Intelligent mine geological environment information evaluation method based on knowledge graph

Abstract

The invention relates to the field of mine geological environment monitoring and evaluation, in particular to an intelligent mine geological environment information evaluation method based on a knowledge graph. The method comprises the steps that mine geological environment data are collected and preprocessed, a knowledge graph is constructed based on the preprocessed mine geological environment data, and embedded vectors of nodes in the knowledge graph are obtained; calculating space-time similarity among the nodes based on the embedded vectors of the nodes; iteratively optimizing the embedded vectors of the nodes by adopting a graph neural network to obtain final embedded vectors of the nodes; a self-adaptive edge weight updating mechanism is introduced, and edge weights among the nodes are dynamically adjusted to obtain final edge weights among the nodes; and comprehensively evaluating the mine geological environment based on the final embedding vector of the nodes and the final edge weight between the nodes. The problems that a traditional monitoring and analysis method neglects spatial and temporal characteristics of a mine geological environment, and accurate prediction and risk early warning are difficult to carry out are solved; and real-time self-adaptive updating cannot be carried out, so that the timeliness of an evaluation result is poor.

Description

Technical Field

[0001] The present invention relates to the field of mine geological environment monitoring and evaluation, and particularly to an intelligent mine geological environment information evaluation method based on a knowledge graph. Background Art

[0002] With the continuous exploitation of mineral resources, the geological environment problems of mines have become increasingly prominent. Especially during the mine exploitation process, the changes in the geological environment have an important impact on mine safety, environmental protection, and sustainable development. The monitoring and evaluation of the mine geological environment are important means to ensure mine production safety, reduce the accident rate, and guarantee the life safety of mine workers.

[0003] Currently, mine geological environment monitoring systems generally rely on technical means such as sensors and remote sensing equipment to obtain various data of the mine environment. These data include geological features, climatic conditions, hydrological changes, ground subsidence, etc. However, with the explosion of monitoring data volume, traditional monitoring and analysis methods face problems such as information overload and data islands, and are unable to effectively integrate and process data from different sources and different types.

[0004] Traditional monitoring and analysis methods have the following technical problems: they often ignore the spatio-temporal characteristics of the mine geological environment and are unable to accurately capture the change laws of the mine geological environment under different time and space conditions, making it difficult for mine safety monitoring to achieve accurate prediction and risk warning; they lack an intelligent evaluation mechanism that can be updated in real time and adaptively, and it is difficult to quickly adapt to the complexity and variability of the mine environment, resulting in poor timeliness of evaluation results and being unable to timely reflect the actual impact of environmental changes on mine safety management. Summary of the Invention

[0005] The present invention provides an intelligent mine geological environment information evaluation method based on a knowledge graph to solve the problems that traditional monitoring and analysis methods often ignore the spatio-temporal characteristics of the mine geological environment and are unable to accurately capture the change laws of the mine geological environment under different time and space conditions, making it difficult for mine safety monitoring to achieve accurate prediction and risk warning; they lack an intelligent evaluation mechanism that can be updated in real time and adaptively, and it is difficult to quickly adapt to the complexity and variability of the mine environment, resulting in poor timeliness of evaluation results and being unable to timely reflect the actual impact of environmental changes on mine safety management.

[0006] An intelligent mine geological environment information evaluation method based on a knowledge graph of the present invention specifically includes the following technical solutions:

[0007] An intelligent mine geological environment information evaluation method based on a knowledge graph includes the following steps:

[0008] S1. Collect mine geological environment data and preprocess it to obtain preprocessed mine geological environment data; based on the preprocessed mine geological environment data, construct a knowledge graph and obtain the embedding vectors of the nodes in the knowledge graph; based on the embedding vectors of the nodes, calculate the spatio-temporal similarity between the nodes;

[0009] S2. Use a graph neural network to iteratively optimize the embedding vectors of the nodes to obtain the final embedding vectors of the nodes; based on the spatio-temporal similarity between the nodes, introduce an adaptive edge weight update mechanism to dynamically adjust the edge weights between the nodes to obtain the final edge weights between the nodes; based on the final embedding vectors of the nodes and the final edge weights between the nodes, comprehensively evaluate the mine geological environment to obtain the comprehensive evaluation value of the mine geological environment.

[0010] Preferably, the S1 specifically includes:

[0011] According to the characteristics and physical properties of the data, perform feature mapping on the preprocessed mine geological environment data to obtain the embedding vectors of each node.

[0012] Preferably, the S1 specifically includes:

[0013] Based on the embedding vectors of each node, introduce a similarity calculation method that fuses spatio-temporal relationships, combines the spatial and temporal information between the nodes, and obtains the spatio-temporal similarity between the nodes.

[0014] Preferably, the S1 specifically includes:

[0015] In the calculation process of the spatio-temporal similarity between the nodes, based on the cosine similarity, introduce the spatio-temporal distance between the nodes. The specific calculation formula for the spatio-temporal similarity between the nodes is as follows:

[0016]

[0017] Among them, S(v, u) represents the spatio-temporal similarity between node v and node u; and represent the embedding vectors of node v and node u at time t; D vu is the spatio-temporal distance between node v and node u; is the Euclidean norm of the embedding vector of node v at time t; is the Euclidean norm of the embedding vector of node u at time t; ∈ is a constant set to avoid division by zero errors; β is a regulation coefficient; ||D vu || represents the Euclidean norm of the spatio-temporal distance D vu of.

[0018] Preferably, the S1 specifically includes:

[0019] The spatio - temporal distance between nodes is calculated by combining the geospatial distance and time delay between the nodes.

[0020] Preferably, the S2 specifically includes:

[0021] Through multiple rounds of message passing mechanism, the graph neural network iteratively updates the embedding vector of the node based on the current embedding vector of the node and the embedding vectors of its neighbor nodes, introducing a weight matrix and the edge weights between nodes, to obtain the final embedding vector of the node.

[0022] Preferably, the S2 specifically includes:

[0023] An adaptive edge weight update mechanism is introduced. Based on the spatio - temporal similarity between nodes, combined with the difference measure of the embedding vectors of nodes, the edge weights between nodes are dynamically adjusted to obtain the final edge weights between nodes.

[0024] Preferably, the S2 specifically includes:

[0025] Based on the final embedding vector of the node and the final edge weights between nodes, and combined with the risk value of the node, a comprehensive assessment of the mine geological environment is carried out to obtain the comprehensive assessment value of the mine geological environment; the comprehensive assessment formula is:

[0026]

[0027] where E is the comprehensive assessment value of the mine geological environment; represents the set of nodes in the knowledge graph; is the embedding vector of node v after the last round of iteration, representing the final embedding vector of the node; K represents the total number of iterations; μ v is the weight coefficient of node v; ω is the weighting coefficient; is the edge weight between node v and node u in the K - th round of iteration, representing the final edge weight between nodes; Risk u is the risk value of node u.

[0028] The beneficial effects of the technical solution of the present invention are:

[0029] 1. The present invention adopts a similarity calculation method that fuses spatio - temporal relationships. It not only considers the spatial relationship between nodes but also comprehensively takes into account the time - varying factors, and can more accurately reflect the spatio - temporal dynamic characteristics in mine geological environment data caused by factors such as seasonal changes, climate fluctuations, or mining activities. It significantly improves the ability to capture changes in the mine geological environment, making the assessment and early warning of the mine geological environment more in line with the actual environmental changes.

[0030] 2. Based on the node embeddings, utilize the multi-round message passing mechanism of the graph neural network for further node information fusion and optimization. Through multiple rounds of iteration, the embedding vectors of the nodes can fuse multi-dimensional information from neighboring nodes, making the embedding vectors of each node more comprehensive, capable of effectively capturing potential relationships and changes in the mine geological environment. This not only improves the information transmission efficiency but also makes the evaluation results of the mine geological environment more timely and accurate.

[0031] 3. To adapt to the real-time changes in the mine geological environment, the present invention designs an adaptive edge weight update mechanism that can dynamically adjust the edge weights between nodes according to the changes in the spatio-temporal similarity between nodes, ensuring that the knowledge graph can reflect the latest state of the mine geological environment in real time and improving the reliability and timeliness of the evaluation results. By comprehensively considering the embedding vectors of the nodes, the edge weights between nodes, and the risk values of the nodes, the comprehensive evaluation value of the mine geological environment is calculated, which not only comprehensively considers different environmental factors but also takes into account the risk levels of each node, providing a scientific basis for mine safety management and assisting decision-makers to conduct risk early warning and resource allocation more accurately. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a flowchart of an intelligent mine geological environment information evaluation method based on a knowledge graph according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0033] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the intended invention purpose, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0034] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0035] The following specifically describes the specific solution of an intelligent mine geological environment information evaluation method based on a knowledge graph provided by the present invention in conjunction with the accompanying drawings.

[0036] Refer to the attached Figure 1 , which shows a flowchart of an intelligent mine geological environment information evaluation method based on a knowledge graph provided by an embodiment of the present invention. The method includes the following steps:

[0037] S1. Collect mine geological environment data and preprocess it to obtain the preprocessed mine geological environment data; based on the preprocessed mine geological environment data, construct a knowledge graph and obtain the embedding vectors of the nodes in the knowledge graph; based on the embedding vectors of the nodes, calculate the spatio-temporal similarity between the nodes.

[0038] By deploying a series of sensors at the mine site, such as soil moisture sensors, gas detectors, groundwater level detectors, etc., to collect mine geological environment data at different locations in the mining area all-weather, including but not limited to soil moisture, groundwater level, meteorological data (such as temperature, air pressure, humidity), ore composition, seismic wave information, and ore body deformation during the mining process. The sensors can measure multiple dimensions of the mine geological environment in real time and transmit the mine geological environment data to the data center through a wireless network.

[0039] Preprocess the collected mine geological environment data to obtain the preprocessed mine geological environment data; the preprocessing process includes cleaning, denoising, and standardization to ensure the quality and consistency of the mine geological environment data; in order to further improve the quality of the mine geological environment data, a missing data imputation method will also be applied during the preprocessing process, and an interpolation algorithm will be used to fill in some missing data to ensure that subsequent analysis will not be affected by data gaps.

[0040] Based on the preprocessed mine geological environment data, construct a knowledge graph. The nodes in the knowledge graph represent different data entities, such as monitoring points, sensors, environmental parameters, etc. The edges between the nodes represent the relationships or interactions between them, such as spatial adjacency, causal association, etc.

[0041] Specifically, when constructing the knowledge graph, it is necessary to convert the preprocessed mine geological environment data into structured graph data. Each data point is mapped to a node in the knowledge graph, and the attributes of the node include the timestamp, value range, and unit of the data; the relationships between the nodes are established through their physical attributes, spatial positions, or interaction relationships, which are represented as edges in the knowledge graph. For example, a node represents the data of a certain monitoring point, and the edge represents the spatial relationship or influence relationship between the monitoring points. The embedding vector of each node in the knowledge graph is generated by mapping the feature vector of the preprocessed mine geological environment data. The specific mapping formula is as follows:

[0042]

[0043] Among them, is the initial embedding vector of node v, and is the feature vector extracted from the preprocessed mine geological environment data d v through the feature mapping function; φ loc is the feature mapping function used to map the preprocessed mine geological environment data dv Map it into a high-dimensional embedded representation; d v is the preprocessed mine geological environment data corresponding to node v. According to the characteristics and physical properties of the preprocessed mine geological environment data, map it into a suitable space and use self-supervised learning methods to train to obtain the feature mapping function φ loc so as to better adapt to the differences of different data sources.

[0044] Based on the embedded vectors of each node, capture the potential correlation between nodes by calculating the similarity between nodes; traditional similarity calculation methods usually use cosine similarity to measure the similarity between nodes, and often cannot fully consider the changes in spatio-temporal features in mine geological environment data. The changes in mine geological environment data are spatio-temporal dynamic and are affected by various factors, such as seasonal changes, climate fluctuations, mining activities, etc., which will lead to differences in the same geological features at different times and spaces; therefore, a spatio-temporal relationship fusion similarity calculation method is designed, which combines the spatial and temporal information between nodes to obtain the spatio-temporal similarity between nodes, so as to more accurately reflect the spatio-temporal changes of mine geological environment data. The spatio-temporal similarity calculation formula between nodes is as follows:

[0045]

[0046] where S(v, u) represents the spatio-temporal similarity between node v and node u, indicating the degree of correlation between the two nodes in time and space; and represent the embedded vectors of node v and node u at time t; D vu is the spatio-temporal distance between node v and node u, indicating the difference between the two nodes in space and time; is the Euclidean norm of the embedded vector of node v at time t; is the Euclidean norm of the embedded vector of node u at time t; ∈ is a constant set to avoid division by zero errors; β is a regulation coefficient used to control the influence degree of spatio-temporal distance on similarity, obtained through experiments; ||D vu || represents the spatio-temporal distance D vuEuclidean norm. The first part of the formula measures the similarity between two nodes through cosine similarity, mainly reflecting the similarity between the embedding vectors of the nodes; the second part further adjusts the similarity between the two nodes by introducing the spatio-temporal distance, making the similarity between the nodes not only depend on the embedding vectors of the nodes, but also consider the factors of spatio-temporal changes, making the calculation of the similarity between the nodes more in line with the actual characteristics of the mine geological environment, because the changes in the mine geological environment are not only related to the spatial position, but also affected by the time factor. Through the similarity calculation method of spatio-temporal relationship fusion, it can better capture the environmental changes in different regions and at different time points in the mine geological environment, so as to provide a more accurate assessment and risk prediction.

[0047] The calculation of spatio-temporal distance takes into account the geographical spatial distance and time delay between two nodes, and can more realistically reflect the possible spatio-temporal relationship between different monitoring points in the mine geological environment. The specific calculation formula of spatio-temporal distance is as follows:

[0048] D vu =α1·||p v -p u ||+α2·|t v -t u |

[0049] Where, p v and p u represent the spatial coordinates (geographical location coordinates) of node v and node u respectively; t v and t u represent the timestamps of node v and node u respectively (such as the acquisition time of monitoring data); α1 and α2 are used to adjust the weights of the geographical spatial distance and time delay in the spatio-temporal distance respectively, and are obtained through experimental optimization.

[0050] S2. Use a graph neural network to iteratively optimize the embedding vectors of the nodes to obtain the final embedding vectors of the nodes; based on the spatio-temporal similarity between the nodes, introduce an adaptive edge weight update mechanism to dynamically adjust the edge weights between the nodes to obtain the final edge weights between the nodes; based on the final embedding vectors of the nodes and the final edge weights between the nodes, conduct a comprehensive assessment of the mine geological environment to obtain the comprehensive assessment value of the mine geological environment.

[0051] The graph neural network (GNN) is used to further process and optimize the embedding vectors of nodes. Through multiple rounds of message passing mechanisms, the embedding vectors of each node can integrate information from neighboring nodes, thereby capturing the relationships in the knowledge graph structure more comprehensively. In each iteration of the graph neural network, the embedding vectors of nodes are updated based on their current embedding vectors and the embedding vectors of neighboring nodes, reflecting the mutual influence and dependence between nodes (such as sensors or geological feature points) in the mine geological environment and other related nodes (such as adjacent sensors or geological feature points). Through repeated iterations, the embedding vectors of each node will gradually integrate more embedding vectors from neighboring nodes, enabling the embedding vectors of nodes to more comprehensively reflect the position of nodes in the knowledge graph and their relevance to other nodes. After the iteration is completed, the final embedding vectors of nodes are obtained. The iteration formula for the embedding vectors of nodes is as follows:

[0052]

[0053] Among them, is the embedding vector of node v at time t + 1, representing the embedding vector of node v after iteration; σ represents the activation function, and commonly used activation functions include ReLU or Sigmoid; W1 is the weight matrix, which is used to adjust the weight of the embedding vector of node v itself and is obtained through experimental tuning; represents the weighted sum of the embedding vectors from neighboring node u; is the neighbor set of node v; w vu is the edge weight between node v and its neighboring node u, representing the relationship strength between nodes. In each iteration, the embedding vector of node v continuously integrates the embedding vectors of surrounding neighboring nodes, gradually enhancing the perception and expression ability of the mine geological environment data in the knowledge graph.

[0054] During the processing of the graph neural network, the relationships between nodes will change in each iteration. As the mine geological environment data is continuously updated, it is necessary to adjust the edge weights in the knowledge graph to ensure that the knowledge graph can adapt to the changes in the mine geological environment data in real time. Therefore, an adaptive edge weight update mechanism is designed to dynamically adjust the edge weights between nodes by analyzing the changes in the spatio-temporal similarity between nodes and combining the difference measurement of the embedding vectors of nodes. After each iteration, the edge weights between nodes are not only affected by their initial values but also optimized according to the current spatio-temporal similarity between nodes, ensuring that the edge weights between nodes can reflect the latest state of the mine geological environment. After the iteration is completed, the final edge weights between nodes are obtained. The update formula for the edge weights between nodes is as follows:

[0055]

[0056] Among them, Denote the edge weight between node v and node u in the (k + 1)-th iteration. is the edge weight between node v and node u in the k-th iteration, representing the edge weight between nodes in the current iteration; γ is a regulation coefficient used to control the smoothness of the edge weight update between nodes, obtained through experiments. and represent the embedding vectors of node v and node u at the k-th iteration. is the difference metric between the embedding vectors of node v and node u at the k-th iteration, calculated using the Euclidean distance. By updating the edge weights between nodes, it is ensured that the knowledge graph can accurately reflect the changing trend of the mine geological environment.

[0057] By weighting the final embedding vectors of all nodes in the knowledge graph, a comprehensive evaluation of the mine geological environment is carried out to obtain the comprehensive evaluation value of the mine geological environment; the comprehensive evaluation value of the mine geological environment takes into account various factors of the mine geological environment, including geological characteristics, environmental monitoring data, and the spatio-temporal relationship between nodes; in addition, the risk values of nodes are also weighted to reflect the importance of different nodes in the evaluation of the mine geological environment. The comprehensive evaluation formula is:

[0058]

[0059] where E is the comprehensive evaluation value of the mine geological environment, representing the risk level obtained after comprehensive evaluation; represents the set of nodes in the knowledge graph; is the embedding vector of node v after the last iteration, representing the final embedding vector of the node; K represents the total number of iterations; μ v is the weight coefficient of node v, representing the importance of node v in the comprehensive evaluation, obtained through experimental tuning; ω is the weighting coefficient used to adjust the weight relationship between the final embedding vector of the node and the risk value of the node, obtained through experimental tuning. is the edge weight between node v and node u in the K-th iteration, representing the final edge weight between nodes; Risk u is the risk value of node u, representing the risk level of node u, set according to the expert experience method.

[0060] The comprehensive evaluation value of the mine geological environment obtained through comprehensive evaluation can reflect the overall state of the mine geological environment, providing a reliable basis for the safety management and decision-making of the mine.

[0061] In summary, an intelligent mine geological environment information evaluation method based on a knowledge graph is completed.

[0062] The sequence of the invention embodiments is only for description and does not represent the superiority or inferiority of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0063] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other. Each embodiment focuses on the differences from other embodiments.

[0064] The above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments or perform equivalent replacements for some of the technical features. These modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention and should all be included in the protection scope of the present invention.

Claims

1. An intelligent mine geological environment information evaluation method based on knowledge graph, characterized in that: The following steps are involved: S1, collecting and preprocessing the mine geological environment data to obtain the preprocessed mine geological environment data; Based on the preprocessed mining geological environment data, a knowledge graph is constructed and the embedding vectors of the nodes in the knowledge graph are obtained; based on the embedding vectors of the nodes, the spatiotemporal similarity between the nodes is calculated; S2. Use graph neural network to iteratively optimize the embedding vector of the node to obtain the final embedding vector of the node; Based on the spatiotemporal similarity between nodes, an adaptive edge weight update mechanism is introduced to dynamically adjust the edge weights between nodes to obtain the final edge weights between nodes; based on the final embedding vectors of the nodes and the final edge weights between nodes, a comprehensive assessment of the mining geological environment is carried out to obtain a comprehensive assessment value of the mining geological environment.

2. According to claim 1, a method for intelligent mine geological environment information evaluation based on knowledge graph, characterized in that: The S1 specifically includes: According to the characteristics and physical properties of the data, feature mapping is performed on the preprocessed mining geological environment data to obtain the embedding vector of each node.

3. According to claim 2, a method for intelligent mine geological environment information evaluation based on knowledge graph, characterized in that: The S1 specifically includes: Based on the embedding vector of each node, a similarity calculation method of spatiotemporal relationship fusion is introduced, and the spatiotemporal similarity between nodes is obtained by combining the spatial and temporal information between nodes.

4. According to claim 3, a method for intelligent mine geological environment information evaluation based on knowledge graph is characterized in that: The S1 specifically includes: In the process of calculating the spatiotemporal similarity between nodes, the spatiotemporal distance between nodes is introduced on the basis of cosine similarity. The specific calculation formula of the spatiotemporal similarity between nodes is as follows: Among them, S(v,u) represents the spatiotemporal similarity between node v and node u; and represents the embedding vector of node v and node u at time t; D vu is the spatiotemporal distance between node v and node u; is the Euclidean norm of the embedding vector of node v at time t; is the Euclidean norm of the embedding vector of node u at time t; ∈ is a constant set to avoid division by zero error; β is the adjustment coefficient; ||D vu || represents the space-time distance D vu The Euclidean norm of .

5. According to claim 4, a method for evaluating intelligent mine geological environment information based on knowledge graph is characterized in that: The S1 specifically includes: By combining the geographic spatial distance and time delay between nodes, the spatiotemporal distance between nodes is calculated.

6. The method for evaluating intelligent mine geological environment information based on knowledge graph according to claim 1 is characterized in that: The S2 specifically includes: The graph neural network, through a multi-round message passing mechanism, introduces a weight matrix and edge weights between nodes based on the current embedding vector of the node and the embedding vectors of neighboring nodes, iteratively updates the embedding vector of the node, and obtains the final embedding vector of the node.

7. The method for evaluating intelligent mine geological environment information based on knowledge graph according to claim 6 is characterized in that: The S2 specifically includes: An adaptive edge weight update mechanism is introduced to dynamically adjust the edge weights between nodes based on the spatiotemporal similarity between nodes and the difference measurement of the node embedding vectors to obtain the final edge weights between nodes.

8. The method for evaluating intelligent mine geological environment information based on knowledge graph according to claim 7 is characterized in that: The S2 specifically includes: Based on the final embedding vector of the node and the final edge weight between the nodes, and combined with the risk value of the node, a comprehensive assessment of the mine geological environment is conducted to obtain a comprehensive assessment value of the mine geological environment; the comprehensive assessment formula is: Among them, E is the comprehensive assessment value of the mine geological environment; Represents a set of nodes in a knowledge graph; is the embedding vector of node v after the last round of iteration, indicating the final embedding vector of the node; K indicates the total number of iterations; μ v is the weight coefficient of node v; ω is the weight coefficient; is the edge weight between node v and node U in the Kth iteration, indicating the final edge weight between nodes; Risk u is the risk value of node u.

Images