a spatial correction tensor for pollution transmission; based on the pollution potential field embedding tensor and the spatial correction tensor for pollution transmission, perform pollution propagation modeling to obtain a pollution diffusion prediction tensor to realize groundwater pollution diffusion prediction.

2. The method for predicting groundwater pollution spread in a chemical park based on artificial intelligence according to claim 1, characterized in that: Said S1 specifically includes: The preprocessed data includes preprocessed pollutant concentration information data and preprocessed geological exploration parameter data; in the process of implementing the self-shaping function embedding mechanism, based on the soil permeability coefficient in the preprocessed pollutant concentration information data and the preprocessed geological exploration parameter data, and introducing a spatial attenuation modulation term, a pollution action function is constructed to quantify the diffusion behavior of pollutants in three-dimensional space.

3. The method for predicting groundwater pollution spread in a chemical park based on artificial intelligence according to claim 2, characterized in that: Said S1 specifically includes: An attention weighting mechanism is introduced to perform weighted processing on the pollution action function and construct an embedding tensor of the pollution potential energy field.

4. The method for predicting groundwater pollution spread in a chemical park based on artificial intelligence according to claim 3, characterized in that: Said S2 specifically includes: The geological disturbance function generator simulates nonlinear disturbance points at spatial positions by setting a disturbance kernel.

5. The method for predicting groundwater pollution spread in a chemical park based on artificial intelligence according to claim 4, characterized in that: Said S2 specifically includes: Based on the preprocessed geological exploration parameter data, the disturbance intensity time weight, disturbance angle and space-time oscillation frequency factor are obtained; based on the disturbance angle and space-time oscillation frequency factor, the spatial attenuation function of the disturbance intensity and the space-time disturbance frequency modulation function are constructed, and combined with the disturbance intensity time weight, the spatial disturbance tensor is constructed.

6. The method for predicting groundwater pollution spread in a chemical park based on artificial intelligence according to claim 5, characterized in that: Said S2 specifically includes: Based on the spatial disturbance tensor and the pollution potential field embedding tensor, an exponential function and a time modulation term are introduced to generate a spatial correction tensor of pollution transmission.

7. The method for predicting groundwater pollution spread in a chemical park based on artificial intelligence according to claim 6, characterized in that: Said S2 specifically includes: The diffusion channel weight factor is introduced and combined with the diffusion frequency modulation factor, and path direction-aware convolution operation is performed on the pollution potential field embedding tensor and the spatial correction tensor of pollution transmission to obtain the pollution diffusion prediction tensor.

8. The method for predicting groundwater pollution spread in a chemical park based on artificial intelligence according to claim 7, characterized in that: Said S2 specifically includes: The diffusion frequency modulation factor dynamically modulates the diffusion process by analyzing the time-frequency characteristics of the pre-processed pollutant concentration information data and combining it with the cosine function.

Images