Mine pollutant migration multi-process coupling prediction method and device based on deep learning

Abstract

The invention relates to a mine pollutant migration multi-process coupling prediction method and device based on deep learning. The method comprises the following steps: acquiring a first feature formed by real-time data of each node of a target mine; calculating a first attention coefficient based on the first feature of the node and the edge feature of the connected node; updating the first feature by using a first attention weight standardized by a first attention coefficient to obtain a second feature of the node; calculating a second attention coefficient of the connected node based on the environment variable of the target mine and the second feature, and updating the second feature based on a second attention weight standardized by the second attention coefficient to obtain a third feature of the node; and calculating a third attention coefficient of the connected node based on the oxidation-reduction potential and the third feature of the node, updating the third feature by using a third attention weight standardized by the third attention coefficient to obtain a fourth feature of the node, and mapping the fourth feature into a predicted pollutant concentration. The method can improve the prediction precision, and can comprehensively carry out risk assessment.

Description

Technical Field

[0001] The present application relates to the fields of environmental science and artificial intelligence technology, and in particular to a method and device for predicting multi-process coupling of mine pollutant migration based on deep learning. Background Art

[0002] When production activities are carried out in the environment, they may generate pollutants. These pollutants can migrate from one location to another, causing their migration and diffusion. For example, during mining, the migration of pollutants such as ammonia nitrogen and heavy metals in fissure water and groundwater is complexly influenced by geological structure, hydrological conditions, and meteorological factors. Therefore, it is necessary to analyze the migration and diffusion of pollutants to predict their concentration distribution at future times, so that timely control measures can be implemented based on these concentration distributions to prevent irreversible damage to the environment.

[0003] Currently, models for predicting pollutant concentrations are mostly based on single processes or static parameters, making it difficult to capture the nonlinear dynamic characteristics of multi-process coupling, resulting in low prediction accuracy. For example, traditional pollutant migration models typically rely on static geological parameters and assumed initial conditions to predict concentrations, while artificial intelligence algorithm models often use single-point sensor data to predict local concentrations, failing to generate a complete spatial concentration field distribution. Local concentrations, however, make it difficult to comprehensively assess pollution risks. Summary of the Invention

[0004] Based on this, it is necessary to provide a deep learning-based multi-process coupling prediction method and device for mine pollutant migration that can improve prediction accuracy and conduct comprehensive risk assessment in response to the above technical problems.

[0005] A multi-process coupled prediction method for mine pollutant migration based on deep learning, the method comprising:

[0006] S1. Obtaining a first feature consisting of real-time data of each node in a target mine; the real-time data includes pollutant data;

[0007] Preferably, the target mine refers to an area where pollution exists.

[0008] Preferably, the pollutant data includes but is not limited to pollutant concentration and pollutant type.

[0009] Preferably, the real-time data of each node can be collected by a real-time data collection device.

[0010] S2. Calculating a first attention coefficient of the connected node based on the first feature of the node and the edge feature between the connected nodes;

[0011] Preferably, the first attention coefficient is a coefficient used to represent the importance or weight of the connection between connected nodes in the graph attention network.

[0012] Preferably, the edge feature is used to describe the interaction relationship between two connected nodes.

[0013] S3. Update the first feature using the first attention weight normalized by the first attention coefficient to obtain a second feature of the node;

[0014] S4. Calculating a second attention coefficient of the connected node based on the environmental variables of the target mine and the second feature, and updating the second feature based on a second attention weight normalized by the second attention coefficient to obtain a third feature of the node; the environmental variables include water flow rate and pollutant diffusion coefficient;

[0015] Preferably, the second attention coefficient is a parameter used to dynamically adjust the weight of the pollutant transmission path.

[0016] Preferably, the pollutant diffusion coefficient is a parameter used to describe the ability of pollutants to diffuse between two connected nodes.

[0017] S5. Based on the redox potential of the node and the third feature, calculate the third attention coefficient of the connected node, and use the third attention weight normalized by the third attention coefficient to update the third feature to obtain the fourth feature of the node, and map the fourth feature to the predicted pollutant concentration.

[0018] Preferably, the third attention coefficient is used to model the redox reaction during the heavy metal chemical form transformation process of each node.

[0019] Preferably, the redox potential is used to reflect the ability of pollutants at a node to undergo redox reactions.

[0020] In one embodiment, the method of obtaining the first feature in step S1 includes:

[0021] Obtaining geological parameters, pollutant parameters, and meteorological parameters of each node in the target mine, and converting the geological parameters, pollutant parameters, and meteorological parameters into vector-type geological vectors, pollutant vectors, and meteorological vectors;

[0022] splicing the geological vector, the pollutant vector, and the meteorological vector to obtain a first feature of each node;

[0023] The expression for concatenating the first features of the nodes is: ;

[0024] in, represents the first feature of node i; represents the geological vector of node i, p i represents the pollutant vector of node i, m i represents the meteorological vector of node i, || represents a vector concatenation operation; , d input represents the total dimension, represents the dimension of the geological vector, represents the dimension of the pollutant vector, Indicates the dimension of the weather vector.

[0025] Preferably, geological parameters include the internal structure and surface features of the target mine. Pollutant parameters include parameters used to describe the characteristics of the pollutants themselves. Meteorological parameters include various physical quantities that describe the atmospheric conditions of the target mine. In one embodiment, the calculation formula of the first attention coefficient in step S2 is: ;

[0026] in, represents the first attention coefficient between the connected nodes i and j, is a nonlinear activation function; for The transposed matrix of the first learnable parameter vector of dimension ; is the first learnable weight matrix, F is the input dimension, is the output dimension; is the first edge feature transformation matrix; is the partition coefficient calculation function, Represents the edge features between connected nodes i and j; represents the first feature of node i, Represents the first feature of node j.

[0027] In one embodiment, the calculation formula of the first attention weight in step S3 is: ;

[0028] The calculation formula of the second feature of the node in step S3 is: ;

[0029] in, is the first attention weight between the connected nodes i and j; softmax is the activation function; represents the first attention coefficient between the connected nodes i and j, Represents the first attention coefficient between the connected nodes i and k; is the set of nodes connected to node i; is the activation function, represents the first feature of node j, is the first learnable weight matrix; is the second feature of node i.

[0030] In one embodiment, the calculation formula of the second attention coefficient in step S4 is: ;

[0031] in, Represents the second attention coefficient between the connected nodes i and j; LeakyReLU is a nonlinear activation function; for The transposed matrix of the second learnable parameter vector of dimension ; is the second learnable weight matrix, is the input dimension, is the output dimension; is the second edge feature transformation matrix; is the water flow velocity vector from node i to node j; is the diffusion coefficient from node i to node j; represents the second feature of node i, Represents the second feature of node j.

[0032] In one embodiment, the environmental variables further include rainfall and mining activity intensity of the target mine. When the rainfall reaches a first threshold and / or the mining activity intensity reaches a second threshold, the calculation formula of the second attention coefficient in step S4 is: ;

[0033] in, Represents the second attention coefficient between the connected nodes i and j; LeakyReLU is a nonlinear activation function; for The transposed matrix of the second learnable parameter vector of dimension ; is the second learnable weight matrix, F is the input dimension, is the output dimension; is the second edge feature transformation matrix; W R and W M is a learnable parameter; R t is the rainfall at time t, M t is the mining activity intensity at time t; is the water flow velocity vector from node i to node j; is the diffusion coefficient from node i to node j.

[0034] Preferably, the mining activity intensity represents the mining frequency of the target mine.

[0035] In one embodiment, the calculation formula of the second attention weight in step S4 is: ;

[0036] The calculation formula of the third feature of the node in step S4 is: ;

[0037] in, is the second attention weight between the connected nodes i and j, softmax is the activation function, represents the second attention coefficient between the connected nodes i and j, represents the second attention coefficient between the connected nodes i and k, is the set of nodes connected to node i; is the activation function, represents the second feature of node j, is the third feature of node i.

[0038] In one embodiment, the calculation formula of the third attention coefficient in step S5 is: ;

[0039] The calculation formula is: ;

[0040] in, Represents the third attention coefficient between the connected nodes i and j; LeakyReLU is a nonlinear activation function; for The transposed matrix of the third learnable parameter vector of dimension ; is the third learnable weight matrix, F is the input dimension, is the output dimension; is the diffusion coefficient from node i to node j; || represents the vector concatenation operation, represents the redox potential of node i With the third feature The result of splicing, represents the redox potential of node j With the third feature The result of splicing.

[0041] In one embodiment, the calculation formula of the third attention weight in step S5 is: ;

[0042] The calculation formula of the fourth feature in step S5 is: ;

[0043] The mapping formula for mapping the fourth feature to the predicted pollutant concentration in step S5 is: ;

[0044] in, is the third attention weight between the connected nodes i and j, softmax is the activation function, represents the third attention coefficient between the connected nodes i and j, represents the third attention coefficient between the connected nodes i and k, is the set of nodes connected to node i; is the activation function, represents the redox potential of node i With the third feature The result of splicing, is the fourth feature of node i, c i is the predicted pollutant concentration at node i, w out is the output weight matrix, b out is the bias vector.

[0045] A multi-process coupled prediction device for mine pollutant migration based on deep learning, the device comprising:

[0046] A first feature acquisition module is used to acquire a first feature composed of real-time data of each node in the target mine; the real-time data includes pollutant data;

[0047] A coefficient calculation module, configured to calculate a first attention coefficient of the connected node based on the first feature of the node and the edge feature between the connected nodes;

[0048] A second feature acquisition module is configured to update the first feature using the first attention weight normalized by the first attention coefficient to obtain a second feature of the node;

[0049] a third feature acquisition module, configured to calculate a second attention coefficient of the connected node based on the environmental variables of the target mine and the second feature, and update the second feature based on a second attention weight normalized by the second attention coefficient to obtain a third feature of the node; the environmental variables include water flow rate and pollutant diffusion coefficient;

[0050] A concentration prediction module is used to calculate the third attention coefficient of the connected node based on the redox potential of the node and the third feature, and update the third feature using the third attention weight normalized by the third attention coefficient to obtain the fourth feature of the node, and map the fourth feature to the predicted pollutant concentration.

[0051] The above-mentioned method and device for multi-process coupling prediction of mine pollutant migration based on deep learning, through S1, obtains the first feature consisting of real-time data of each node in the target mine; the real-time data includes pollutant data, S2, based on the first feature of the node and the edge feature between the connected nodes, calculates the first attention coefficient of the connected node, S3, uses the first attention weight standardized by the first attention coefficient to update the first feature, and obtains the second feature of the node, S4, based on the environmental variables and the second feature of the target mine, calculates the second attention coefficient of the connected node, and updates the second feature based on the second attention weight standardized by the second attention coefficient, and obtains the third feature of the node; the environmental variables include water Flow velocity and pollutant diffusion coefficient, S5, based on the redox potential and the third feature of the node, calculate the third attention coefficient of the connected node, and use the third attention weight standardized by the third attention coefficient to update the third feature to obtain the fourth feature of the node, and map the fourth feature to the predicted pollutant concentration. On the one hand, the predicted pollutant concentration of each node in the target mine can be obtained, so as to conduct a comprehensive risk assessment of the target mine based on the predicted pollutant concentration of each node. On the other hand, since the real-time data of the node is used, the dynamic update of the edge feature can be realized, so that the entire prediction process can adapt to the dynamic changes of the environment in the target mine, reduce the deviation caused by static assumptions, and improve the prediction accuracy. In addition, by fusing the first feature, edge feature, environmental variable, second feature, redox potential and the third feature of each node layer by layer, multi-process coupling modeling and migration prediction under the synergistic effect of multiple mechanisms can be realized. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 This is a diagram illustrating an application environment of a multi-process coupled prediction method for mine pollutant migration based on deep learning in one embodiment;

[0053] Figure 2 Schematic diagram of a process for predicting multi-process coupling of mine pollutant migration based on deep learning in one embodiment;

[0054] Figure 3 This is a structural block diagram of a multi-process coupled prediction device for mine pollutant migration based on deep learning in one embodiment;

[0055] Figure 4 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0057] The multi-process coupling prediction method for mine pollutant migration based on deep learning provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. The terminal 102 interacts with the server 104 via a wired / wireless channel. The data storage system can store data that the server 104 needs to process. S1. Obtain a first feature consisting of real-time data of each node in the target mine; the real-time data includes pollutant data; S2. Calculate the first attention coefficient of the connected node based on the first feature of the node and the edge feature between the connected nodes; S3. Update the first feature using the first attention weight normalized by the first attention coefficient to obtain the second feature of the node; S4. Calculate the second attention coefficient of the connected node based on the environmental variables and the second feature of the target mine, and update the second feature based on the second attention weight normalized by the second attention coefficient to obtain the third feature of the node; the environmental variables include water flow rate and pollutant diffusion coefficient; S5. Calculate the third attention coefficient of the connected node based on the redox potential of the node and the third feature, and update the third feature using the third attention weight normalized by the third attention coefficient to obtain the fourth feature of the node, and map the fourth feature to the predicted pollutant concentration. The terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, etc. The server 104 may be a single server, a server cluster composed of multiple servers, or a cloud computing center composed of multiple servers.

[0058] In one embodiment, Figure 2 As shown in the figure, a multi-process coupling prediction method for mine pollutant migration based on deep learning is provided. Figure 1 Taking the server 104 in the example as an example, the following steps are included:

[0059] S1. Acquire a first feature of real-time data of each node in the target mine; the real-time data includes pollutant data.

[0060] Target mines are those with contamination. Specifically, during the ore extraction process, chemicals are used to extract the minerals. If these chemicals are not properly managed during the extraction process, they can leak into the surrounding environment, causing soil and water pollution. Mines include, but are not limited to, metal mines and rare earth mines. Metal mines include, but are not limited to, iron and gold mines.

[0061] The distribution of nodes throughout the target mine can be uniform or non-uniform. Nodes refer to locations in the target mine where real-time data acquisition devices have been deployed, as well as locations in the target mine where real-time data acquisition devices are relatively important but not deployed. Specifically, in the target mine, nodes include locations where real-time data acquisition devices have been deployed, as well as locations where real-time data acquisition devices have not been deployed, such as deep fissures, remote slopes, and locations where monitoring costs are too high. Furthermore, real-time data acquisition devices include, but are not limited to, sensors and data recorders. Furthermore, real-time data from nodes is collected by the deployed real-time data acquisition devices, and real-time data from nodes where real-time data acquisition devices are not deployed is determined based on prior geological knowledge of the target mine, which includes, but is not limited to, historical fracture distribution and permeability gradient. Historical fracture distribution refers to the formation, development, and spatial distribution of fractures or faults in rocks in a certain area during geological history. Permeability gradient refers to the trend or rate of change of permeability with location, that is, how the permeability changes with distance in a specific direction. Permeability is a parameter that describes the ability of fluid to pass through a porous medium.

[0062] The first feature is a feature vector obtained from the real-time data of the node. Furthermore, when the real-time data is of one type, the first feature is the feature vector of the real-time data; when the real-time data is of multiple types, the first feature is a feature vector formed by combining the various real-time data.

[0063] Pollutant data, also known as pollutant parameters, includes but is not limited to pollutant concentrations and types at the node's location. In some embodiments, pollutants can be transferred from one node to another through water flow, soil, and other processes. Therefore, the node's real-time data is constantly changing, and so are the node's primary characteristics.

[0064] In some embodiments, the process of obtaining a first feature formed by real-time data from each node in a target mine includes: obtaining initial data from each node in the target mine in real time, performing data preprocessing on the initial data to obtain real-time data, and obtaining the first feature of the real-time data. The data preprocessing includes removing outliers from the initial data and standardizing the initial data. Data preprocessing can ensure input consistency.

[0065] To predict pollutant concentrations at each node, real-time data from each node in the target mine is fed into a trained graph attention network. This network then predicts pollutant concentrations through its various layers. When inputting real-time data into the graph attention network, it is accompanied by node information, which clarifies the node from which the input real-time data originates. Graph attention networks are a type of deep learning. Specifically, they are a subclass of graph neural networks, specifically designed for processing graph-structured data.

[0066] S2. Calculate the first attention coefficient of the connected nodes based on the first feature of the node and the edge features between the connected nodes.

[0067] Among them, connected nodes refer to nodes with physical or chemical connections. Physical connections include but are not limited to groundwater flow direction and fracture connectivity. Groundwater flow direction refers to the direction of groundwater flow in underground porous media, and fracture connectivity refers to the degree of interconnection between fractures in the rock. Fracture connectivity reflects the openness and continuity of the fracture network. Chemical connections include but are not limited to pollutant migration paths. Pollutant migration paths refer to the routes that pollutants move from the source to other locations. For example, in the target mine, groundwater from node A can flow to node B, then nodes A and node B are connected nodes; there is a main fracture near node C, which is connected to the intersection of the main fracture zone (node ​​D) and a deep hidden fracture (node ​​E), forming a fracture network channel, then nodes C, node D, and node E are connected nodes; a certain pollutant leaked in the node F area has a diffusion path that is often detected by nodes G and node H, and there is a correlation between the concentration changes, then nodes F, node G, and node H are connected nodes.

[0068] Edge features describe the interaction between two connected nodes. For example, if nodes A and B are connected, the edge feature between them could be the adsorption intensity of a pollutant when it migrates from node A to node B. Edge features are calculated in real time based on the first feature of the connected nodes. Since the first feature of the connected nodes changes in real time, the edge feature will also change continuously.

[0069] The first attention coefficient is a coefficient used to represent the importance or weight of the connection between connected nodes in the graph attention network. The calculation formula of the first attention coefficient can be , Represents the first attention coefficient between the connected nodes i and j, and LeakyReLU is a nonlinear activation function; for The transposed matrix of the first learnable parameter vector of dimension ; is the partition coefficient calculation function, Represents the edge features between connected nodes i and j; represents the first feature of node i, Represents the first feature of node j.

[0070] S3. Update the first feature using the first attention weight normalized by the first attention coefficient to obtain the second feature of the node.

[0071] The first attention weight is obtained by normalizing the first attention coefficient. The normalization methods include but are not limited to normalization using a softmax function, normalization, threshold-based truncation, sparse processing, and local normalization.

[0072] The first attention weight, the second attention weight, and the third attention weight are the attention weights between two connected nodes, which are mainly used to measure the connection strength or importance between two connected nodes.

[0073] Methods for updating the first feature of a node include, but are not limited to, updating through weighted summation, updating through a high-order aggregation strategy, updating through a skip connection, and updating through a self-attention mechanism. The formula for updating the first feature through a high-order aggregation strategy is: ;

[0074] The calculation formula for updating the first feature through skip connection is: ;

[0075] The calculation formula for updating the first feature through the self-attention mechanism is: ;

[0076] in, is the second feature of node i, h i is the first feature of node i, W is the weight matrix, is the first attention weight between the connected nodes i and j, is the gate vector between the connected nodes i and j, N i is the set of nodes connected to node i, h j is the first feature of node j, is the self-attention weight of node i, is the activation function.

[0077] S4. Based on the environmental variables and the second feature of the target mine, calculate the second attention coefficient of the connected nodes, and update the second feature based on the second attention weight after normalization of the second attention coefficient to obtain the third feature of the node.

[0078] Environmental variables include water flow rate and pollutant diffusion coefficient. Water flow rate is the speed of water flowing between two connected nodes in the target mine. Pollutant diffusion coefficient is a parameter used to describe the ability of pollutants to diffuse between two connected nodes. Environmental variables can be acquired in advance or in real time.

[0079] The second attention coefficient is a parameter used to dynamically adjust the weight of the pollutant transmission path. The calculation formula of the second attention coefficient can be ,in, Represents the second attention coefficient between the connected nodes i and j; LeakyReLU is a nonlinear activation function; for The transposed matrix of the second learnable parameter vector of dimension ; is the water flow velocity vector from node i to node j; is the diffusion coefficient from node i to node j; represents the second feature of node i, Represents the second feature of node j.

[0080] The second attention weight is obtained by normalizing the second attention coefficient. Normalization methods include, but are not limited to, using a softmax function, normalization, threshold-based truncation, sparse processing, and local normalization. The second attention coefficient can be obtained by concatenating the second feature and the features of the environment variable.

[0081] The updating methods of the second feature of the node include but are not limited to updating through weighted summation, updating through high-order aggregation strategy, updating through skip connection, and updating through self-attention mechanism.

[0082] S5. Based on the redox potential and the third feature of the node, calculate the third attention coefficient of the connected node, and use the third attention weight normalized by the third attention coefficient to update the third feature to obtain the fourth feature of the node, and map the fourth feature to the predicted pollutant concentration.

[0083] The redox potential reflects the ability of pollutants at a node to undergo redox reactions. In a graph attention network, the redox potential is used to control whether the pollutant's form and toxicity change at the node. This allows for simulation and prediction of the chemical forms of heavy metals and their evolution under different environmental conditions. The redox potential of a node can be measured using sensors.

[0084] The third attention coefficient is used to model the redox reaction during the transformation of heavy metal chemical forms at each node. Specifically, by introducing the redox potential as part of the node feature, the interaction weights between connected nodes are dynamically adjusted, that is, the third attention coefficient is dynamically adjusted. The third attention coefficient can be obtained by splicing the redox potential and the third feature. The calculation formula of the third attention coefficient can be ,in, Represents the third attention coefficient between the connected nodes i and j; LeakyReLU is a nonlinear activation function; for The transposed matrix of the third learnable parameter vector of dimension ; is the pollutant diffusion coefficient from node i to node j; || represents the vector concatenation operation, represents the redox potential of node i With the third feature The result of splicing, represents the redox potential of node j With the third feature The result of splicing.

[0085] The third attention weight is obtained by normalizing the third attention coefficient. The normalization methods include but are not limited to normalization using a softmax function, normalization, threshold-based truncation, sparse processing, and local normalization.

[0086] Methods for updating the third feature of a node include but are not limited to updating through weighted summation, updating through a high-order aggregation strategy, updating through a skip connection, and updating through a self-attention mechanism.

[0087] Methods for mapping the fourth feature to the predicted pollutant concentration include, but are not limited to, implementation via a linear regression model or a multilayer perceptron. Specifically, the process of implementing the fourth feature mapping via a linear regression model involves fitting the training features and the pollutant concentration to obtain a fitted relationship between the training features and the pollutant concentration, and then obtaining the predicted pollutant concentration based on the fitted relationship and the actual fourth feature. The process of implementing the fourth feature mapping via a multilayer perceptron involves using the fourth feature as data for the input layer of the multilayer perceptron and outputting the predicted pollutant concentration value through learning and transformation in the hidden layers of the multilayer perceptron.

[0088] The predicted pollutant concentration obtained through mapping is the pollutant concentration of each node in the target mine at a future moment. By predicting the pollutant concentration of each node at a future moment, the migration of pollutants in the target mine can be analyzed, so that remediation actions can be taken on the target mine based on the predicted pollutant concentration. Specifically, the pollutant concentration alarm threshold is determined, and when the predicted pollutant concentration at any node reaches the alarm threshold, an early warning signal is triggered. Among them, the alarm threshold includes a low-risk alarm threshold, a medium-risk alarm threshold, and a high-risk alarm threshold. Specifically, the nodes where the predicted pollutant concentration reaches the high-risk alarm threshold are automatically pushed to the remediation system to remind management personnel to remediate the nodes where the predicted pollutant concentration reaches the high-risk alarm threshold.

[0089] In some embodiments, the pollutant concentration of each node in the target mine that has been treated is collected to obtain the collected pollutant concentration; the pollutant concentration collected at each node is differentially analyzed with the predicted pollutant concentration. If the difference between the collected pollutant concentration and the predicted pollutant concentration is greater than a preset difference, it is determined that the treatment measures for the target mine are better.

[0090] In some embodiments, the predicted pollutant concentration of each node is pushed to a mobile terminal to enable real-time monitoring and remote decision-making through the mobile terminal.

[0091] Since the nodes are distributed throughout the target mine, when the predicted pollutant concentration of each node is obtained, the concentration field distribution of the target mine can be determined based on the predicted pollutant concentration of each node. Specifically, according to the spatial coordinates of each node and predict pollutant concentrations, generating a continuous three-dimensional pollutant concentration field through interpolation algorithms , thus realizing the multi-process coupled prediction of pollutant migration in target mines.

[0092] The above-mentioned multi-process coupling prediction method for mine pollutant migration based on deep learning, through S1, obtains the first feature consisting of real-time data of each node in the target mine; the real-time data includes pollutant data, S2, based on the first feature of the node and the edge feature between the connected nodes, calculates the first attention coefficient of the connected node, S3, uses the first attention weight standardized by the first attention coefficient to update the first feature, and obtains the second feature of the node, S4, based on the environmental variables and the second feature of the target mine, calculates the second attention coefficient of the connected node, and updates the second feature based on the second attention weight standardized by the second attention coefficient, and obtains the third feature of the node; the environmental variables include water flow rate and the pollutant diffusion coefficient, S5, based on the redox potential and the third feature of the node, calculate the third attention coefficient of the connected nodes, and use the third attention weight standardized by the third attention coefficient to update the third feature to obtain the fourth feature of the node, and map the fourth feature to the predicted pollutant concentration. On the one hand, the predicted pollutant concentration of each node in the target mine can be obtained, so as to conduct a comprehensive risk assessment of the target mine based on the predicted pollutant concentration of each node. On the other hand, since the real-time data of the node is used, the dynamic update of the edge feature can be realized, so that the entire prediction process can adapt to the dynamic changes of the environment in the target mine, reduce the deviation caused by static assumptions, and improve the prediction accuracy. In addition, by fusing the first feature, edge feature, environmental variable, second feature, redox potential and third feature of each node layer by layer, multi-process coupling modeling and migration prediction under the synergistic effect of multiple mechanisms can be realized.

[0093] In one embodiment, the method of obtaining the first feature in step S1 includes:

[0094] The geological parameters, pollutant parameters and meteorological parameters of each node in the target mine are obtained, and the geological parameters, pollutant parameters and meteorological parameters are converted into geological vectors, pollutant vectors and meteorological vectors in vector form.

[0095] The geological vector, pollutant vector and meteorological vector are spliced ​​together to obtain the first feature of each node.

[0096] The expression for concatenating the first features of each node is: ;

[0097] in, represents the first feature of node i; represents the geological vector of node i, p i represents the pollutant vector of node i, m i represents the meteorological vector of node i, || represents the vector concatenation operation; , d input represents the total dimension, represents the dimension of the geological vector, represents the dimension of the pollutant vector, Indicates the dimension of the weather vector.

[0098] Geological parameters, pollutant parameters, and meteorological parameters are real-time data for nodes. Geological parameters include the internal structure and surface characteristics of the target mine. For example, geological parameters include the type and properties of the rocks in the target mine, the proportion of pore space in the rock or soil to the total volume, the ability to affect fluid passage, the ease of fluid movement in porous media, the distribution of cracks in the rock, and the degree of connectivity between them. Pollutant parameters include parameters used to describe the characteristics of the pollutants themselves. For example, the concentration of pollutants in the target mine and the type of pollutants. Meteorological parameters include various physical quantities that describe the atmospheric conditions of the target mine. For example, the temperature, humidity, wind speed and direction, precipitation, and sunshine hours of the target mine.

[0099] The geological parameters are converted into geological vectors, the pollutant parameters are converted into pollutant vectors, and the meteorological parameters are converted into meteorological vectors. The method of converting geological parameters, pollutant parameters and meteorological parameters into geological vectors, pollutant vectors and meteorological vectors in vector form is mainly determined by the type of geological parameters, pollutant parameters and meteorological parameters. Specifically, if the geological parameters, pollutant parameters and meteorological parameters are numerical data, the numerical data can be directly used as vectors; if the geological parameters, pollutant parameters and meteorological parameters are type data, the geological parameters, pollutant parameters and meteorological parameters can be converted into geological vectors, pollutant vectors and meteorological vectors through one-hot encoding. One-hot encoding is a method of converting type data into a format that can be provided to a machine learning algorithm for processing. One-hot encoding achieves conversion by representing each classification value as a binary vector. In the converted vector, only one element is 1, and the rest are 0.

[0100] Furthermore, the numerical geological parameters, pollutant parameters, and meteorological parameters are standardized, and the standardized data are determined as geological vectors, pollutant vectors, and meteorological vectors. Furthermore, the types of geological parameters include, but are not limited to, rock layer types and surface cover types; the types of pollutant parameters include, but are not limited to, pollutant types; and the types of meteorological parameters include, but are not limited to, sunny and rainy days.

[0101] Furthermore, the process of converting geological parameters, pollutant parameters and meteorological parameters into geological vectors, pollutant vectors and meteorological vectors through one-hot encoding is as follows: presetting the one-hot encoding of each geological type, each pollutant type and each meteorological type; and according to the target geological type of the acquired geological parameters, the target pollutant type of the pollutant parameters and the target meteorological type of the meteorological parameters, querying the matching geological type that matches the target geological type from each geological type, querying the matching pollutant type that matches the target pollutant type from each pollutant type, and querying the matching meteorological type that matches the target meteorological type from each meteorological type; using the one-hot encoding of the matching geological type as the geological vector, the one-hot encoding of the matching pollutant type as the pollutant type vector, and the one-hot encoding of the matching meteorological type as the meteorological type vector.

[0102] Total dimension d input It is the sum of the dimensions of the geological vector, pollutant vector and meteorological vector after the geological parameters, pollutant parameters and meteorological parameters are converted into the graph attention network. It can also be called the sum of the vector lengths. For example, the geological vector has 5 values, the pollutant vector has 3 values ​​and the meteorological vector has 2 values, then the total dimension d is input It is 5+3+2=10.

[0103] Splicing is the process of connecting two or more vectors together to form a new vector. Splicing does not change the contents of the geological vector, pollutant vector, and meteorological vector. It combines the geological vector, pollutant vector, and meteorological vector to facilitate subsequent data processing. Specifically, the geological vector is [0.6,75], d 地质 =2, the pollutant vector is [1,0], d 污染物 =2, the meteorological vector is [20,80,10], d 气象 =3, then the first feature after splicing is [0.6, 75, 1, 0, 20, 80, 10], d input =d 地质 +d 污染物 +d 气象 =7.

[0104] In this embodiment, by converting geological parameters, pollutant parameters, and meteorological parameters into vector form, different types of geological parameters, pollutant parameters, and meteorological parameters can be converted into a unified vector form, ensuring that the geological parameters, pollutant parameters, and meteorological parameters can be processed and analyzed. By splicing the geological vectors, pollutant vectors, and meteorological vectors, a comprehensive first feature can be created, providing more comprehensive information for the graph attention network.

[0105] In one embodiment, the calculation formula of the first attention coefficient in step S2 is: ;

[0106] in, Represents the first attention coefficient between the connected nodes i and j, and LeakyReLU is a nonlinear activation function; for The transposed matrix of the first learnable parameter vector of dimension ; is the first learnable weight matrix, F is the input dimension, is the output dimension; is the first edge feature transformation matrix; is the partition coefficient calculation function, Represents the edge features between connected nodes i and j; represents the first feature of node i, Represents the first feature of node j.

[0107] Edge characteristics describe the properties of the interaction between two connected nodes. For example, edge characteristics describe the strength of adsorption behavior when a pollutant migrates from node A to node B. A high edge characteristic indicates that the pollutant is easily adsorbed in the soil and is unlikely to migrate with groundwater or desorb into water, resulting in a low migration risk. A low edge characteristic indicates that the pollutant is not easily adsorbed in the soil and is likely to migrate with groundwater or desorb into water, resulting in a high migration risk.

[0108] Specifically, after determining the first learnable weight matrix , the first feature of node i , the first feature of node j , the first side feature transformation matrix , edge features between connected nodes i and j After obtaining the specific value, the specific value is directly substituted into the calculation formula of the first attention coefficient for calculation, so as to obtain the first attention coefficient.

[0109] In this embodiment, by considering the edge features between two connected nodes and the first features of the two connected nodes when calculating the first attention coefficient, this conforms to the complex interaction mechanism between nodes in the actual environment, can more comprehensively characterize the relationship between nodes, and obtain a more accurate first attention coefficient.

[0110] In one embodiment, the calculation formula of the first attention weight in step S3 is: ;

[0111] The calculation formula of the second feature of the node in step S3 is: ;

[0112] in, is the first attention weight between the connected nodes i and j; softmax is the activation function; represents the first attention coefficient between the connected nodes i and j, Represents the first attention coefficient between the connected nodes i and k; is the set of nodes connected to node i; is the activation function, represents the first feature of node j, is the first learnable weight matrix; is the second feature of node i.

[0113] The Softmax function is used to normalize the first attention coefficient to obtain the first attention weight. The second feature is obtained by weighted summing the first features of connected nodes. The second feature is an enhanced expression based on the aggregation of information from adjacent nodes in the graph structure, reflecting the local context of the node in the graph structure. For example, the second feature of node A is obtained by weighted summing the first features of the nodes connected to node A.

[0114] Pollutant migration is a process of "distribution-transmission-reaction" coupled in series. When calculating the first attention coefficient, the edge features between two connected nodes and the first features of the two connected nodes are considered. The first feature is updated using the first attention weight normalized by the first attention coefficient. The resulting second feature of the node is the distribution process during this series coupling process. A hierarchical design enables graph attention networks to learn physically interpretable decision logic. By physically decoupling the attention mechanism through staged feature transformation, this design significantly improves accuracy and robustness.

[0115] Specifically, after determining the first attention coefficient between connected nodes After obtaining the specific value, the specific value is directly substituted into the calculation formula of the first attention weight for calculation, so as to obtain the first attention weight.

[0116] Specifically, after determining the first attention weight between connected nodes , the first feature of the connected nodes , the first learnable weight matrix After obtaining the specific value of , the specific value can be directly substituted into the calculation formula of the second feature for calculation, thereby obtaining the second feature.

[0117] In this embodiment, the first attention coefficient is normalized by using the softmax function to obtain the first attention weight, and the second feature of each node is obtained by weighted summing the first features of the connected nodes. In this way, the second feature of the node can integrate the first features of the connected nodes, strengthen the features of the connected nodes, suppress the influence of the first features of the unconnected nodes, and avoid information overload or noise interference.

[0118] In one embodiment, the calculation formula of the second attention coefficient in step S4 is: ;

[0119] in, Represents the second attention coefficient between the connected nodes i and j; LeakyReLU is a nonlinear activation function; for The transposed matrix of the second learnable parameter vector of dimension ; is the second learnable weight matrix, F is the input dimension, is the output dimension; is the second edge feature transformation matrix; is the water flow velocity vector from node i to node j; is the pollutant diffusion coefficient from node i to node j; represents the second feature of node i, Represents the second feature of node j.

[0120] Specifically, after determining the specific values ​​of the second learnable weight matrix, the second feature of node i, the second feature of node j, the second edge feature transformation matrix, the pollutant diffusion coefficient, and the water flow rate vector, the specific values ​​are directly substituted into the calculation formula of the second attention coefficient for calculation, so that the second attention coefficient can be obtained.

[0121] In some embodiments, the contaminant diffusion coefficient It can be determined based on the temperature and / or medium type between nodes i and j. Specifically, the initial diffusion coefficient of each medium is preset, and based on the target medium between the connected nodes, a matching medium that matches the target medium is searched from each medium, and the initial diffusion coefficient of the matching medium and the current temperature between the connected nodes are obtained; the initial diffusion coefficient is corrected by the current temperature to obtain the pollutant diffusion coefficient between the connected nodes. The higher the temperature, the more intense the molecular motion and the faster the diffusion speed, so the pollutant diffusion coefficient It will increase with increasing temperature. Different media have different resistances to the diffusion of pollutants. Generally speaking, the resistance of air medium is greater than that of soil medium, and the resistance of soil medium is greater than that of water medium.

[0122] In this embodiment, by considering the water flow rate and pollutant diffusion coefficient between connected nodes when calculating the second attention coefficient, the migration process of pollutants can be modeled more accurately, so that the simulated migration process of pollutants is more consistent with the actual migration process.

[0123] In one embodiment, the environmental variables also include rainfall and mining activity intensity of the target mine. The calculation formula of the second attention coefficient in step S4 is: ;

[0124] in, Represents the second attention coefficient between the connected nodes i and j; LeakyReLU is a nonlinear activation function; for The transposed matrix of the second learnable parameter vector of dimension ; is the second learnable weight matrix, F is the input dimension, is the output dimension; is the second edge feature transformation matrix; W R and W M is a learnable parameter; R t is the rainfall at time t, M t is the mining activity intensity at time t; is the water flow velocity vector from node i to node j; is the pollutant diffusion coefficient from node i to node j.

[0125] Learnable parameter W R and W M , respectively controlling the impact of rainfall and mining activities on the second attention weight. t and mining activity intensity M t is the normalized value. t When W increases R R tItem enhancement, the second attention coefficient of the edge related to water flow Significantly improved; similarly, the mining activity intensity M t By W M M t The second attention weights between connected nodes are dynamically adjusted to reflect the impact of human activities on pollutant migration. The mining activity intensity represents the mining frequency of the target mine.

[0126] Specifically, after determining the specific values ​​of the two learnable weight matrices, the second feature of node i, the second feature of node j, the second edge feature transformation matrix, the pollutant diffusion coefficient, the vector of water flow rate, the learnable parameters, the rainfall, and the intensity of mining activities, the specific values ​​are directly substituted into the calculation formula of the second attention coefficient for calculation, so that the second attention coefficient can be obtained.

[0127] In some embodiments, the contaminant diffusion coefficient It can be determined based on the temperature and / or medium type between node i and node j.

[0128] In this embodiment, when the rainfall in the target mine reaches a rainfall threshold and / or the mining activity intensity in the target mine reaches a second threshold, the second attention coefficient is calculated by combining the rainfall and the mining activity intensity. This can filter out data that has no significant impact on the prediction results, thereby avoiding the graph attention network from becoming complicated or overfitting due to processing too much meaningless data.

[0129] In one embodiment, the calculation formula of the second attention weight in step S4 is: ;

[0130] The calculation formula of the third feature of the node in step S4 is: ;

[0131] in, is the second attention weight between the connected nodes i and j, softmax is the activation function, represents the second attention coefficient between the connected nodes i and j, represents the second attention coefficient between the connected nodes i and k, is the set of nodes connected to node i; is the activation function, represents the second feature of node j, is the third feature of node i.

[0132] Specifically, after determining the second attention coefficient between connected nodes After obtaining the specific value, the specific value is directly substituted into the calculation formula of the second attention weight for calculation, so as to obtain the second attention weight.

[0133] Specifically, after determining the specific values ​​of the second attention weight between connected nodes, the second features of the connected nodes, and the second learnable weight matrix, the specific values ​​can be directly substituted into the calculation formula of the third feature for calculation, thereby obtaining the third feature.

[0134] The Softmax function is used to normalize the second attention coefficient to obtain the second attention weight. The third feature is obtained by weighted summing the second features of connected nodes. For example, the third feature of node A is obtained by weighted summing the second features of the nodes connected to node A.

[0135] Pollutant migration is a process of "distribution-transmission-reaction" coupled in series. When calculating the second attention coefficient, the water flow rate between two connected nodes, the pollutant diffusion coefficient, and the second characteristics of the two connected nodes are considered. The second attention weight, normalized by the second attention coefficient, is used to update the second characteristic. The third characteristic of the node is the transmission process.

[0136] In this embodiment, the second attention coefficient is normalized by using the softmax function to obtain the second attention weight, and the third feature of each node is obtained by weighted summing the second features of the connected nodes. In this way, the third feature of the node can integrate the second features of the connected nodes, strengthen the features of the connected nodes, suppress the influence of the second features of the unconnected nodes, and avoid information overload or noise interference.

[0137] In one embodiment, the calculation formula of the third attention coefficient in step S5 is: ;

[0138] The calculation formula is: ;

[0139] in, Represents the third attention coefficient between the connected nodes i and j; LeakyReLU is a nonlinear activation function; for The transposed matrix of the third learnable parameter vector of dimension ; is the third learnable weight matrix, F is the input dimension, is the output dimension; is the pollutant diffusion coefficient from node i to node j; || represents the vector concatenation operation, represents the redox potential of node i With the third feature The result of splicing, represents the redox potential of node j With the third feature The result of splicing.

[0140] Specifically, after determining the third characteristic and redox potential of the node, the result of splicing the redox potential of the node and the third characteristic can be obtained. The result of splicing the redox potential and the third characteristic is substituted into the calculation formula of the third attention coefficient for calculation, so that the third attention coefficient can be obtained.

[0141] In this embodiment, by considering the redox potential of each node when calculating the third attention coefficient, the chemical form and change process of pollutants in each node under different environmental conditions can be simulated, thereby achieving more detailed modeling and prediction of the migration and transformation behavior of pollutants.

[0142] In one embodiment, the calculation formula of the third attention weight in step S5 is: ;

[0143] The calculation formula of the fourth feature in step S5 is: ;

[0144] in, is the third attention weight between the connected nodes i and j, softmax is the activation function, represents the third attention coefficient between the connected nodes i and j, represents the third attention coefficient between the connected nodes i and k, is the set of nodes connected to node i; is the activation function, represents the redox potential of node i With the third feature The result of splicing, is the fourth feature of node i.

[0145] The Softmax function is used to normalize the third attention coefficient to obtain the third attention weight. The fourth feature is the weighted sum of the redox potential of the connected nodes and the concatenation of the third feature. For example, the fourth feature of node A is the weighted sum of the redox potential of the nodes connected to node A and the concatenation of the third feature.

[0146] Pollutant migration is a process of "distribution, transmission, and reaction" coupled in series. When calculating the third attention coefficient, the redox potential of the connected nodes and the result of concatenating the third feature are considered. The third feature is then updated using the third attention weight normalized by the third attention coefficient. The fourth feature of the node is the reaction process involved.

[0147] Specifically, after determining the third attention coefficient between connected nodes After obtaining the specific value, the specific value is directly substituted into the calculation formula of the third attention weight for calculation, so as to obtain the third attention weight.

[0148] Specifically, after determining the third attention weight between connected nodes, the concatenation result of the third feature and redox potential of the connected nodes, and the specific value of the third learnable weight matrix, the specific value can be directly substituted into the calculation formula of the fourth feature for calculation, thereby obtaining the fourth feature.

[0149] In this embodiment, the third attention coefficient is normalized by using the softmax function to obtain the third attention weight, and the fourth feature is obtained by weighted summing the redox potential of the connected nodes and the splicing result of the third feature. In this way, the fourth feature of the node can integrate the third feature and the redox reaction of the connected nodes, suppress the influence of the features of unconnected nodes, and avoid information overload or noise interference.

[0150] In one embodiment, the mapping formula for mapping the fourth feature to the predicted pollutant concentration in step S5 is: ;

[0151] Among them, c i is the predicted pollutant concentration at node i, is the fourth feature of node i, w out is the output weight matrix, b out is the bias vector. Specifically, the output weight matrix w out and the bias vector b out For known data, after obtaining the specific value of the fourth feature of the node, the output weight matrix w can be out , bias vector b out The specific value of the fourth feature is entered into the mapping formula for calculation, so that the predicted pollutant concentration of each node can be obtained.

[0152] By formula The process of calculating the predicted pollutant concentration is carried out in the fully connected layer of the graph attention network. Perform linear regression prediction.

[0153] In this embodiment, by using Performing linear regression prediction on pollutant concentrations can simplify the prediction process and quickly obtain the predicted pollutant concentrations at each node.

[0154] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0155] Based on the same inventive concept, the embodiments of the present application also provide a device for multi-process coupled prediction of mine pollutant migration based on deep learning, which is used to implement the multi-process coupled prediction method for mine pollutant migration based on deep learning. The implementation solution provided by this device is similar to the implementation solution described in the above method. Therefore, the specific limitations of one or more embodiments of the multi-process coupled prediction device for mine pollutant migration based on deep learning provided below can be referred to the limitations of the multi-process coupled prediction method for mine pollutant migration based on deep learning above, and will not be repeated here.

[0156] In one embodiment, Figure 3 As shown, a multi-process coupled prediction device for mine pollutant migration based on deep learning is provided, including:

[0157] A first feature acquisition module 302 is configured to acquire a first feature consisting of real-time data of each node in the target mine; the real-time data includes pollutant data;

[0158] A coefficient calculation module 304 is configured to calculate a first attention coefficient of the connected node based on the first feature of the node and the edge feature between the connected nodes;

[0159] A second feature acquisition module 306 is configured to update the first feature using the first attention weight normalized by the first attention coefficient to obtain a second feature of the node;

[0160] a third feature acquisition module 308 for calculating a second attention coefficient of the connected node based on the environmental variables of the target mine and the second feature, and updating the second feature based on a second attention weight normalized by the second attention coefficient to obtain a third feature of the node; the environmental variables include water flow rate and pollutant diffusion coefficient;

[0161] The concentration prediction module 310 is used to calculate the third attention coefficient of the connected node based on the redox potential of the node and the third feature, and update the third feature using the third attention weight normalized by the third attention coefficient to obtain the fourth feature of the node, and map the fourth feature to the predicted pollutant concentration.

[0162] Each module in the deep learning-based multi-process coupled prediction device for mine pollutant migration can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in hardware form, or stored in a computer device memory in software form, allowing the processor to call and execute the corresponding operations of each module.

[0163] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 4 As shown. The computer device includes a processor, a memory and a network interface connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store real-time data, a first feature, an edge feature, a first attention coefficient, a first attention weight, a second feature, an environmental variable, a second attention coefficient, a second attention weight, a third feature, an oxidation-reduction potential, a third attention coefficient, a third attention weight, a fourth feature and predicted pollutant concentrations. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a multi-process coupling prediction method for mine pollutant migration based on deep learning is implemented.

[0164] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0165] In one embodiment, a computer device is further provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.

[0166] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0167] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0168] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0169] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned embodiments. In particular, any reference to memory, database, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processors involved in the various embodiments provided herein may be, but are not limited to, general-purpose processors, central processing units (CPUs), graphics processing units (GPUs), digital signal processors (DSPs), programmable logic devices (PLDs), data processing logic devices based on quantum computing, and the like.

[0170] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0171] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A multi-process coupling prediction method for mine pollutant migration based on deep learning, characterized by: The method comprises: S1. Obtaining a first feature consisting of real-time data of each node in a target mine; the real-time data includes pollutant data; S2. Calculating a first attention coefficient of the connected node based on the first feature of the node and the edge feature between the connected nodes; S3. Update the first feature using the first attention weight normalized by the first attention coefficient to obtain a second feature of the node; S4. Calculating a second attention coefficient of the connected node based on the environmental variables of the target mine and the second feature, and updating the second feature based on a second attention weight normalized by the second attention coefficient to obtain a third feature of the node; the environmental variables include water flow rate and pollutant diffusion coefficient; S5. Based on the redox potential of the node and the third feature, calculate the third attention coefficient of the connected node, and use the third attention weight normalized by the third attention coefficient to update the third feature to obtain the fourth feature of the node, and map the fourth feature to the predicted pollutant concentration.

2. The method according to claim 1, characterized in that The method of obtaining the first feature in step S1 includes: Obtaining geological parameters, pollutant parameters, and meteorological parameters of each node in the target mine, and converting the geological parameters, pollutant parameters, and meteorological parameters into vector-type geological vectors, pollutant vectors, and meteorological vectors; splicing the geological vector, the pollutant vector, and the meteorological vector to obtain a first feature of each node; The expression for concatenating the first features of the nodes is: ； in, represents the first feature of node i; represents the geological vector of node i, p i represents the pollutant vector of node i, m i represents the meteorological vector of node i, || represents a vector concatenation operation; , d input represents the total dimension, represents the dimension of the geological vector, represents the dimension of the pollutant vector, Represents the dimension of the weather vector.

3. The method according to claim 1 or 2, characterized in that The calculation formula of the first attention coefficient in step S2 is: ； in, Represents the first attention coefficient between the connected nodes i and j, and LeakyReLU is a nonlinear activation function; for The transposed matrix of the first learnable parameter vector of dimension ; is the first learnable weight matrix, F is the input dimension, is the output dimension; is the first edge feature transformation matrix; is the partition coefficient calculation function, Represents the edge features between connected nodes i and j; represents the first feature of node i, Represents the first feature of node j.

4. The method according to claim 3, characterized in that The calculation formula of the first attention weight in step S3 is: ； The calculation formula of the second feature of the node in step S3 is: ； in, is the first attention weight between the connected nodes i and j; softmax is the activation function; represents the first attention coefficient between the connected nodes i and j, Represents the first attention coefficient between the connected nodes i and k; is the set of nodes connected to node i; is the activation function, represents the first feature of node j, is the first learnable weight matrix; is the second feature of node i.

5. The method according to claim 1, characterized in that The calculation formula of the second attention coefficient in step S4 is: ； in, Represents the second attention coefficient between the connected nodes i and j; LeakyReLU is a nonlinear activation function; for The transposed matrix of the second learnable parameter vector of dimension ; is the second learnable weight matrix, F is the input dimension, is the output dimension; is the second edge feature transformation matrix; is the water flow velocity vector from node i to node j; is the diffusion coefficient from node i to node j; represents the second feature of node i, Represents the second feature of node j.

6. The method according to claim 1, characterized in that The environmental variables also include the rainfall and mining activity intensity of the target mine. When the rainfall reaches a first threshold and / or the mining activity intensity reaches a second threshold, the calculation formula of the second attention coefficient in step S4 is: ； in, Represents the second attention coefficient between the connected nodes i and j; LeakyReLU is a nonlinear activation function; for The transposed matrix of the second learnable parameter vector of dimension ; is the second learnable weight matrix, F is the input dimension, is the output dimension; is the second edge feature transformation matrix; W R and W M is a learnable parameter; R t is the rainfall at time t, M t is the mining activity intensity at time t; is the water flow velocity vector from node i to node j; is the diffusion coefficient from node i to node j.

7. The method according to claim 5 or 6, characterized in that The calculation formula of the second attention weight in step S4 is: ； The calculation formula of the third feature of the node in step S4 is: ； in, is the second attention weight between the connected nodes i and j, softmax is the activation function, represents the second attention coefficient between the connected nodes i and j, represents the second attention coefficient between the connected nodes i and k, is the set of nodes connected to node i; is the activation function, represents the second feature of node j, is the third feature of node i.

8. The method according to claim 1, characterized in that The calculation formula of the third attention coefficient in step S5 is: ； The calculation formula is: ； in, Represents the third attention coefficient between the connected nodes i and j; LeakyReLU is a nonlinear activation function; for The transposed matrix of the third learnable parameter vector of dimension ; is the third learnable weight matrix, F is the input dimension, is the output dimension; is the diffusion coefficient from node i to node j; || represents the vector concatenation operation, represents the redox potential of node i With the third feature The result of splicing, represents the redox potential of node j With the third feature The result of splicing.

9. The method according to claim 8, characterized in that The calculation formula of the third attention weight in step S5 is: ； The calculation formula of the fourth feature in step S5 is: ； The mapping formula for mapping the fourth feature to the predicted pollutant concentration in step S5 is: ； in, is the third attention weight between the connected nodes i and j, softmax is the activation function, represents the third attention coefficient between the connected nodes i and j, represents the third attention coefficient between the connected nodes i and k, is the set of nodes connected to node i; is the activation function, represents the redox potential of node i With the third feature The result of splicing, is the fourth feature of node i; c i is the predicted pollutant concentration at node i, w out is the output weight matrix, b out is the bias vector.

10. A multi-process coupling prediction device for mine pollutant migration based on deep learning, characterized in that: The device comprises: A first feature acquisition module is used to acquire a first feature composed of real-time data of each node in the target mine; the real-time data includes pollutant data; A coefficient calculation module, configured to calculate a first attention coefficient of the connected node based on the first feature of the node and the edge feature between the connected nodes; A second feature acquisition module is configured to update the first feature using the first attention weight normalized by the first attention coefficient to obtain a second feature of the node; a third feature acquisition module, configured to calculate a second attention coefficient of the connected node based on the environmental variables of the target mine and the second feature, and update the second feature based on a second attention weight normalized by the second attention coefficient to obtain a third feature of the node; the environmental variables include water flow rate and pollutant diffusion coefficient; A concentration prediction module is used to calculate the third attention coefficient of the connected node based on the redox potential of the node and the third feature, and update the third feature using the third attention weight normalized by the third attention coefficient to obtain the fourth feature of the node, and map the fourth feature to the predicted pollutant concentration.

Images