Mine underground water pollution multi-element sensing and intelligent early warning method and system

Abstract

The application discloses a mine underground water pollution multi-element sensing and intelligent early warning method and system, and the method comprises the following steps: constructing a mine underground water pollution full-element sensing network, collecting pollution source, hydrogeology, underground water dynamic monitoring and environmental correlation data, and storing the standardized data into a unified database. The pollution source, monitoring well and hydrogeology zoning data vector layers are superimposed to generate a pollution risk distribution base containing the correlation of "pollution source-aquifer-monitoring point". A migration model coupling pollution dissolution and seepage is established to quantize the influence weight of elements on pollution diffusion. The pollution concentration time sequence is extracted, and the real-time data sliding window is combined to output the pollution diffusion path, concentration peak value and sensitive target time to arrival by using LSTM. Based on the risk base and the prediction result, the sensitive target grading is combined to form a research conclusion, and the key prevention and control areas and key nodes are determined. The application provides a scientific basis for mine pollution prevention and control.

Description

Technical Field

[0001] This invention belongs to the field of monitoring and early warning technology for groundwater pollution in mines, and more specifically, relates to a method and system for multi-element perception and intelligent early warning of groundwater pollution in mines. Background Technology

[0002] With the long-term operation of mining activities, a large amount of mine pollutants pose a serious threat to the groundwater environment. The pollutants in mining areas are complex in type and scattered in distribution, and can easily spread to surrounding areas through groundwater seepage, disrupting the balance of the groundwater ecosystem, affecting the safety of drinking water for residents and the quality of agricultural irrigation, while also hindering the process of regional ecological restoration, posing a huge challenge to ecological and environmental governance.

[0003] The core challenge in current mine groundwater pollution control lies in the lack of a systematic perception and early warning system. Existing monitoring technologies are mostly limited to single indicators or localized areas, failing to achieve coordinated capture of multiple dimensions such as pollution sources, hydrogeological conditions, and environmental influencing factors. This makes it difficult to fully grasp the complete chain of pollution generation and migration. Fragmented monitoring data lacks standardized integration, and the incompatibility between different types of data is poor, failing to provide comprehensive and reliable data support for pollution risk analysis, thus affecting the scientific rigor of pollution assessment.

[0004] In terms of pollution prediction and control, existing methods struggle to depict the complex coupling relationship between pollutant dissolution, seepage, and migration. Predictions of pollution diffusion trends often rely on empirical judgments or simple models, resulting in insufficient accuracy and timeliness. They are unable to predict pollution peaks, diffusion paths, and impacts on sensitive targets in advance. Furthermore, existing technologies lack differentiated consideration for sensitive targets, making it difficult to delineate key control areas and critical nodes. This leads to irrational allocation of governance resources and insufficient targeting and effectiveness in pollution control.

[0005] Therefore, constructing a set of technical methods that take into account multiple factors of perception, prediction and early warning, and solve the problems of current single monitoring dimensions, insufficient prediction accuracy and weak control, is of great practical significance and application value for improving the initiative of mine groundwater pollution control, ensuring regional groundwater ecological security and promoting high-quality restoration of mine ecological environment. Summary of the Invention

[0006] This invention aims to overcome the technical challenges of limited monitoring dimensions, difficulty in data integration, and insufficient prediction accuracy in mine groundwater pollution monitoring. By employing multi-element collaborative sensing, coupled model analysis, time-series prediction, and hierarchical assessment, it enables the identification and early warning of pollution risks, providing a scientific basis for pollution prevention and control and ensuring the ecological security of regional groundwater.

[0007] To address the aforementioned deficiencies or improvement needs of existing technologies, as a first aspect of this invention, the present invention provides a method for multi-element perception and intelligent early warning of groundwater pollution in mines, comprising:

[0008] S1. Construct a sensing network covering all elements of groundwater pollution in mines, collect pollution source data, hydrogeological data, groundwater dynamic monitoring data and environmental correlation data from mines, and store all data in a unified database after standardization;

[0009] S2. The spatial coordinates of pollution sources, the location of monitoring wells, and the boundary data of hydrogeological zones are overlaid in the geographic information system to generate a base map of groundwater pollution risk distribution in the mining area that includes the spatial relationship between "pollution source-aquifer-monitoring point".

[0010] S3. Establish a pollutant migration model that couples the dissolution rate of mine pollutants with the characteristics of groundwater seepage field. By using the control variable method, correlate the dynamic response relationship between the stock of pollution sources, the permeability coefficient of aquifers and the concentration of characteristic pollutants in groundwater. Based on the analytic hierarchy process, quantify the influence weight of each factor on pollution diffusion.

[0011] S4. Extract the pollution concentration time series from historical monitoring data, combine it with the sliding window characteristics of real-time data, and use the Long Short-Term Memory (LSTM) time series analysis method to output the diffusion path, concentration peak and time prediction of groundwater pollution reaching sensitive targets in the future prediction period.

[0012] S5. Based on the pollution risk distribution baseline and diffusion prediction results, and combined with the priority classification of sensitive target protection, a comprehensive judgment conclusion on mine groundwater pollution is formed, and key areas and critical control nodes for pollution prevention and control are identified.

[0013] Furthermore, the construction process of the full-element perception network in S1 is as follows:

[0014] With the diffusion path and impact range of groundwater pollution in mines as the core, a three-level sensing node system of "pollution source - migration path - sensitive target" is set up; fixed monitoring nodes are set up in the mine pollution area, aquifer recharge area and runoff zone, and mobile supplementary monitoring nodes are set up around sensitive targets to form a sensing network layout with full coverage and key densification.

[0015] The fixed monitoring node is equipped with a multi-parameter water quality sensor, a water level sensor, and a flow sensor, while the mobile supplementary monitoring node is equipped with a portable rapid water quality analyzer.

[0016] By combining wired and wireless communication technologies, the system enables real-time data collection and transmission from each monitoring node. It also integrates existing survey data, hydrogeological exploration data, and environmental data including regional meteorology and surface runoff from the mining area, thus completing the construction of a comprehensive sensing network.

[0017] Furthermore, the pollution source data, hydrogeological data, groundwater dynamic monitoring data, and environmental correlation data in S1 are as follows:

[0018] Pollution source data includes basic attributes, pollutant characteristics, and spatial distribution correlation data of pollution sources related to groundwater pollution in and around mines;

[0019] Hydrogeological data includes geological structural parameters, hydrodynamic conditions, and background water quality characteristics related to groundwater occurrence and migration in the mining area;

[0020] Groundwater dynamic monitoring data includes real-time dynamic data on the physical and chemical properties of groundwater, concentrations of characteristic pollutants, and hydrodynamic parameters collected through a sensing network.

[0021] Environmental data includes meteorological, surface, soil, land use, and sensitive target protection data that affect the generation and migration of groundwater pollution in mining areas.

[0022] Furthermore, the generation process of the water pollution risk distribution base in S2 is as follows:

[0023] Based on a unified geographic coordinate system, let the first layer of the pollution source spatial coordinate vector map be defined. The spatial coordinates of the pollution sources are ,in , They are respectively Cartesian coordinates. The elevation of the pollution source; the first layer in the hydrogeological zoning boundary vector map layer. The set of vertex coordinates of the boundary polygons of each hydrogeological zone is as follows: , The number of boundary vertices; the first in the vector layer of the monitoring well deployment location. The spatial coordinates of each monitoring well are ,in These represent the abscissa, ordinate, and elevation of the wellhead.

[0024] Based on the formula for determining the attribution of spatial points and surfaces Determine the coordinates of the pollution source and monitoring well coordinates Does it fall in the first place? Hydrogeological zones Within the boundary polygon, where The coordinates of the point to be determined are in the plane. When the formula holds true, it indicates that the point to be determined belongs to this hydrogeological zone.

[0025] Using the formula for calculating spatial distance Calculate the spatial straight-line distance between pollution sources belonging to the same hydrogeological zone and monitoring wells. ;

[0026] Based on the attribution determination results and spatial distance calculation results, a groundwater pollution risk distribution baseline for the mining area is integrated, which includes the spatial correlation between "pollution source - hydrogeological zone - monitoring well". The attribution determination results represent the spatial subordination relationship among these three elements, and the spatial distance... Characterizes the spatial proximity between pollution sources and monitoring wells.

[0027] Furthermore, the pollutant migration model in S3 is specifically as follows:

[0028] First, taking the aquifer control volume as the research object, according to the law of conservation of mass, the rate of change of pollutant mass within the control volume is equal to the algebraic sum of the divergence of pollutant dispersion flux, the divergence of convective flux, and source-sink terms. Based on this, a pollutant mass balance equation is established: ,in For total pollutant flux, Porosity is the porosity of an aquifer, which characterizes the proportion of pores in the total volume of the aquifer. The concentration of characteristic pollutants in groundwater; For time; It is a vector differential operator that characterizes the spatial gradient change;

[0029] Second, total pollutant flux Composed of diffuse flux and convective flux, according to Fick's law and convective transport theory, the expression for diffuse flux is: The expression for circulation volume is: ,in The pollutant dispersion coefficient is determined by the aquifer particle composition, pore structure, and groundwater seepage velocity. The actual seepage velocity of groundwater reflects the speed of groundwater movement within the aquifer; therefore, the total flux... ;

[0030] Furthermore, considering the leaching and replenishment effects of mine pollutants, a source-sink project is introduced. Characterizing the leaching intensity of pollutants, and combining the above mass balance equation and total flux expression, the core expression of the pollutant migration model is derived:

[0031]

[0032] Among them, the source and sink items Based on the kinetics of pollutant leaching, a quantitative derivation is made: the amount of pollutant leached is positively correlated with the leaching rate constant, total mass, and contact area, and negatively correlated with the groundwater volume of the hydrogeological unit. , The leaching rate constant of pollutants is determined by the physicochemical properties of the pollutants, the temperature and humidity of the surrounding environment, and the groundwater immersion conditions. The total mass of mine pollutants; The contact area between the pollutants and groundwater; The volume of groundwater in the hydrogeological unit where the pollutant is located;

[0033] Actual groundwater seepage velocity Based on Darcy's law: Darcy's law states that seepage velocity is positively correlated with permeability coefficient and hydraulic gradient. By correcting the actual seepage velocity using porosity, we obtain... , The aquifer permeability coefficient characterizes the water permeability of an aquifer. The hydraulic gradient is determined by the ratio of the groundwater level difference to the length of the water flow path within the hydrogeological zone.

[0034] Furthermore, the quantification process for the influence weights of each element on pollution diffusion in S3 is as follows:

[0035] First, a pairwise comparison judgment matrix is ​​constructed: based on the differences in the degree of influence of each indicator on pollution diffusion, the element values ​​of the judgment matrix are determined by comparing the relative importance of the indicators. These element values ​​are supported by pollution diffusion simulation experimental data; that is, with other indicators fixed, the change in the concentration of a characteristic pollutant caused by a change in a single indicator is observed using the controlled variable method. The ratio of the change in magnitude determines the relative importance level between the indicators, thus forming the judgment matrix. ,in To evaluate the number of indicators, Indicates the first The first indicator is relative to the first The relative importance values ​​of each indicator, and satisfying , ;

[0036] Secondly, calculate the eigenvectors and the largest eigenvalue of the judgment matrix: by solving the characteristic equation of the judgment matrix. To obtain the largest eigenvalue and the corresponding normalized feature vector The elements in the normalized eigenvector This refers to the initial weight values ​​of each evaluation indicator;

[0037] Finally, a consistency check is performed: the consistency index is calculated. And combined with the average random consistency index Obtain the consistency ratio ;like If the judgment matrix satisfies the consistency requirement, the corresponding normalized eigenvector is the final impact weight of each element on pollution diffusion.

[0038] If the consistency requirement is not met, the values ​​of the judgment matrix elements are adjusted based on the pollution diffusion simulation experiment data, and the above steps are repeated until the consistency requirement is met, ultimately realizing the quantification of the influence weight of each element.

[0039] Furthermore, the prediction process for the prediction result in S4 is as follows:

[0040] First, extract the time series of characteristic pollutant concentrations from historical monitoring data, denoted as... ,in For historical monitoring moments; construct a sliding window from real-time monitoring data at fixed time steps, with the first [time point] within the window being [the next data point]. The input feature vector at each time step is ,in The length of the sliding window. The pollutant concentration at that time step. This represents the groundwater seepage velocity at the corresponding time. The hydraulic gradient at the corresponding time point is obtained from the parameters of the pollutant migration model, thus forming the input sample set. , The total sample size is given; the sample set is then normalized using the following formula: ,in , The minimum and maximum values ​​of the corresponding features in the sample set are respectively used to obtain the standardized sample set. ;

[0041] Secondly, construct an LSTM model, where the hidden layer state update formula is: ,in This is the weight matrix from the input layer to the hidden layer. This is the weight matrix from hidden layer to hidden layer. , These are the corresponding bias vectors. It is the Sigmoid activation function. Let be the hyperbolic tangent activation function, and o be the Hadamard product. The hidden layer state is the value from the previous time step; a standardized sample set is input into the model for training, and the model outputs predicted pollutant concentration values ​​for consecutive future time steps. , The predicted time step number is used to predict the time span, which covers a preset duration.

[0042] Finally, based on the baseline of groundwater pollution risk distribution in the mining area, spatial interpolation formulas were used. Calculate any spatial location The concentration of pollutants, among which For the spatial coordinates of the monitoring point, For position and The spatial distance is obtained from the spatial distance formula. The number of monitoring points; through Determine the peak concentration, corresponding to the time point. Based on the spatial coordinates of sensitive targets Solve the equation Obtain the time it takes for pollutants to reach sensitive targets , For sensitive target pollutant concentration thresholds, the final integration Spatial distribution , and This forms the prediction results.

[0043] Furthermore, the priority classification of sensitive target protection in S5 is based on the functional attributes, spatial correlation characteristics, and risk exposure degree of sensitive targets, combined with the pollution risk distribution base and diffusion prediction results, to construct a multi-dimensional classification system; taking the functional importance of the sensitive target itself as the core basis, it is associated with its spatial location relationship with the pollution source and the degree of pollution threat to form a priority gradient sequence, and high, medium and low protection levels are divided accordingly, providing a priority judgment benchmark for the delineation of key pollution prevention and control areas and key control nodes.

[0044] As a second aspect of the present invention, the present invention also provides a multi-element sensing and intelligent early warning system for groundwater pollution in mines, comprising:

[0045] The full-element sensing network construction unit is used to build a sensing network covering all elements of groundwater pollution in mines, collect pollution source data, hydrogeological data, groundwater dynamic monitoring data and environmental correlation data in mines, and store all data in a unified database after standardization.

[0046] The risk distribution base generation unit is used to overlay vector layers on the spatial coordinates of pollution sources, the location of monitoring wells, and the boundary data of hydrogeological zones in a geographic information system to generate a groundwater pollution risk distribution base in the mining area that includes the spatial relationship between "pollution source-aquifer-monitoring point".

[0047] The weighted quantification unit is used to establish a pollutant migration model that couples the dissolution rate of mine pollutants with the characteristics of groundwater seepage field. It uses the control variable method to correlate the dynamic response relationship between the stock of pollution sources, the permeability coefficient of aquifers and the concentration of characteristic pollutants in groundwater, and quantifies the influence weight of each factor on pollution diffusion based on the analytic hierarchy process.

[0048] The pollution prediction unit is used to extract the time series of pollution concentration from historical monitoring data, combine the sliding window characteristics of real-time data, and use the Long Short-Term Memory (LSTM) time series analysis method to output the predicted diffusion path, concentration peak and time of arrival of groundwater pollution to sensitive targets in the future prediction period.

[0049] The risk assessment unit is used to form a comprehensive assessment conclusion on mine groundwater pollution based on the pollution risk distribution base and diffusion prediction results, combined with the priority classification of sensitive target protection, and to identify key areas and critical control nodes for pollution prevention and control.

[0050] As a third aspect of the invention, the invention also provides a computer-readable storage medium having a computer program stored thereon, the computer program being executed by a processor of any one of the methods for multi-element perception and intelligent early warning of groundwater pollution in mines.

[0051] In summary, compared with the prior art, the above-described technical solutions conceived by this invention can achieve the following beneficial effects:

[0052] 1. The multi-element perception and intelligent early warning method for mine groundwater pollution of the present invention constructs a perception network covering all pollution elements, collects pollution source, hydrogeological, groundwater dynamic monitoring and environmental correlation data and completes standardized storage, providing comprehensive data source support for subsequent pollution risk analysis. This perception network breaks through the limitations of traditional monitoring data being single and having limited coverage, realizing full-domain monitoring of mine pollution sources, migration paths and surrounding environmental influencing factors. The standardized multi-source data can eliminate format barriers between different data types, ensuring the compatibility and effectiveness of data in subsequent model calculation and analysis, laying a solid data foundation for pollution risk assessment.

[0053] 2. The multi-factor perception and intelligent early warning method for mine groundwater pollution of the present invention generates a spatial correlation base for pollution risk through multi-dimensional vector layer overlay. Combined with a pollutant migration model coupled with leaching and seepage, it quantifies the influence weight of each factor on pollution diffusion, and constructs a spatial-mechanistic dual analysis system for pollution risk. The vector layer overlay realizes the spatial correlation mapping of pollution sources, aquifers and monitoring points, clearly presenting the spatial distribution characteristics of pollution risk. The coupled migration model is derived based on the law of conservation of mass, depicting the dynamic correlation process of pollutant leaching, seepage and migration. The factor weights quantified by the analytic hierarchy process clarify the contribution of each influencing factor to pollution diffusion, providing core model support and quantitative basis for the scientific prediction of pollution diffusion trends.

[0054] 3. The multi-element perception and intelligent early warning method for mine groundwater pollution of the present invention mines historical and real-time monitoring data using LSTM time-series analysis, outputting prediction results of pollution diffusion paths, concentration peaks, and arrival times at sensitive targets. Combined with the priority classification of sensitive targets, a comprehensive judgment conclusion is formed, clearly identifying key areas and critical control nodes for pollution prevention and control. The LSTM model can effectively capture long-term dependencies in time-series data, improving the accuracy and reliability of pollution prediction. The priority classification of sensitive targets delineates protection sequences based on functional attributes and risk exposure levels, ensuring that prevention and control resources are tilted towards high-value, high-risk targets. The final comprehensive judgment conclusion can provide an intuitive and feasible decision-making reference for the prevention and scientific management of mine groundwater pollution. Attached Figure Description

[0055] Figure 1 This is a flowchart of the multi-element perception and intelligent early warning method for mine groundwater pollution according to an embodiment of the present invention;

[0056] Figure 2 This is a schematic diagram of groundwater pollution simulation and prediction according to an embodiment of the present invention;

[0057] Figure 3 This is a system unit diagram of an embodiment of the present invention;

[0058] Figure 4 This is a schematic diagram of the system pollution source management interface according to an embodiment of the present invention;

[0059] Figure 5 This is a schematic diagram of online groundwater monitoring and early warning according to an embodiment of the present invention;

[0060] Figure 6 This is a panoramic schematic diagram of the system's integrated groundwater decision-making in an embodiment of the present invention. Detailed Implementation

[0061] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0062] Example 1

[0063] Please refer to Figure 1 This embodiment 1 provides a method for multi-element perception and intelligent early warning of groundwater pollution in mines, including:

[0064] S1. Construct a sensing network covering all elements of groundwater pollution in mines, collect pollution source data, hydrogeological data, groundwater dynamic monitoring data and environmental correlation data from mines, and store all data in a unified database after standardization;

[0065] S2. The spatial coordinates of pollution sources, the location of monitoring wells, and the boundary data of hydrogeological zones are overlaid in the geographic information system to generate a base map of groundwater pollution risk distribution in the mining area that includes the spatial relationship between "pollution source-aquifer-monitoring point".

[0066] S3. Establish a pollutant migration model that couples the dissolution rate of mine pollutants with the characteristics of groundwater seepage field. By using the control variable method, correlate the dynamic response relationship between the stock of pollution sources, the permeability coefficient of aquifers and the concentration of characteristic pollutants in groundwater. Based on the analytic hierarchy process, quantify the influence weight of each factor on pollution diffusion.

[0067] S4. Extract the pollution concentration time series from historical monitoring data, combine it with the sliding window characteristics of real-time data, and use the Long Short-Term Memory (LSTM) time series analysis method to output the diffusion path, concentration peak and time prediction of groundwater pollution reaching sensitive targets in the future prediction period.

[0068] S5. Based on the pollution risk distribution baseline and diffusion prediction results, and combined with the priority classification of sensitive target protection, a comprehensive judgment conclusion on mine groundwater pollution is formed, and key areas and critical control nodes for pollution prevention and control are identified.

[0069] This embodiment 1 further elaborates on the above steps.

[0070] (1) Construction of a full-element perception network

[0071] Given that the sources of groundwater pollution in mines are dispersed, the migration paths are complex, and the impact range is wide, in order to achieve comprehensive capture of pollution-related elements, it is necessary to first build a sensing network covering all elements to provide support for subsequent data integration and analysis.

[0072] The construction process is based on the diffusion path and potential impact range of groundwater pollution in mines. It adopts a three-tiered sensing node deployment scheme of "pollution source - migration path - sensitive target," forming a three-dimensional monitoring layout with full coverage and increased density in key areas. Fixed monitoring nodes are deployed in high-pollution and critical migration areas such as the core pollution area of ​​the mine, aquifer recharge area, and runoff zone to achieve continuous and stable monitoring of pollution sources and migration processes. Mobile supplementary monitoring nodes are deployed around sensitive targets such as residential areas and drinking water sources to fill coverage gaps of fixed nodes and address emergency monitoring needs in the event of sudden pollution.

[0073] Fixed monitoring nodes are uniformly equipped with multi-parameter water quality sensors, water level sensors, and flow sensors, which can capture core indicators such as the physical and chemical properties of groundwater and hydrodynamic parameters in real time; mobile supplementary monitoring nodes are equipped with portable rapid water quality analyzers, which facilitate the rapid acquisition of key data such as pollutant concentrations and improve monitoring flexibility.

[0074] Data transmission employs a combination of wired and wireless communication technologies. Wired transmission ensures stable data transmission from fixed nodes, while wireless communication adapts to the dynamic monitoring needs of mobile nodes, enabling real-time data acquisition and synchronous transmission from each monitoring node. Simultaneously, existing survey data, hydrogeological exploration results, and environmental data related to regional meteorology and surface runoff are integrated to form a multi-source data collaborative acquisition system, ultimately completing the construction of a comprehensive sensing network.

[0075] The collected data were categorized and organized by attribute to ensure relevance and completeness. Pollution source data covered the basic attributes of pollution sources related to groundwater pollution in and around mines, including pollutant types, concentrations, and spatial distribution. Basic pollution source attributes included the location, formation time, and volume of mine pits, tailings ponds, and waste dumps. Pollutant types included characteristic pollutants such as heavy metals, acids and alkalis, and soluble inorganic salts. Pollutant concentrations were the measured values ​​and concentration gradients of different pollutants at each pollution point. Spatial distribution information clarified the spatial relationships between various pollution sources and surrounding aquifers and groundwater runoff paths, thereby identifying the characteristics of the pollution sources.

[0076] Hydrogeological data includes geological structural parameters, hydrodynamic conditions, and background water quality characteristics related to groundwater occurrence and migration in the mining area. Geological structural parameters include information on lithology of the strata, aquifer type and thickness, distribution characteristics of aquitards, and the degree of fault and fracture development in the mining area. Hydrodynamic conditions cover key parameters such as groundwater level, flow direction, flow velocity, and recharge-discharge relationships. Background water quality characteristics include physicochemical indicators and background value ranges of groundwater in uncontaminated areas, providing a solid geological and hydrological foundation for analyzing pollution migration patterns.

[0077] Groundwater dynamic monitoring data consists of real-time data collected through a sensing network, detailing the dynamic changes in the physicochemical properties, characteristic pollutant concentrations, and hydrodynamic parameters of groundwater. The physicochemical properties include indicators such as water temperature, pH value, conductivity, and dissolved oxygen; the characteristic pollutant concentrations are the real-time data on the content of pollutants such as heavy metals and organic matter detected at each monitoring point; and the hydrodynamic parameters include real-time data on changes in water level, flow velocity, and flow direction, dynamically reflecting the real-time state and migration trends of pollution.

[0078] Environmental data includes information related to meteorology, surface conditions, soil, land use, and the protection of sensitive targets. Meteorological data covers regional rainfall, rainfall intensity, temperature, and evaporation, used to analyze the impact of atmospheric precipitation on groundwater recharge and pollutant leaching and migration. Surface data includes information on surface runoff paths, flow rates, and catchment areas. Soil data includes parameters such as soil type, porosity, and permeability. Land use data covers the distribution of land types such as cultivated land, construction land, and ecological protection zones in and around mining areas. Information related to the protection of sensitive targets includes the location, extent, and protection level of drinking water sources, residential areas, and ecological protection zones.

[0079] These data directly influence the intensity and migration pathways of groundwater pollution in mining areas, providing support for the analysis of pollution influencing factors. After all data is collected, it needs to be standardized to unify data format and accuracy, eliminate compatibility barriers between data from different sources and of different types, and finally stored in a unified database to achieve centralized data management and efficient retrieval.

[0080] (2) Risk distribution base generation

[0081] Considering that the spatial migration characteristics of groundwater pollution in mines are closely related to hydrogeological conditions, in order to intuitively present the spatial distribution pattern of pollution risks, it is necessary to rely on geographic information systems to complete the overlay of vector layers of multiple types of spatial data, and then generate a pollution risk distribution base that includes the spatial correlation between pollution sources, aquifers and monitoring points.

[0082] Before performing layer overlay operations, it is necessary to unify the geographic coordinate system of all spatial data to ensure the consistency of spatial information from different sources. For the spatial coordinate vector map layer of pollution sources, it is necessary to clarify the Cartesian coordinates and elevation of each pollution source to locate the spatial location of the pollution source. For the boundary vector map layer of hydrogeological zones, it is necessary to sort out the vertex coordinates of the boundary polygons of each hydrogeological zone to define the scope of different hydrogeological units. The stratigraphic lithology and aquifer structure of these units are different, which will directly affect the migration path of pollutants. For the vector map layer of monitoring well locations, it is necessary to mark the plane coordinates and elevation of each monitoring wellhead to clarify the spatial location of the monitoring point.

[0083] Specifically, based on a unified geographic coordinate system, let the first layer of the pollution source spatial coordinate vector map be defined. The spatial coordinates of the pollution sources are ,in , They are respectively Cartesian coordinates. The elevation of the pollution source; the first layer in the hydrogeological zoning boundary vector map layer. The set of vertex coordinates of the boundary polygons of each hydrogeological zone is as follows: , The number of boundary vertices; the first in the vector layer of the monitoring well deployment location. The spatial coordinates of each monitoring well are ,in These represent the abscissa, ordinate, and elevation of the wellhead.

[0084] Based on the formula for determining the attribution of spatial points and surfaces Determine the coordinates of the pollution source and monitoring well coordinates Does it fall in the first place? Hydrogeological zones Within the boundary polygon, where The coordinates of the point to be determined are in the plane. When the formula holds true, it indicates that the point to be determined belongs to this hydrogeological zone.

[0085] Using the formula for calculating spatial distance Calculate the spatial straight-line distance between pollution sources belonging to the same hydrogeological zone and monitoring wells. ;

[0086] Based on the attribution determination results and spatial distance calculation results, a groundwater pollution risk distribution baseline for the mining area is integrated, which includes the spatial correlation between "pollution source - hydrogeological zone - monitoring well". The attribution determination results represent the spatial subordination relationship among these three elements, and the spatial distance... Characterizes the spatial proximity between pollution sources and monitoring wells.

[0087] (3) Weight quantization

[0088] Considering that the migration process of groundwater pollution in mines is affected by the coupling of multiple factors such as pollutant leaching and groundwater seepage, it is difficult to characterize the pollution diffusion pattern by analyzing a single factor. It is necessary to establish a targeted pollutant migration model and quantify the influence weight of each factor on pollution diffusion to provide mechanistic support and quantitative basis for subsequent pollution prediction.

[0089] The model is constructed using the aquifer control volume as the research object. According to the law of conservation of mass, the rate of change of pollutant mass within the control volume is equal to the algebraic sum of the divergence of pollutant dispersion flux, the divergence of convective flux, and source-sink terms. Based on this, the pollutant mass balance equation is established: ,in For total pollutant flux, Porosity is the porosity of an aquifer, which characterizes the proportion of pores in the total volume of the aquifer. The concentration of characteristic pollutants in groundwater; For time; It is a vector differential operator that characterizes the spatial gradient change;

[0090] Second, total pollutant flux Composed of diffuse flux and convective flux, according to Fick's law and convective transport theory, the expression for diffuse flux is: The expression for circulation volume is: ,in The pollutant dispersion coefficient is determined by the aquifer particle composition, pore structure, and groundwater seepage velocity. The actual seepage velocity of groundwater reflects the speed of groundwater movement within the aquifer; therefore, the total flux... ;

[0091] Furthermore, considering the leaching and replenishment effects of mine pollutants, a source-sink project is introduced. Characterizing the leaching intensity of pollutants, and combining the above mass balance equation and total flux expression, the core expression of the pollutant migration model is derived:

[0092]

[0093] Among them, the source and sink items Based on the kinetics of pollutant leaching, a quantitative derivation is made: the amount of pollutant leached is positively correlated with the leaching rate constant, total mass, and contact area, and negatively correlated with the groundwater volume of the hydrogeological unit. , The leaching rate constant of pollutants is determined by the physicochemical properties of the pollutants, the temperature and humidity of the surrounding environment, and the groundwater immersion conditions. The total mass of mine pollutants; The contact area between the pollutants and groundwater; The volume of groundwater in the hydrogeological unit where the pollutant is located;

[0094] Actual groundwater seepage velocity Based on Darcy's law: Darcy's law states that seepage velocity is positively correlated with permeability coefficient and hydraulic gradient. By correcting the actual seepage velocity using porosity, we obtain... , The aquifer permeability coefficient characterizes the water permeability of an aquifer. The hydraulic gradient is determined by the ratio of the groundwater level difference to the length of the water flow path within the hydrogeological zone.

[0095] To clarify the contribution of each key element to pollution diffusion, the influence weights are quantified based on the analytic hierarchy process. The evaluation index system selects pollution source stock, aquifer permeability coefficient, pollutant dissolution rate constant, and groundwater hydraulic gradient, all of which are derived from the inherent parameters and related factors of the migration model, ensuring a logical closed loop with the model construction.

[0096] First, a pairwise comparison judgment matrix is ​​constructed: based on the differences in the degree of influence of each indicator on pollution diffusion, the element values ​​of the judgment matrix are determined by comparing the relative importance of the indicators. These element values ​​are supported by pollution diffusion simulation experimental data; that is, with other indicators fixed, the change in the concentration of a characteristic pollutant caused by a change in a single indicator is observed using the controlled variable method. The ratio of the change in magnitude determines the relative importance level between the indicators, thus forming the judgment matrix. ,in To evaluate the number of indicators, Indicates the first The first indicator is relative to the first The relative importance values ​​of each indicator, and satisfying , ;

[0097] Secondly, calculate the eigenvectors and the largest eigenvalue of the judgment matrix: by solving the characteristic equation of the judgment matrix. To obtain the largest eigenvalue and the corresponding normalized feature vector The elements in the normalized eigenvector This refers to the initial weight values ​​of each evaluation indicator;

[0098] Finally, a consistency check is performed: the consistency index is calculated. And combined with the average random consistency index Obtain the consistency ratio ;like If the judgment matrix satisfies the consistency requirement, the corresponding normalized eigenvector is the final impact weight of each element on pollution diffusion.

[0099] If the consistency requirement is not met, the values ​​of the judgment matrix elements are adjusted based on the pollution diffusion simulation experiment data, and the above steps are repeated until the consistency requirement is met, ultimately realizing the quantification of the influence weight of each element.

[0100] The process of observing the magnitude of characteristic pollutant concentration changes caused by changes in a single indicator using the controlled variable method is as follows: Based on a migration model coupling pollutant leaching and groundwater seepage characteristics, when using the controlled variable method to correlate these relationships, the core variables, response variables, and control variables are first identified. The core variables are the stock of the pollution source and the aquifer permeability coefficient; the response variable is the concentration of characteristic pollutants in the groundwater. Simultaneously, parameters such as the pollutant leaching rate constant, hydraulic gradient, and aquifer porosity are set to fixed values ​​based on measured hydrogeological data from the mining area. A simulation scenario closely matching actual working conditions is built to eliminate interference from irrelevant variables in the correlation analysis, ensuring the relevance and reliability of the experimental results.

[0101] The experiment was conducted in two phases: In the first phase, the aquifer permeability coefficient was fixed as the measured baseline value, and the stock of low, medium, and high-gradient pollution sources covering the actual area of ​​the mining area was selected. The model was run to monitor the changes in pollutant concentration at different time points, and the correlation between stock and concentration peak, stable value, and trend was analyzed. In the second phase, the stock of pollution sources was fixed as the baseline gradient, and the permeability coefficients of aquifers with weak, medium, and strong permeability were selected. The concentration rise rate and peak occurrence time were monitored simultaneously to clarify the influence of permeability coefficient on pollutant diffusion rate and concentration distribution. Finally, by comparing the data from the two phases, the dynamic response relationship between the core variable and the response variable was fully established.

[0102] (4) Pollution prediction

[0103] Considering the dynamic changes in groundwater pollution concentration over time, and the combined influence of factors such as seepage velocity and hydraulic gradient, relying solely on historical data is insufficient to predict future pollution trends. Therefore, a combination of time-series analysis methods and spatial correlation information is needed to achieve multi-dimensional prediction of pollution diffusion. The prediction process is based on historical and real-time monitoring data, leveraging the temporal feature capture capabilities of Long Short-Term Memory (LSTM) networks and incorporating spatial interpolation techniques to output complete pollution prediction results.

[0104] First, extract the time series of characteristic pollutant concentrations from historical monitoring data, denoted as... ,in For historical monitoring moments; construct a sliding window from real-time monitoring data at fixed time steps, with the first [time point] within the window being [the next data point]. The input feature vector at each time step is ,in The length of the sliding window. The pollutant concentration at that time step. This represents the groundwater seepage velocity at the corresponding time. The hydraulic gradient at the corresponding time point is obtained from the parameters of the pollutant migration model, thus forming the input sample set. , The total sample size is given; the sample set is then normalized using the following formula: ,in , The minimum and maximum values ​​of the corresponding features in the sample set are respectively used to obtain the standardized sample set. ;

[0105] Secondly, construct an LSTM model, where the hidden layer state update formula is: ,in This is the weight matrix from the input layer to the hidden layer. This is the weight matrix from hidden layer to hidden layer. , These are the corresponding bias vectors. It is the Sigmoid activation function. Let be the hyperbolic tangent activation function, and o be the Hadamard product. The hidden layer state is the value from the previous time step; a standardized sample set is input into the model for training, and the model outputs predicted pollutant concentration values ​​for consecutive future time steps. , The predicted time step number is used to predict the time span, which covers a preset duration.

[0106] Finally, based on the baseline of groundwater pollution risk distribution in the mining area, spatial interpolation formulas were used. Calculate any spatial location The concentration of pollutants, among which For the spatial coordinates of the monitoring point, For position and The spatial distance is obtained from the spatial distance formula. The number of monitoring points; through Determine the peak concentration, corresponding to the time point. Based on the spatial coordinates of sensitive targets Solve the equation Obtain the time it takes for pollutants to reach sensitive targets , For sensitive target pollutant concentration thresholds, the final integration Spatial distribution , and This forms the prediction results.

[0107] like Figure 2 As shown, this groundwater pollution simulation and prediction diagram presents the core data and results of the pollution prediction process in this embodiment. The left side of the diagram shows a three-dimensional geological structure model of the study area, clearly displaying the stratigraphic distribution, spatial morphology, and contact relationships between sandy clay, silty clay, and silty clay. This provides a geological medium basis for spatial interpolation calculations during the pollution prediction process, clarifying the stratigraphic environment in which pollutants migrate. The right side shows a spatial distribution cloud map of pollutant concentrations, using color gradients to distinguish different pollutant concentration ranges. This corresponds to the spatial distribution results of pollutants obtained after time-series analysis and spatial interpolation processing using a Long Short-Term Memory (LSTM) network. It intuitively presents the diffusion range, concentration gradient distribution, and spatial location of concentration peak areas of pollutants, corresponding to the pollution diffusion paths and concentration peaks output in the pollution prediction process described above. This fully presents the core basic conditions and key outputs of the pollution prediction process.

[0108] (5) Risk assessment

[0109] To effectively manage groundwater pollution in mines, it is necessary to integrate the pollution risk distribution baseline and diffusion prediction results, combined with the priority classification of sensitive targets, to form a scientific and comprehensive assessment conclusion, clarify the key points and critical nodes for prevention and control, and provide a basis for pollution control decisions. The classification and assessment process must take into account the intrinsic value of sensitive targets, their spatial location, and the degree of pollution threat, ensuring that prevention and control resources are tilted towards high-value, high-risk areas.

[0110] A multi-dimensional evaluation system for prioritizing the protection of sensitive targets is constructed, with the functional attributes of sensitive targets as the core basis, combined with spatial correlation characteristics and risk exposure levels for comprehensive judgment. At the functional attribute level, based on the differences in ecological and social value, drinking water sources and core residential areas are designated as core protection types, followed by cultivated land and general ecological areas, clarifying the basic protection weights for different targets;

[0111] Based on the pollution risk distribution baseline, the spatial correlation characteristics are analyzed to determine the spatial distance between sensitive targets and pollution sources, as well as the pollution susceptibility of the hydrogeological units in which they are located. Targets that are closer to pollution sources and have a higher pollution susceptibility are given higher protection priority.

[0112] By combining risk exposure levels with diffusion prediction results, the probability of a target being affected by pollutants, the expected pollution concentration, and the duration of exposure are quantified. Targets that are predicted to be affected by high concentrations of pollution and have a longer exposure period are further prioritized. Through the synergistic analysis of these three factors, a three-tiered protection sequence of high, medium, and low is formed, providing a clear priority benchmark for the assessment work.

[0113] In the comprehensive assessment phase, the classification results are deeply integrated with the pollution risk distribution baseline and diffusion prediction results. The pollution diffusion path coverage area and the range of influence of the concentration peak corresponding to high-priority sensitive targets are directly designated as key pollution prevention and control areas. These areas require priority control measures to reduce the damage of pollution to core targets. At the same time, combined with the hydrogeological conditions and monitoring point layout in the pollution risk distribution baseline, key nodes of groundwater runoff around high-priority targets and hydrogeological boundaries that pollutants must diffuse through are identified as key control nodes. These nodes are the core locations for blocking pollution diffusion and monitoring and early warning.

[0114] The final comprehensive assessment not only defined the scope and core locations of prevention and control, but also took into account the matching of protection priorities and pollution risks, providing reliable support for targeted treatment and efficient management of groundwater pollution in mines.

[0115] The application prospects of this embodiment are broad, covering multiple fields such as mine groundwater pollution control, ecological restoration, and environmental supervision, and can be effectively adapted to mine scenarios of different scales and pollution types. At the ecological governance level, it can provide pollution risk data support for mine ecological restoration projects, identify key pollution prevention and control areas and critical control nodes, help formulate targeted restoration plans, reduce pollution damage to groundwater ecosystems and surrounding sensitive targets, and promote the gradual restoration of ecological functions in mining areas. At the environmental supervision level, the all-element sensing network and intelligent early warning mechanism can realize real-time monitoring, prediction, and analysis of pollution, helping regulatory departments to predict pollution spread trends in advance, improve the initiative and efficiency of environmental supervision, and solve the problems of passive response and insufficient accuracy in traditional supervision.

[0116] Meanwhile, the technical method of this embodiment can be further extended to similar groundwater pollution scenarios such as mineral resource mining areas and industrial contaminated sites, possessing strong technical versatility and promotional value. From an industrial application perspective, it can provide environmental technology companies with standardized pollution perception and early warning technology solutions, promoting the industrial application and iterative upgrading of related monitoring equipment and intelligent analysis models. From a social value perspective, by controlling groundwater pollution, it can effectively ensure the safety of drinking water and agricultural production for surrounding residents, alleviate environmental conflicts caused by mining pollution, and provide solid technical support for regional ecological environment security and coordinated economic and social development, demonstrating significant practical application significance and long-term promotion prospects.

[0117] Example 2

[0118] Please refer to Figure 3 This embodiment 2 provides a multi-element perception and intelligent early warning system for groundwater pollution in mines, including:

[0119] The full-element sensing network construction unit is used to build a sensing network covering all elements of groundwater pollution in mines, collect pollution source data, hydrogeological data, groundwater dynamic monitoring data and environmental correlation data in mines, and store all data in a unified database after standardization.

[0120] The risk distribution base generation unit is used to overlay vector layers on the spatial coordinates of pollution sources, the location of monitoring wells, and the boundary data of hydrogeological zones in a geographic information system to generate a groundwater pollution risk distribution base in the mining area that includes the spatial relationship between "pollution source-aquifer-monitoring point".

[0121] The weighted quantification unit is used to establish a pollutant migration model that couples the dissolution rate of mine pollutants with the characteristics of groundwater seepage field. It uses the control variable method to correlate the dynamic response relationship between the stock of pollution sources, the permeability coefficient of aquifers and the concentration of characteristic pollutants in groundwater, and quantifies the influence weight of each factor on pollution diffusion based on the analytic hierarchy process.

[0122] The pollution prediction unit is used to extract the time series of pollution concentration from historical monitoring data, combine the sliding window characteristics of real-time data, and use the Long Short-Term Memory (LSTM) time series analysis method to output the predicted diffusion path, concentration peak and time of arrival of groundwater pollution to sensitive targets in the future prediction period.

[0123] The risk assessment unit is used to form a comprehensive assessment conclusion on mine groundwater pollution based on the pollution risk distribution base and diffusion prediction results, combined with the priority classification of sensitive target protection, and to identify key areas and critical control nodes for pollution prevention and control.

[0124] like Figure 4 As shown, the functional outputs of the corresponding all-element perception network construction unit and risk distribution base generation unit, using the geospatial base map as a carrier, visualize the spatial distribution of pollution sources, and simultaneously present core attribute information such as pollution source name, spatial coordinates, and pollution type. This enables systematic management and spatial correlation of pollution source data, provides an intuitive data verification and management entry point for risk distribution base construction, and ensures the accuracy of the data source required for vector layer overlay.

[0125] like Figure 5 As shown, in accordance with the operational coordination requirements of the all-element sensing network construction unit and the pollution prediction unit, the system integrates the location of monitoring stations, real-time groundwater monitoring data, time-series variation curves of characteristic pollutant concentrations, and early warning threshold indicators. On the one hand, it provides real-time feedback of dynamic monitoring data collected by the all-element sensing network, enabling real-time data presentation, anomaly early warning, and source tracing. On the other hand, it provides continuous real-time data input to the pollution prediction unit, supporting the implementation of time-series analysis and prediction work.

[0126] like Figure 6 As shown, as the core output carrier of the system, it integrates the core outputs of five functional units, superimposes multiple types of information such as risk distribution base, pollution diffusion prediction results, sensitive target protection classification, key prevention and control areas and key control nodes, and corresponds to the functional output of the risk assessment unit. It realizes the integrated and visualized presentation of the results of the entire process of "data collection-risk modeling-pollution prediction-comprehensive assessment", and provides management departments with comprehensive and accurate decision support.

[0127] Example 3

[0128] This embodiment 3 also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it can realize any step of a multi-element perception and intelligent early warning method for groundwater pollution in mines.

[0129] The computer-readable storage medium may include various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0130] For a description of the computer-readable storage medium provided in this application, please refer to the above method embodiments; further details will not be repeated here.

[0131] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for multi-element perception and intelligent early warning of groundwater pollution in mines, characterized in that, include: S1. Construct a sensing network covering all elements of groundwater pollution in mines, collect pollution source data, hydrogeological data, groundwater dynamic monitoring data and environmental correlation data from mines, and store all data in a unified database after standardization; S2. The spatial coordinates of pollution sources, the location of monitoring wells, and the boundary data of hydrogeological zones are overlaid in the geographic information system to generate a base map of groundwater pollution risk distribution in the mining area that includes the spatial relationship between "pollution source-aquifer-monitoring point". S3. Establish a pollutant migration model that couples the dissolution rate of mine pollutants with the characteristics of groundwater seepage field. By using the control variable method, correlate the dynamic response relationship between the stock of pollution sources, the permeability coefficient of aquifers and the concentration of characteristic pollutants in groundwater. Based on the analytic hierarchy process, quantify the influence weight of each factor on pollution diffusion. S4. Extract the pollution concentration time series from historical monitoring data, combine it with the sliding window characteristics of real-time data, and use the Long Short-Term Memory (LSTM) time series analysis method to output the diffusion path, concentration peak and time prediction of groundwater pollution reaching sensitive targets in the future prediction period. S5. Based on the pollution risk distribution baseline and diffusion prediction results, and combined with the priority classification of sensitive target protection, a comprehensive judgment conclusion on mine groundwater pollution is formed, and key areas and critical control nodes for pollution prevention and control are identified. The generation process of the water pollution risk distribution base in S2 is as follows: Based on a unified geographic coordinate system, let the first layer of the pollution source spatial coordinate vector map be defined. The spatial coordinates of the pollution sources are ,in , They are respectively Cartesian coordinates. The elevation of the pollution source; the first layer in the hydrogeological zoning boundary vector map layer. The set of vertex coordinates of the boundary polygons of each hydrogeological zone is as follows: , The number of boundary vertices; the first in the vector layer of the monitoring well deployment location. The spatial coordinates of each monitoring well are ,in These represent the abscissa, ordinate, and elevation of the wellhead. Based on the formula for determining the attribution of spatial points and surfaces Determine the coordinates of the pollution source and monitoring well coordinates Does it fall in the first place? Hydrogeological zones Within the boundary polygon, where The coordinates of the point to be determined are in the plane. When the formula holds true, it indicates that the point to be determined belongs to this hydrogeological zone. Using the formula for calculating spatial distance Calculate the spatial straight-line distance between pollution sources belonging to the same hydrogeological zone and monitoring wells. ; Based on the attribution determination results and spatial distance calculation results, a groundwater pollution risk distribution baseline for the mining area is integrated, which includes the spatial correlation between "pollution source - hydrogeological zone - monitoring well". The attribution determination results represent the spatial subordination relationship among these three elements, and the spatial distance... Characterizes the spatial proximity between pollution sources and monitoring wells.

2. The method for multi-element perception and intelligent early warning of groundwater pollution in mines according to claim 1, characterized in that, The construction process of the full-element perception network in S1 is as follows: With the diffusion path and impact range of groundwater pollution in mines as the core, a three-level sensing node system of "pollution source - migration path - sensitive target" is deployed; fixed monitoring nodes are deployed in the mine pollution area, aquifer recharge area and runoff zone, and mobile supplementary monitoring nodes are deployed around sensitive targets to form a sensing network layout with full coverage and key densification. The fixed monitoring node is equipped with a multi-parameter water quality sensor, a water level sensor, and a flow sensor, while the mobile supplementary monitoring node is equipped with a portable rapid water quality analyzer. By combining wired and wireless communication technologies, the system enables real-time data collection and transmission from each monitoring node. It also integrates existing survey data, hydrogeological exploration data, and environmental data including regional meteorology and surface runoff from the mining area, thus completing the construction of a comprehensive sensing network.

3. The method for multi-element perception and intelligent early warning of groundwater pollution in mines according to claim 2, characterized in that, The pollution source data, hydrogeological data, groundwater dynamic monitoring data, and environmental correlation data in S1 are as follows: Pollution source data includes basic attributes, pollutant characteristics, and spatial distribution correlation data of pollution sources related to groundwater pollution in and around mines; Hydrogeological data includes geological structural parameters, hydrodynamic conditions, and background water quality characteristics related to groundwater occurrence and migration in the mining area; Groundwater dynamic monitoring data includes real-time dynamic data on the physical and chemical properties of groundwater, concentrations of characteristic pollutants, and hydrodynamic parameters collected through a sensing network. Environmental data includes meteorological, surface, soil, land use, and sensitive target protection data that affect the generation and migration of groundwater pollution in mining areas.

4. The method for multi-element perception and intelligent early warning of groundwater pollution in mines according to claim 1, characterized in that, The pollutant migration model in S3 is specifically as follows: First, taking the aquifer control volume as the research object, according to the law of conservation of mass, the rate of change of pollutant mass within the control volume is equal to the algebraic sum of the divergence of pollutant dispersion flux, the divergence of convective flux, and source-sink terms. Based on this, a pollutant mass balance equation is established: ,in For total pollutant flux, Porosity is the porosity of an aquifer, which characterizes the proportion of pores in the total volume of the aquifer. The concentration of characteristic pollutants in groundwater; For time; It is a vector differential operator that characterizes the spatial gradient change; Second, total pollutant flux Composed of diffuse flux and convective flux, according to Fick's law and convective transport theory, the expression for diffuse flux is: The expression for circulation volume is: ,in The pollutant dispersion coefficient is determined by the aquifer particle composition, pore structure, and groundwater seepage velocity. The actual seepage velocity of groundwater reflects the speed of groundwater movement within the aquifer; therefore, the total flux... ; Furthermore, considering the leaching and replenishment effects of mine pollutants, a source-sink project is introduced. Characterizing the leaching intensity of pollutants, and combining the above mass balance equation and total flux expression, the core expression of the pollutant migration model is derived: Among them, the source and sink items Based on the kinetics of pollutant leaching, a quantitative derivation is made: the amount of pollutant leached is positively correlated with the leaching rate constant, total mass, and contact area, and negatively correlated with the groundwater volume of the hydrogeological unit. , The leaching rate constant of pollutants is determined by the physicochemical properties of the pollutants, the temperature and humidity of the surrounding environment, and the groundwater immersion conditions. The total mass of mine pollutants; The contact area between the pollutants and groundwater; The volume of groundwater in the hydrogeological unit where the pollutant is located; Actual groundwater seepage velocity Based on Darcy's law: Darcy's law states that seepage velocity is positively correlated with permeability coefficient and hydraulic gradient. By correcting the actual seepage velocity using porosity, we obtain... , The aquifer permeability coefficient characterizes the water permeability of an aquifer. The hydraulic gradient is determined by the ratio of the groundwater level difference to the length of the water flow path within the hydrogeological zone.

5. The method for multi-element perception and intelligent early warning of groundwater pollution in mines according to claim 1, characterized in that, The quantification process of the influence weights of each element on pollution diffusion in S3 is as follows: First, a pairwise comparison judgment matrix is ​​constructed: based on the differences in the degree of influence of each indicator on pollution diffusion, the element values ​​of the judgment matrix are determined by comparing the relative importance of the indicators. These element values ​​are supported by pollution diffusion simulation experimental data; that is, with other indicators fixed, the change in the concentration of a characteristic pollutant caused by a change in a single indicator is observed using the controlled variable method. The ratio of the change in magnitude determines the relative importance level between the indicators, thus forming the judgment matrix. ,in To evaluate the number of indicators, Indicates the first The first indicator is relative to the first The relative importance values ​​of each indicator, and satisfying , ; Secondly, calculate the eigenvectors and the largest eigenvalue of the judgment matrix: by solving the characteristic equation of the judgment matrix. To obtain the largest eigenvalue and the corresponding normalized feature vector The elements in the normalized eigenvector This refers to the initial weight values ​​of each evaluation indicator; Finally, a consistency check is performed: the consistency index is calculated. And combined with the average random consistency index Obtain the consistency ratio ;like If the judgment matrix satisfies the consistency requirement, the corresponding normalized eigenvector is the final impact weight of each element on pollution diffusion. If the consistency requirement is not met, the values ​​of the judgment matrix elements are adjusted based on the pollution diffusion simulation experiment data, and the above steps are repeated until the consistency requirement is met, ultimately realizing the quantification of the influence weight of each element.

6. The method for multi-element perception and intelligent early warning of groundwater pollution in mines according to claim 1, characterized in that, The prediction process for the prediction results in S4 is as follows: First, extract the time series of characteristic pollutant concentrations from historical monitoring data, denoted as... ,in For historical monitoring moments; construct a sliding window from real-time monitoring data at fixed time steps, with the first [time point] within the window being [the next data point]. The input feature vector at each time step is ,in The length of the sliding window. The pollutant concentration at that time step. This represents the groundwater seepage velocity at the corresponding time. The hydraulic gradient at the corresponding time point is obtained from the parameters of the pollutant migration model, thus forming the input sample set. , The total sample size is given; the sample set is then normalized using the following formula: ,in , The minimum and maximum values ​​of the corresponding features in the sample set are respectively used to obtain the standardized sample set. ; Secondly, construct an LSTM model, where the hidden layer state update formula is: ,in This is the weight matrix from the input layer to the hidden layer. This is the weight matrix from hidden layer to hidden layer. , These are the corresponding bias vectors. It is the Sigmoid activation function. The hyperbolic tangent activation function is used. For Hadama accumulation, This represents the hidden layer state from the previous time step. The model is trained by inputting a standardized sample set and outputs predicted pollutant concentrations for consecutive future time periods. , The predicted time step number is used to predict the time span, which covers a preset duration. Finally, based on the baseline of groundwater pollution risk distribution in the mining area, spatial interpolation formulas were used. Calculate any spatial location The concentration of pollutants, among which For the spatial coordinates of the monitoring point, For position and The spatial distance is obtained from the spatial distance formula. The number of monitoring points; through Determine the peak concentration, corresponding to the time point. Based on the spatial coordinates of sensitive targets Solve the equation Obtain the time it takes for pollutants to reach sensitive targets , For sensitive target pollutant concentration thresholds, the final integration Spatial distribution , and This forms the prediction results.

7. The method for multi-element perception and intelligent early warning of groundwater pollution in mines according to claim 1, characterized in that, The priority classification of sensitive target protection in S5 is based on the functional attributes, spatial correlation characteristics and risk exposure degree of sensitive targets, combined with the pollution risk distribution base and diffusion prediction results, to construct a multi-dimensional classification system; Based on the functional importance of sensitive targets themselves, and in conjunction with their spatial location relationship with pollution sources and the degree of pollution threat they are subject to, a priority gradient sequence is formed. Based on this, high, medium and low protection levels are divided, providing a priority judgment benchmark for the delineation of key pollution prevention and control areas and critical control nodes.

8. A multi-element sensing and intelligent early warning system for groundwater pollution in mines, characterized in that, include: The full-element sensing network construction unit is used to build a sensing network covering all elements of groundwater pollution in mines, collect pollution source data, hydrogeological data, groundwater dynamic monitoring data and environmental correlation data in mines, and store all data in a unified database after standardization. The risk distribution base generation unit is used to overlay vector layers on the spatial coordinates of pollution sources, the location of monitoring wells, and the boundary data of hydrogeological zones in a geographic information system to generate a groundwater pollution risk distribution base in the mining area that includes the spatial relationship between "pollution source-aquifer-monitoring point". The weighted quantification unit is used to establish a pollutant migration model that couples the dissolution rate of mine pollutants with the characteristics of groundwater seepage field. It uses the control variable method to correlate the dynamic response relationship between the stock of pollution sources, the permeability coefficient of aquifers and the concentration of characteristic pollutants in groundwater, and quantifies the influence weight of each factor on pollution diffusion based on the analytic hierarchy process. The pollution prediction unit is used to extract the time series of pollution concentration from historical monitoring data, combine the sliding window characteristics of real-time data, and use the Long Short-Term Memory (LSTM) time series analysis method to output the predicted diffusion path, concentration peak and time of arrival of groundwater pollution to sensitive targets in the future prediction period. The risk assessment unit is used to form a comprehensive assessment conclusion on mine groundwater pollution based on the pollution risk distribution base and diffusion prediction results, combined with the priority classification of sensitive target protection, and to identify key areas and critical control nodes for pollution prevention and control. The process of generating the water pollution risk distribution base in the risk distribution base generation unit is as follows: Based on a unified geographic coordinate system, let the first layer of the pollution source spatial coordinate vector map be defined. The spatial coordinates of the pollution sources are ,in , They are respectively Cartesian coordinates. The elevation of the pollution source; the first layer in the hydrogeological zoning boundary vector map layer. The set of vertex coordinates of the boundary polygons of each hydrogeological zone is as follows: , The number of boundary vertices; the first in the vector layer of the monitoring well deployment location. The spatial coordinates of each monitoring well are ,in These represent the abscissa, ordinate, and elevation of the wellhead. Based on the formula for determining the attribution of spatial points and surfaces Determine the coordinates of the pollution source and monitoring well coordinates Does it fall in the first place? Hydrogeological zones Within the boundary polygon, where The coordinates of the point to be determined are in the plane. When the formula holds true, it indicates that the point to be determined belongs to this hydrogeological zone. Using the formula for calculating spatial distance Calculate the spatial straight-line distance between pollution sources belonging to the same hydrogeological zone and monitoring wells. ; Based on the attribution determination results and spatial distance calculation results, a groundwater pollution risk distribution baseline for the mining area is integrated, which includes the spatial correlation between "pollution source - hydrogeological zone - monitoring well". The attribution determination results represent the spatial subordination relationship among these three elements, and the spatial distance... Characterizes the spatial proximity between pollution sources and monitoring wells.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, The computer program is executed by a processor as described in any one of claims 1-7: a method for multi-element perception and intelligent early warning of groundwater pollution in mines.

Images