Mine disaster water source intelligent traceability system and method

Abstract

The invention provides a mine disaster water source intelligent traceability system and method, and belongs to the technical field of mine safety production, flood control and disaster emergency treatment. The method comprises the steps that a hydraulic network diagram is constructed based on hydrological data collected by sensors at all observation points in a mine laneway in real time, and then a mine global flow field is reconstructed in the hydraulic network diagram; on the basis of a mine global flow field, starting from each observation point, performing path backtracking along a reverse water flow direction, and generating a plurality of reverse backtracking paths; and according to the probability weight of each reverse backtracking path, calculating a cumulative probability weight corresponding to each position of the mine, and drawing a probability thermal distribution diagram by using the cumulative probability weight of each position. According to the method, path backtracking is carried out in the reverse water flow direction, potential water source positions are speculated from all observation points at the same time, multiple reverse paths are formed, the probability weights of the reverse paths are calculated, the most probable water source area can be recognized under the conditions of a complex roadway network and multi-source water gushing, and the decision reference value of the recognition result is improved.

Description

Technical Field

[0001] This application relates to the fields of mine safety production, flood prevention and disaster emergency response, and in particular to an intelligent source tracing system and method for mine disaster water sources. Background Technology

[0002] Mine water source identification is a crucial foundation for water control efforts. Accurate and efficient identification of water inrush sources is of great significance for coal mine safety, as mine water hazards are a major factor restricting mine safety. Existing technologies largely rely on manual experience and simple hydrogeological analysis to infer water sources, which suffers from problems such as untimely identification, low accuracy, and high uncertainty in results. Especially in situations with complex hydrogeological conditions and sparse sensor distribution, traditional methods struggle to accurately identify the location of water inrush sources.

[0003] In recent years, some scholars have proposed forward simulation or inversion methods based on hydraulic equations, but most of these methods have failed to fully integrate actual observation data and have poor adaptability to complex roadway networks and multi-source water inflow scenarios. Therefore, there is an urgent need for a mine disaster water source tracing system and method that can achieve high efficiency, accuracy, and intelligence under limited monitoring data conditions. Summary of the Invention

[0004] The purpose of this application is to provide an intelligent source tracing system and method for mine disaster water sources, so as to solve or alleviate the problems existing in the above-mentioned prior art.

[0005] To achieve the above objectives, this application provides the following technical solution: Firstly, this application provides an intelligent source tracing method for mine disaster water sources, including: Based on the real-time hydrological data collected by sensors at various observation points in the mine roadway, a hydraulic network diagram is constructed, and then the global flow field of the mine is reconstructed in the hydraulic network diagram. Based on the global flow field of the mine, starting from each observation point, the path is traced back along the direction of the reverse flow to generate multiple reverse tracing paths; Based on the probability weights of each reverse backtracking path, the cumulative probability weights corresponding to each location in the mine are calculated, and a probability heat map is drawn using the cumulative probability weights of each location. The probability heat map is used to indicate the location of one or more candidate water inrush points. Using candidate water inrush points as input, the final mine water inrush probability distribution map is obtained through forward hydrodynamic simulation and probability update.

[0006] In conjunction with the first aspect, in one possible implementation, the construction of the hydraulic network diagram includes: The mine roadway system is discretized into a topological structure consisting of nodes and edges, where nodes represent roadway intersections and key locations, and edges represent roadway segments connecting nodes. Assign geometric properties and hydraulic parameters to each edge to obtain a hydraulic network diagram.

[0007] In conjunction with the first aspect, in one possible implementation, reconstructing the global flow field of the mine in the hydraulic network diagram includes: By using the hydrological data collected in real time by sensors at various observation points and combining it with the continuity equation, the flow rate of each tunnel segment in the hydraulic network diagram is determined. By setting boundary conditions, flow direction data for unobserved areas are generated using interpolation methods to obtain a continuous flow field covering the entire mine.

[0008] In conjunction with the first aspect, in one possible implementation, based on the global flow field of the mine, multiple reverse backtracking paths are generated by tracing back along the direction of the reverse flow from each observation point, including: Calculate the unit reverse flow vector at each observation point; Based on the global flow field of the mine, starting from each observation point, the reverse backtracking path is generated by iteratively advancing along the direction of the unit reverse flow vector with a preset step size. The expression for the unit reverse flow vector is as follows: , In the formula, Represents the unit reverse flow vector. Indicates the position of any observation point (node). The velocity vector at that point express The model.

[0009] In conjunction with the first aspect, in one possible implementation, the generation of the reverse backtracking path further includes: When the reverse backtracking is advanced to the intersection of the roadways, the velocity vector and flow rate of each connected branch of the intersection are obtained; The inflow branch set is selected based on the dot product of the velocity vectors of each connected branch and the unit direction vectors pointing outward from the intersection node. For the set of inflow branches, the probability weight of the reverse path corresponding to each inflow branch is calculated according to the flow ratio, and then the probability weight of multiple reverse backtracking paths corresponding to the intersection node is generated.

[0010] In conjunction with the first aspect, in one possible implementation, based on the probability weights of each reverse backtracking path, the cumulative probability weights corresponding to each location in the mine are calculated, and a probability heatmap is drawn using the cumulative probability weights of each location, including: The frequency of each location in the mine roadway being traversed by the reverse backtracking path is counted, and the probability weight of the corresponding reverse backtracking path is accumulated and superimposed on the grid cell where each location is located to obtain the accumulated probability weight. A probability heat map is generated based on the cumulative probability weight, and the regions in the probability heat map with a cumulative probability weight greater than a preset weight threshold are taken as the locations of candidate water inrush points. The grid cells are obtained by discretizing the space of the mine roadway.

[0011] In conjunction with the first aspect, in one possible implementation, candidate water inrush points are used as input, and through forward hydrodynamic simulation and probability updates, the final mine water inrush probability distribution map is obtained, including: The flow rate values ​​of candidate inrush points in the probabilistic heat map are used as source terms input into the positive hydrodynamic model; A forward hydrodynamic simulation was performed using the Saint-Venant equation combined with the Manning formula to calculate the simulated values ​​of water level and flow velocity at each observation point. The simulated values ​​are compared with the actual observed values ​​to calculate the comprehensive error of the candidate water inrush points, and the probability distribution of each candidate water inrush point is updated according to the magnitude of the error.

[0012] In conjunction with the first aspect, in one possible implementation, updating the probability distribution of each candidate inrush point based on the magnitude of the error includes: An error adjustment factor is introduced to control the probability decay rate, and probability normalization is performed after each adjustment to complete the probability update for this iteration. Repeat the forward simulation and probability update process until the change in probability distribution between two consecutive iterations is less than a preset probability threshold, at which point the iteration stops.

[0013] In conjunction with the first aspect, in one possible implementation, the sensors are deployed in the goaf, fault fracture zone, roadway intersection and drainage pipeline node of the mine roadway, and are locally densified in key areas. The hydrological data is uploaded to the main control center via a wireless network and, after data cleaning and missing data interpolation, is used to reconstruct the global flow field of the mine.

[0014] Secondly, this embodiment provides an intelligent source tracing system for mine disaster water sources. This system is used to execute the intelligent source tracing method for mine disaster water sources provided in any of the above embodiments, including: The flow field construction unit is used to construct a hydraulic network diagram based on the hydrological data collected in real time by sensors at various observation points in the mine roadway, and then reconstruct the global flow field of the mine in the hydraulic network diagram. The backtracking unit is used to backtrack along the direction of the reverse water flow from each observation point based on the global flow field of the mine, generating multiple reverse backtracking paths; The candidate water inflow point generation unit is used to calculate the cumulative probability weight corresponding to each location in the mine based on the probability weight of each reverse backtracking path, and to draw a probability heat map using the cumulative probability weight of each location. The probability heat map is used to indicate the location of one or more candidate water inflow points. The iterative update unit is configured to take candidate water inrush points as input and obtain the final mine water inrush probability distribution map through forward hydrodynamic simulation and probability update.

[0015] The technical solution of this application embodiment has the following beneficial effects: 1. By abstracting the mine roadway system into a hydraulic network diagram and reconstructing the global flow field by combining the hydrological data collected in real time by sensors at various observation points, the flow direction and flow rate of each roadway section can be quantitatively characterized. This overcomes the limitations of traditional methods that rely solely on local water levels or experience and cannot reflect the overall flow characteristics, and significantly improves the spatial resolution of the mine flow process.

[0016] 2. By tracing back along the direction of the reverse water flow, the location of potential water sources can be inferred from various observation points simultaneously, forming multiple reverse paths and calculating their probability weights. This can identify the most likely water source area in complex tunnel networks and multi-source water inflow situations, avoiding the accumulation of errors caused by traditional one-way simulation or single-point inference. At the same time, the probability weights of each reverse path are accumulated to create a probability heat map, which intuitively reflects the spatial probability characteristics of the potential water source distribution, improving the interpretability and decision-making reference value of the water inflow point identification results.

[0017] 3. Based on the combination of probabilistic modeling and probabilistic thermal distribution maps, even with a small number of observation points, reliable water source location estimates can be obtained through flow field reconstruction and probabilistic diffusion, which significantly enhances the robustness and adaptability of the system under complex geological conditions. Furthermore, the sensor network can be linked with the computational models formed in each step to realize dynamic updates and real-time identification of water inrush points, providing timely data support for water inrush early warning and emergency decision-making.

[0018] In summary, the technical solution of this embodiment, by constructing a hydraulic network and combining reverse flow field backtracking and probability analysis, can achieve efficient, accurate, and visualized intelligent source tracing of mine disaster water sources under limited observation conditions, significantly improving the scientific nature and real-time performance of mine water control work. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the structure of an electronic device provided according to some embodiments of this application.

[0020] Figure 2 This is a flowchart illustrating a method for intelligent tracing of mine disaster water sources according to some embodiments of this application.

[0021] Figure 3 This is an example of a system module structure.

[0022] Figure 4 This is a technical logic block diagram of the monitoring module.

[0023] Figure 5 This is a technical logic block diagram of the data processing module.

[0024] Figure 6 This is the technical logic diagram of the algorithm module.

[0025] Figure 7 The technical logic diagram for the verification module and the error calculation module.

[0026] Figure 8 This is another flowchart illustrating the intelligent source tracing method for mine disaster water sources provided in this embodiment.

[0027] Figure 9 This is an example of a hydraulic network diagram.

[0028] Figure 10 This is an example of a probability distribution map (i.e., a probability heatmap) for mine water inrush. Detailed Implementation

[0029] The embodiments of this application will now be described with reference to the accompanying drawings.

[0030] The method provided in the embodiments of this application can be applied to... Figure 1 Among the electronic devices shown, the electronic devices may be, but are not limited to, mobile terminals such as mobile phones, tablets, handheld computers, and personal digital assistants (PDAs), smart home devices such as smart TVs and smart cameras, wearable devices such as smart bracelets, smartwatches, and smart glasses, or other computer devices such as desktop, laptop, notebook, ultra-mobile personal computer (UMPC), netbook, and smart screen.

[0031] like Figure 1 As shown, the electronic device 200 may include one or more of the following components: a processor 201, a memory 203, a communication interface 202, and a communication bus 204. The memory 203 can be connected to the processor 201 via the bus 204. The bus can transfer data between the processor 201 and the memory 203. The bus can be divided into an address bus, a data bus, a control bus, etc.

[0032] Processor 201 may include one or more processing cores. Processor 201 can connect to various parts within the electronic device 200 using various interfaces and lines. It performs various functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in memory 203, and by calling data stored in memory 203. For example, processor 201 may include an application processor (AP), a modem processor, a CPU, a graphics processing unit (GPU), an image signal processor (ISP), a controller, a video codec, a digital signal processor (DSP), a field-programmable gate array (FPGA), a programmable logic array (PLA), and / or a neural network processing unit (NPU). The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content to be displayed; the NPU implements artificial intelligence (AI) functions; and the modem handles wireless communication. Different processing units can be independent devices or integrated into one or more processors. For example, the multiple processing units shown above are all integrated into a single SoC, or the AP is a separate semiconductor chip, while other processing units are integrated into a single SoC. This application does not limit this to any particular type.

[0033] The memory 203 may include random access memory (RAM), read-only memory (ROM), or non-transitory computer-readable storage medium. The memory 203 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 203 may include a program storage area and a data storage area. The program storage area may store instructions for implementing an operating system or instructions for at least one function, such as a smart source tracing method for mine disaster water sources. The data storage area may store data created based on the use of the electronic device 200, such as real-time hydrological data collected by sensors at various observation points, probability weight calculation results, and probability heat map distribution.

[0034] In addition, those skilled in the art will understand that the structure of the electronic device 200 shown in the above figures does not constitute a limitation on the electronic device 200. The electronic device may include more or fewer components than shown, or combine certain components, or have different component arrangements. For example, the electronic device 200 may also include components such as a microphone, speaker, radio frequency circuit, sensor, audio circuit, power supply, and Bluetooth module, which will not be described in detail here.

[0035] This embodiment provides a method for intelligent tracing of water sources in mine disasters, such as... Figure 2 As shown, the method includes: Step S1: Based on the hydrological data collected in real time by sensors at various observation points in the mine roadway, construct a hydraulic network diagram, and then reconstruct the global flow field of the mine in the hydraulic network diagram.

[0036] In this embodiment, multiple observation points are pre-deployed in the mine roadway system. Each observation point is equipped with a hydrological sensing device, and the multi-point deployment of water level, flow velocity, and pressure sensors constitutes a monitoring module. The hydrological sensing devices may include, but are not limited to, water level sensors, water pressure sensors, flow velocity sensors, conductivity sensors, and temperature sensors. These multi-point deployment of water level, flow velocity, and pressure sensors form a corresponding monitoring network for real-time monitoring of hydrological state parameters (i.e., hydrological data) in various local areas of the roadway, including water level and flow velocity in each roadway.

[0037] Furthermore, sensors are deployed in the goaf, fault fracture zone, roadway intersections, and drainage pipeline nodes of the mine roadways, with localized denser deployment in key areas; hydrological data is uploaded to the main control center via wireless network, and after data cleaning and missing interpolation processing, it is used to reconstruct the overall flow field of the mine.

[0038] The monitoring module is also used to determine key monitoring areas (critical locations) based on the mine's hydrogeological conditions, historical water inrush point distribution, and roadway layout. Key areas include goaf areas, areas adjacent to old goaf water bodies, fault fracture zones, roadway intersections, and drainage pipeline nodes. Roadway intersections and critical locations are defined as "nodes" and their coordinates are recorded to characterize key control points in the water flow path.

[0039] Sensors are deployed at appropriate locations (such as intersections or key monitoring areas). A differentiated deployment strategy based on regional variations is adopted, meaning multiple types of sensors, such as water level gauges and flow meters, are deployed across different areas at varying elevations and with varying hydrological conditions. Sensor deployment follows the principle of even distribution with localized density; for example, more points are placed at the entrances of key roadways, around the working face, and in old workings to ensure accurate collection of water level and flow data at regular time intervals in the event of mine water disasters.

[0040] Sensors at each observation point upload the collected data to the mine's main control center (also known as the monitoring center) via wireless communication (such as 5G). The main control center is equipped with data acquisition terminals for various processing operations on the collected data.

[0041] Furthermore, data transmission can employ 5G multi-redundant transmission links. Specifically, this utilizes 5G communication technology to simultaneously establish two or more independent wireless data transmission channels (i.e., "multi-path"), transmitting the same sensor data in parallel through multiple physical or logical paths. When one link is interrupted due to tunnel collapse, equipment failure, electromagnetic interference, or signal obstruction, the remaining links can continue transmitting data, thus achieving redundant backup and seamless switching of communication links. Specific implementation methods include multi-base station redundant access or multi-band / multi-carrier aggregation. Multi-base station redundant access can involve deploying multiple 5G small cells or mining 5G private network base stations in key areas of the tunnel. Sensor terminals (such as 5G industrial gateways) can register with multiple base stations simultaneously, dynamically selecting the link with the best signal quality, or simultaneously transmitting the same data through multiple base stations. Multi-band / multi-carrier aggregation utilizes different frequency bands supported by 5G, such as Sub-6GHz and millimeter wave (mmWave), or multiple carriers within the same frequency band, to construct multiple logical transmission channels. By establishing 5G multi-redundant transmission links, stable data transmission is ensured in complex environments before and after water inrush, enhancing the system's disaster recovery capabilities and stability.

[0042] In addition, the system is designed with an automatic verification mechanism, in which the front-end device filters outliers and the back-end platform performs consistency comparison and missing data interpolation.

[0043] To ensure the quality of data acquisition, the monitoring center is also equipped with an inspection function, which is used to conduct regular online inspections of all sensor devices and promptly repair any problems found.

[0044] After acquiring multi-source hydrological data collected by sensors at the mine site, the data processing module performs standardized preprocessing to ensure that the data input into the inversion algorithm has consistency, integrity, and high confidence, providing a reliable basis for subsequent flow field reconstruction and reverse tracing.

[0045] Specifically, the data acquisition terminal in the main control center receives water level and flow velocity data collected by sensors, converts the data uploaded by different sensors into a unified data structure and encoding format, assigns a unique label to each data point, and stores it in a standard MySQL database for subsequent algorithm modules to retrieve the data.

[0046] To further improve data quality, sliding window local analysis and statistical discrimination methods are used to automatically identify and remove outliers.

[0047] Among them, sliding window local analysis is used to detect local anomalies in the data stream without being affected by long-term trends. The specific method is as follows: for each data stream acquired by the sensor, a sliding window of length W is set, and the window is slid one step at a time to calculate the mean of the data within the window. and standard deviation Compare the data value at the center of the window with the mean. Compare, if the following conditions are met: Then mark For local outliers, where, It is a positive integer, and its value is 2 or 3.

[0048] Local outliers marked The Grubbs test was used to confirm the results. If the results were confirmed as extreme outliers, they were removed; otherwise, they were retained.

[0049] The Grubbs test is a classic method for detecting single outliers in a univariate dataset, and it is suitable for small samples. The specific steps are as follows: First, calculate the global mean. and standard deviation Then for each local outlier The formula for calculating the G value is as follows: (1) Set the confidence level (e.g., 95%) and look up the critical value in the table. ,if If so, the point is confirmed as an extreme outlier and removed; if If the condition is met, the point is confirmed as a normal point and retained. An anomaly data log is then recorded.

[0050] Furthermore, preprocessing includes supplementing reasonable values ​​using kriging interpolation. Kriging is an optimal unbiased linear estimation method based on spatial correlation and variogram; for example, it can be used to estimate values ​​for a removed outlier. The corresponding water level value is Known outliers There are N known observation points in the surrounding area, denoted by { If} represents the number of observation points, then outliers can be estimated using N observation points. The location data values ​​are as follows: , (2) In the formula, These are the weighting coefficients for each known observation point. For the first Known observation points water level value, outlier The estimated location.

[0051] in, The solution can be obtained using the Lagrange multiplier method, i.e., by constructing the following system of equations: (3) In the formula, For any two known observation points and The semivariogram values ​​between It is a Lagrange multiplier.

[0052] The above system of equations can be solved using linear algebra to obtain the optimal weighting coefficients for each known observation point relative to the missing point, i.e. .

[0053] In this embodiment, the hydraulic network diagram is a graph theory model used to abstractly represent the water flow path and its hydraulic connection relationship in mine roadways. The construction of the hydraulic network diagram includes the following steps: discretizing the mine roadway system into a topological structure composed of nodes and edges, where nodes represent roadway intersections and key locations, and edges represent roadway segments connecting nodes; assigning geometric properties and hydraulic parameters to each edge to obtain the hydraulic network diagram.

[0054] In this embodiment, the monitoring module performs topological discretization processing on the mine roadway system, discretizing the mine roadway system into a topological structure composed of nodes and edges. For example, based on the three-dimensional mine roadway map, three-dimensional laser scanning data, or roadway digital model, key structural features in the roadway can be identified, including roadway intersections and key locations (also known as key monitoring areas).

[0055] Adjacent nodes are connected by actual roadway segments. Each roadway segment is abstracted as an "edge" in the graph structure. That is, each edge corresponds to a physical roadway segment, which connects two nodes and has clear geometric and hydraulic properties.

[0056] Furthermore, the geometric attributes of each side include: tunnel length, cross-sectional shape (such as rectangle, arch or circle), cross-sectional area, etc. These geometric attributes can be obtained from tunnel design drawings, digital tunnel models (such as BIM models) or by scanning with LiDAR carried by tunnel inspection robots. This embodiment does not limit the content and acquisition method of the geometric attributes.

[0057] Hydraulic parameters are used to describe the transmission characteristics of water flow in the tunnel section, and mainly include: water level, water flow velocity, etc.

[0058] This embodiment uses discretization to transform the entire complex three-dimensional tunnel system into a directed or undirected graph consisting of a finite number of nodes and edges, i.e., a topological graph. The hydraulic network graph constructed therefrom not only preserves the spatial topological relationship of the mine tunnel system, but also integrates key hydraulic characteristics, providing a structured and computable digital foundation for flow field deduction, water inrush path tracing, and risk area identification based on graph models.

[0059] In this embodiment, the global flow field of the mine is reconstructed in the hydraulic network diagram. That is, based on the hydrological data of the nodes and the hydraulic relationships of the edges in the hydraulic network diagram, the flow field is reconstructed through numerical simulation or data-driven methods, and the velocity field, pressure field and flow direction distribution of the water flow in the entire mine roadway system are deduced.

[0060] In this embodiment, reconstructing the global flow field of the mine in the hydraulic network diagram includes: using hydrological data collected in real time by sensors at various observation points and combining it with the continuity equation to determine the flow rate of each roadway segment in the hydraulic network diagram; setting boundary conditions and generating flow direction data of unobserved areas through interpolation methods to obtain a continuous flow field covering the entire mine.

[0061] The main purpose of the above steps is to combine the velocity and flow rate data of the observation points, utilize the continuity equation and velocity-flow rate relationship, and supplement the flow direction data of the unobserved areas through interpolation and boundary conditions, thereby achieving a complete reconstruction of the hydrodynamic state of the entire mine.

[0062] In fluid mechanics, for incompressible fluids (such as mine water), the mass conservation can be expressed by the continuity equation as follows: within any control volume, the inflow rate equals the outflow rate (steady state) or equals the rate of change of the water volume within the volume (non-steady state). In this embodiment, for the roadway section (i.e., the "edge" in the hydraulic network diagram), combining the velocity and flow rate data at the observation points, a simplified one-dimensional flow model is adopted using the continuity equation and the velocity-flow relationship. The basic relationship is as follows: (4) in, For the cross-sectional flow rate of the tunnel, The cross-sectional area of ​​the tunnel. Flow rate.

[0063] Formula (4) serves as a bridge connecting the observed velocity and the flow distribution. In practical applications, if the velocity at a certain node is known... Then the cross-sectional area of ​​the edge connected to that node can be used. Calculate the flow rate on the corresponding edge. Conversely, if the flow rate is known... and cross-sectional area The flow rate can also be deduced from this. .

[0064] Treating each node in the hydraulic network diagram as a "sink," and based on the principle of mass conservation, a flow balance equation is established for each node. For example, the steady-state flow balance equation is: (5) In the formula, , These are the inflow and outflow of the sink, respectively.

[0065] The system of linear equations consists of the flow balance equations of each node, with the unknowns being the flow of each edge. By solving this system of equations, the flow distribution on all unobserved edges can be derived, thus reconstructing the global flow field from local observation.

[0066] Since sensors cannot cover all roadway sections, direct velocity or flow rate data is lacking in some areas. In such cases, interpolation and boundary conditions are used to supplement flow direction data in unobserved areas. (1) Spatial interpolation method.

[0067] Based on the velocity / flow rate values ​​of nearby observation points, the velocity or flow rate of unobserved edges is estimated using inverse distance weighted (IDW), Kriging, or finite element interpolation methods.

[0068] (2) Boundary condition constraint method.

[0069] At the boundaries of the tunnel entrance (such as the aquifer recharge point) and exit (such as the drainage pumping station or water tank), set known flow or head boundary conditions as constraints for the equation system to guide the reasonable distribution of the internal flow field.

[0070] Ultimately, the flow of each side will be... With flow rate Mapped to three-dimensional tunnel space coordinates, a vectorized global flow field is formed. This flow field includes: the flow rate and direction of each tunnel segment, and the flow rate and velocity vectors at each node. This global flow field can be directly used for subsequent disaster water source identification and tracing.

[0071] Step S2: Based on the global flow field of the mine, starting from each observation point, backtrack along the direction of the reverse flow to generate multiple reverse backtracking paths.

[0072] In this embodiment, in order to achieve dynamic tracking and reverse tracing of mine water inrush risk, based on the successful reconstruction of the mine global flow field in step S1, multiple possible upstream paths, i.e. reverse tracing paths, are automatically tracked and generated from each observation point as the starting point, according to the reverse direction information of water flow in the flow field.

[0073] In this context, given the known global flow field (i.e., the flow direction and velocity of each tunnel segment), the reverse flow direction refers to the direction towards potential water sources or recharge areas. Step S2 utilizes this physical characteristic, using the reverse flow vector in the flow field as the guiding direction for path tracing, and tracing back from the downstream observation point to the upstream to reveal the possible incoming path of the water flow.

[0074] In one specific embodiment, based on the global flow field of the mine, multiple reverse backtracking paths are generated by tracing back along the direction of the reverse water flow from each observation point, including: Step S21: Calculate the unit reverse flow vector at each observation point.

[0075] Assuming the location of any observation point in the mine is... This means that, based on the overall flow field of the mine, it can be extracted... Flow velocity vector at location The unit reverse flow vector is calculated using the following formula: (6) In the formula, Represents the unit reverse flow vector. Indicates the position of any observation point (node). The velocity vector at that point express The model.

[0076] Step S22: Based on the global flow field of the mine, starting from each observation point, iteratively advance along the direction of the unit reverse flow vector with a preset step size to generate a reverse backtracking path.

[0077] The preset step size is a fixed, small step size (such as 0.1m or 0.5m), used... express.

[0078] In this embodiment, a reverse backtracking path composed of discrete coordinate points is generated by starting from the observation point and gradually tracing back along the direction of water flow in the three-dimensional tunnel space.

[0079] Specifically, starting from each observation point, the direction of the unit reverse flow vector is used as the initial direction for backtracking, using a small step size. Backtracking along the reverse unit flow (e.g., performing a reverse path search in a hydraulic network graph), the position update formula during iterative advancement is: (7) In the formula, Indicates the first Step to the current iteration position, This is the next iteration position. Indicates position The unit reverse flow vector at the location is obtained by formula (6). It is the iteration step size (i.e., the preset step size).

[0080] In formula (7), each step Each point moves a small distance upstream from the current point, thus gradually constructing a continuous path from the downstream observation point to the potential water source, i.e., a reverse backtracking path.

[0081] Record the set of points traversed by the reverse backtracking path, denoted as . This forms a reverse trajectory.

[0082] Furthermore, during the reverse backtracking process in step S2, it is inevitable that the roadway intersections (such as three-way intersections, four-way intersections, and the connection points between the main roadway and the branch roadway) will be encountered. The intersection is defined as a node with an in-degree greater than 1. In this embodiment, a branch decision strategy is adopted—that is, to determine which branch is the direction from which the water flow actually originates, starting from the current node, so as to ensure that the backtracking path continues to extend along the direction of the real water source.

[0083] Specifically, when the reverse backtracking progresses to the intersection of the tunnels, the following operations are performed: Step S2a: Obtain the velocity vector and flow rate of each connected branch of the intersection node.

[0084] Step S2b: Based on the velocity vectors of each connected branch and the dot product of the unit direction vectors pointing outward from the intersection node, the set of inflow branches is selected.

[0085] Step S2c: For the set of inflow branches, calculate the probability weight of the reverse path corresponding to each inflow branch according to the flow ratio, and then generate the probability weight of multiple reverse backtracking paths (also known as sub-paths) corresponding to the intersection node.

[0086] During the reverse trajectory advancement, each step checks whether the current position is at the intersection node. If so, a branch decision process is triggered.

[0087] In the branch decision-making process, the first step is to query all adjacent edges (i.e., connected branches) of the current node from the hydraulic network graph and extract the velocity vector of each edge. and traffic and the unit direction vector of each edge. .

[0088] Wherein, the unit direction vector Used to indicate the direction in which a branch leaves the node at a junction, it is defined as: for each branch path connected to the current node, The vector pointing from a node to the outside is calculated as follows: If a branch channel is a straight line segment, The coordinates are obtained by subtracting the node coordinates from the end coordinates of the branch and then normalizing. If the branch is a curve or a complex geometric structure, the tangent direction at the node can be taken as the branch's coordinates. .

[0089] Then, based on the velocity vectors of each connected branch and the dot product of the unit direction vectors pointing outward from the intersection node, the branch direction is determined, and the inflow branches flowing into the node in the positive flow field are selected. The determination logic is as follows: (8) In the formula, Let be the set of inflow branches of the current intersection node. For branches The unit direction vector pointing outward from the node.

[0090] Formula (8) retains only the branches from which water flows into the current node from the outside as candidate paths for backtracking.

[0091] After obtaining the candidate paths for reverse backtracking, in step S2c, the probability weights of the reverse sub-paths are calculated based on the flow of the inflow branch, as follows: Suppose that at the current intersection node, the set of inflow branches selected is For each inflow branch , Since the water flow also follows the law of mass conservation at the confluence node, the sum of the flow rates of each inflow branch equals the total inflow. Therefore, in this embodiment, the probability weight of the reverse sub-path of each inflow branch is calculated according to the flow rate ratio. The calculation formula is as follows: (9) In the formula, Indicates the first The flow weight, i.e., the probability weight, of the reverse sub-path corresponding to each inflow branch. Indicates the first The flow of each inflow branch This represents the total flow of all inflow branches.

[0092] The denominator of formula (9) represents the sum of the flow rates of all inflow branches, i.e. Molecules are branches Traffic ,therefore, It is calculated based on the proportion of traffic flow. The value of represents the probability that water flows from that branch into the node.

[0093] Subsequently, a full traversal approach is used to continue backtracking along each incoming branch, generating the reverse backtracking path corresponding to each incoming branch. The specific steps are as follows: Traverse the set of inflow branches Each inflow branch This generates a corresponding reverse sub-path. The step size is used to continue backtracking along the reverse sub-path. If a new intersection node is reached again, the set of inflow branches at the new intersection is selected according to the aforementioned steps, and the probability weight of the reverse path corresponding to each inflow branch at the new intersection node is calculated using formula (9). This process is repeated until the boundary node is reached, at which point the backtracking stops, and the reverse backtracking trajectory is recorded. The boundary node is defined as a preset node.

[0094] After the above steps, when tracing back to the junction, each sub-path starts from the current junction node and continues to trace back against the direction of the corresponding inflow branch. Whenever the reverse tracing path reaches a junction node, if the node has several inflow branches, the path will split into several reverse sub-paths. Each reverse sub-path is calculated with different probability weights and proceeds independently, eventually forming a tree-like reverse tracing path network that covers all possible combinations of water source paths.

[0095] For each reverse sub-path, a new reverse backtracking is initiated, that is, traversing each reverse sub-path in the hydraulic network diagram one by one, starting again from the other end of the inflow branch (i.e., the upstream node) with a step size. Perform reverse path backtracking until the preset termination condition is met (e.g., reaching the preset boundary).

[0096] Step S3: Calculate the cumulative probability weight corresponding to each location in the mine based on the probability weight of each reverse backtracking path, and draw a probability heat map using the cumulative probability weight of each location. The probability heat map is used to indicate the location of one or more candidate water inflow points.

[0097] Furthermore, step S3 includes the following sub-steps: Step S31: Count the frequency of each location in the mine roadway being traversed by the reverse backtracking path, and accumulate the probability weight of the corresponding reverse backtracking path and superimpose it onto the grid cell where each location is located to obtain the accumulated probability weight.

[0098] Step S32: Generate a probability heat map based on the cumulative probability weight, and take the area in the probability heat map where the cumulative probability weight is greater than the preset weight threshold as the location of the candidate water inrush point.

[0099] The grid cell is obtained by discretizing the space of the mine roadway.

[0100] In this embodiment, to facilitate statistics and visualization, the mine roadway space is discretized to obtain a regular grid (such as a 1m×1m grid or a 0.5m×0.5m grid), and each grid cell is denoted as... , These are the coordinates of the center of the grid cell.

[0101] This involves counting the frequency with which each location in a mine roadway is traversed by a reverse tracing path. Specifically, it means counting how many paths cover each location or grid cell. More specifically, it involves counting each reverse sub-path. Define whether it covers Location indicator function for: (10) Since a path is a set of points, therefore, when determining a path... Have you passed through the location? At this time, path judgment can be used. The previous point fell into the grid cell Internal methods for identifying grid cells Path Coverage, i.e., path Passing through the location The above judgment can be achieved using overlay analysis in existing GIS software, which will not be elaborated upon here for the sake of brevity.

[0102] Finally, for each spatial location (or grid cell), calculate its weighted probability of being covered by all paths, i.e., its spatial location. The cumulative probability weights of each sub-path are: (11) In the formula, Spatial location The cumulative probability weight of each sub-path, also known as a grid cell. The cumulative weight, For path The cumulative allocation weight, For path In grid cells The indicator function at the location, according to formula (10), when the path After grid cells hour, ,otherwise, .

[0103] Formula (11) shows that if a certain grid cell The more high-probability paths cover a location, the higher its cumulative probability weight. Equivalent to covering the grid cell The sum of probability weights for all paths to the location. The higher the value, the more grid cells it represents. The more likely the location is to be a source of sudden water flow.

[0104] By spatial discretization, path coverage statistics, and probability accumulation, discrete path information is aggregated into a continuous probability field, which is then used to identify high-risk candidate water inrush points.

[0105] Based on the calculation results of formula (11), a probability heat map of the candidate regions for water inrush points is generated. The probability heat map can be generated as follows: [The text abruptly ends here, likely due to an incomplete sentence or missing information.] After normalization, the data is mapped to color intensity, such as blue-yellow-red, corresponding to low probability-medium probability-high probability. A heat map is generated based on the color intensity mapping results. Since the higher the cumulative probability weight of a grid cell, the more likely it is to become a water inrush point, regions (there may be one or more regions) with cumulative probability weights exceeding a preset weight threshold can be extracted and marked as candidate water inrush point locations.

[0106] The above steps transform the water hazard source tracing results (probability field) into a spatial distribution map of water inrush risk that can be intuitively interpreted and quantitatively assessed. In this embodiment, the probability thermal distribution map can indicate the location of one or more candidate water inrush points, thereby alleviating the problem of poor adaptability of traditional methods to multi-source water inrush scenarios and providing technical support for rapid identification of mine disaster water sources.

[0107] Step S4: Using candidate water inrush points as input, the final mine water inrush probability distribution map is obtained through forward hydrodynamic simulation and probability update.

[0108] Further, step S4 includes the following sub-steps: inputting the flow rate values ​​of candidate water inrush points in the probability thermal distribution map as source terms into the forward hydrodynamic model; performing forward hydrodynamic simulation using the Saint-Venant equation combined with the Manning formula to calculate the simulated values ​​of water level and flow velocity at each observation point; comparing the simulated values ​​with the actual observed values ​​to calculate the comprehensive error of the candidate water inrush points, and iteratively updating the probability distribution of each candidate water inrush point according to the magnitude of the comprehensive error to obtain the final mine water inrush probability distribution map.

[0109] In this embodiment, the global flow field of the mine established in step S2 and the flow rate value at the location of the candidate water inrush point area (i.e., candidate water inrush point) extracted in step S3 are used as the input values ​​for the forward hydrodynamic simulation. The Saint-Venant equation combined with the Manning formula is used to perform the forward hydrodynamic simulation and calculate the simulated values ​​of water level and flow velocity at each observation point.

[0110] The forward hydrodynamic simulation uses the Saint-Venant equations for calculation, with parameters including friction gradient. Then it can be calculated using Manning's formula.

[0111] Specifically, the Saint-Venant equations include the continuity equation and the momentum equation, which respectively represent the conservation of mass and momentum of water flow, as expressed below: (12) (13) In the formula, For any point, For time, The cross-sectional area of ​​the tunnel. For the cross-sectional flow rate of the tunnel, For hydraulic head, For the slope of the alley, It is the acceleration due to gravity. This refers to the friction slope.

[0112] Furthermore, according to Manning's formula, the friction slope The calculation formula is as follows: (14) In the formula, This is the Manning coefficient. The radius is the hydraulic radius.

[0113] Since the Saint-Venant equation composed of formulas (12) and (13) is a partial differential equation, the continuous analytical solution is difficult to obtain directly. Therefore, the continuous partial differential equation is transformed into a system of algebraic equations, and the solution is calculated point by point using numerical methods.

[0114] Define the space step size as The time step is The discrete grid points required to solve the partial differential equation are: The time layer is , , The current iteration step number is represented by the grid points and the time layer, forming a discrete node, expressed as ( , If ), then the continuity equation is discretized as: (15) The momentum equation is discretized as follows: (16) In the formula, Represents a node ( , The cross-sectional area of ​​the tunnel at point ) and the cross-sectional flow rate of the tunnel. and hydraulic head The rules for superscripts and subscripts are similar and will not be elaborated here.

[0115] Through forward hydrodynamic simulation, simulated values ​​of water level and flow velocity at each observation point were calculated, and then... and To indicate, among which, For the observation point, The observation period is used to calculate the simulated values ​​of water level and flow velocity. and Compared with the actual values ​​at the observation points and The results are compared, the comprehensive error of the candidate water inrush points is calculated, and the probability distribution of each candidate water inrush point is updated according to the magnitude of the error.

[0116] The formula for calculating the overall error of the candidate inrush points is as follows: (17) In the formula, For the first The comprehensive error corresponding to each candidate inrush point , These are the weights corresponding to water level and flow velocity, respectively. , The first Simulated values ​​of water level and flow velocity at each candidate inrush point. , These are the measured values ​​of water level and flow velocity at the observation points, respectively.

[0117] Furthermore, the probability distribution of each candidate water inrush point is updated according to the magnitude of the error, including: generating a new probability by probability decay based on the cumulative probability weight of the candidate water inrush point location generated in the previous iteration, introducing an error adjustment factor to control the probability decay rate, and performing probability normalization after each decay (i.e., probability adjustment) to obtain the normalized new probability; redrawing the probability heat map based on the normalized new probability, and obtaining new candidate water inrush points based on the new probability heat map, and then repeating the forward simulation and probability update process until the change in probability distribution between two adjacent iterations is less than a preset probability threshold, at which point the iteration stops.

[0118] In the initial probability heatmap (i.e., the probability heatmap generated in the previous iteration), it is assumed that the probability at each candidate water inrush point location is... (referred to as the old probability), then the probability update formula is: (18) In the formula, This is the updated probability (also known as the new probability). This is the error adjustment factor, used to control the sensitivity of the error to probability adjustment. The denominator term is " "Used to control the probability decay rate, error adjustment factor" The larger the value, the faster the probability decays.

[0119] Furthermore, to ensure that the total probability of the probability heatmap is 1, a probability normalization operation is performed, calculated as follows: (19) In the formula, This is the new probability after normalization.

[0120] according to Redraw the probabilistic heat map, re-input the locations of one or more candidate inrush points indicated in the newly drawn probabilistic heat map into the Saint-Venant equation for forward hydrodynamic simulation, and then perform probability updates again until the preset convergence condition is met, at which point the iteration is terminated.

[0121] The convergence conditions are as follows: , like (20) In the formula, , For any candidate water inflow point in two adjacent iterations The probability of new and old items. This represents the change in probability distribution between two iterations, i.e., the sum of the absolute values ​​of the differences between the old and new probabilities of all candidate water inrush points. This is a preset probability threshold.

[0122] Finally, based on the results of the iteration, all possible water inrush points are output.

[0123] In this embodiment, the flow rate value of the candidate inrush point is used as the source term input, and a forward hydrodynamic model is used for verification. An error feedback update mechanism is used to realize the closed-loop coupling of reverse inference and forward simulation, which significantly improves the accuracy and robustness of source tracing.

[0124] By setting an error adjustment factor, the probability decay rate can be dynamically adjusted to prevent the results from oscillating due to excessively rapid changes in probability; when the error is large, the update is accelerated, and when the error is small, the adjustment is smoothed, thereby improving the numerical stability and convergence efficiency of the model under complex mining conditions.

[0125] As an example, the following combines Figure 8 The method provided in this embodiment will be described in detail again. For example... Figure 8 As shown, this method can be performed according to the following steps: Step 1, input sensor data; Step 2: Use a sliding window for local analysis and combine it with the Grubbs test to determine if there is any outlier. If there is no outlier, proceed directly to step 3; if there is outlier, remove the outlier and use the Kriging method to interpolate the missing data to obtain the preprocessed data. Step 3: Construct the global flow field of the mine; Step 4: Extract observation point data and calculate the reverse flow vector; Step 5: Backtrack the path and record it; Step 6: Generate a probability heatmap of the candidate areas for water inrush points, i.e., draw a probability heatmap to obtain candidate water inrush points; Step 7: Using the Saint-Venant equations, simulate the spread path in the forward direction with the candidate inrush points as initial conditions; Step 8: Calculate the error based on the simulation results and update the probability; Step 9: Determine if the error is greater than the set value. If so, regenerate the probability heatmap according to the updated probability and return to step 7 to perform the forward simulation again. If the error meets the convergence condition, proceed to step 10. Step 10: Output possible water inrush points; process ends.

[0126] The following example illustrates the intelligent source tracing method for mine disaster water sources provided in this embodiment.

[0127] Taking the 15th minute of a water inrush in a mine as an example, water level and flow velocity values ​​were collected at various locations underground using monitoring modules deployed at different locations. These values ​​were then uploaded to the mine's main control center via the mine's 5G network. The uploaded data underwent standardized preprocessing and was stored in a database. The standardized preprocessed data is shown in Table 1. Table 1 is as follows: Table 1. Data after standardized preprocessing

[0128] Using sliding window local analysis and statistical discrimination methods, outliers and missing values ​​were automatically identified and removed. Kriging interpolation was then used to supplement the removed outlier data to obtain all the observation point data of the mine, as shown in Table 2. Table 2 is as follows: Table 2 Interpolated observation data

[0129] The specific methods for data removal and interpolation are described above and will not be repeated here.

[0130] Based on the cleaned and supplemented complete data, a hydraulic network diagram is constructed in the 3D mine roadway map. Each roadway segment is discretized into a node-edge structure, resulting in the final mine structure diagram (i.e., the hydraulic network diagram). Figure 9 As shown. Figure 9 In the diagram, circles represent nodes, each node has a number, such as A1, A2, B1, etc. Nodes are connected by edges, and each edge stores the corresponding geometric and hydraulic properties.

[0131] The global flow field of the mine was reconstructed based on the observation data, and the flow field data of the unobserved areas were supplemented by Kriging interpolation and boundary conditions. The final global flow field data is shown in Table 3. Table 3. Partial data of the global flow field

[0132] Next, a reverse path backtracking was performed based on the global flow field. The partial backtracking paths of all nodes are shown in Table 4. Table 4 is as follows: Table 4. Partial Reverse Backtracking Paths

[0133] For the set of backtracked path points, the probability weight of the branch reverse path is calculated according to the traffic ratio, and then the cumulative probability weight of each node is calculated. The results are shown in Table 5. Table 5 is as follows: Table 5 Cumulative probability weights of each node

[0134] The iterative process involves the following steps: performing forward hydrodynamic simulation using candidate water inrush points as input; calculating the comprehensive error of the candidate water inrush points; updating the probability; and calculating the change in probability distribution until a preset convergence condition is met, i.e., the iteration stops when the change in probability distribution between two adjacent iterations is less than a preset probability threshold. Based on the final updated probability, a mine water inrush probability distribution map is plotted, as shown below. Figure 10 As shown. From Figure 10 It can be seen that H1 and its surrounding area are areas with a high probability of water inrush, while the peripheral areas may have water inrush, but the probability is low.

[0135] In summary, the method provided in this embodiment combines computer simulation technology with hydraulic model analysis technology. Utilizing existing observational data such as mine roadway water level and flow velocity, it accurately and quickly uses relevant hydraulic formulas and probabilistic models to inversely deduce the possible locations of water inrush points. This forms a highly efficient and stable inversion algorithm for intelligent tracing of mine water inrush points, ensuring the accuracy and stability of water inrush point location. In conclusion, this method employs interdisciplinary modeling techniques in mine safety and intelligent technology to achieve efficient, accurate, and intelligent tracing of mine disaster water sources under limited monitoring data conditions.

[0136] Based on the same inventive concept, this embodiment also provides an intelligent source tracing system for mine disaster water sources. This system is used to execute the intelligent source tracing method for mine disaster water sources provided in any of the above embodiments, including: The flow field construction unit is used to construct a hydraulic network diagram based on the hydrological data collected in real time by sensors at various observation points in the mine roadway, and then reconstruct the global flow field of the mine in the hydraulic network diagram. The backtracking unit is used to backtrack along the direction of the reverse water flow from each observation point based on the global flow field of the mine, generating multiple reverse backtracking paths; The candidate water inflow point generation unit is used to calculate the cumulative probability weight corresponding to each location in the mine based on the probability weight of each reverse backtracking path, and to draw a probability heat map using the cumulative probability weight of each location. The probability heat map is used to indicate the location of one or more candidate water inflow points. The iterative update unit is configured to take candidate water inrush points as input and obtain the final mine water inrush probability distribution map through forward hydrodynamic simulation and probability update.

[0137] The intelligent source tracing system for mine disaster water sources provided in this embodiment can realize the steps and processes of the intelligent source tracing method for mine disaster water sources provided in any of the above embodiments, and achieve the same technical effect, which will not be described in detail here.

[0138] The following reference Figures 3-7 The intelligent source tracing system for mine disaster water sources provided in this embodiment is illustrated with an example.

[0139] In one example, such as Figures 3-7As shown, the intelligent source tracing system for mine disaster water sources can include four parts: a monitoring module, a data processing module, an algorithm module, and a verification module. The monitoring module consists of multiple deployed water level, flow velocity, and pressure sensors that collect real-time hydrological data from the mine, including water level and flow velocity in various roadways. Different types of sensors form corresponding sensor networks (such as water level sensor networks and flow velocity sensor networks). The collected data is uploaded to the mine's main control center via a 5G multi-route transmission link, where a data acquisition terminal is located. The data processing module cleans, preprocesses, interpolates, and performs consistency checks on the data collected by the monitoring module. This module first performs data standardization and structural transformation, then performs outlier detection and removal (achieved through sliding window analysis and Grubbs' test), and finally uses Kriging space interpolation to interpolate the data and perform consistency checks. The algorithm module, based on an improved reverse flow field reconstruction method, a probability weight allocation model, and time series analysis technology, dynamically reverse-tracks water inflow points and generates candidate regions and probability distributions. This is the core module, which includes three sub-modules: flow field reconstruction, reverse tracing, and potential water inflow point generation. Flow field reconstruction encompasses steps such as constructing the mine hydraulic network diagram and global flow field reconstruction and interpolation. Reverse tracing includes steps such as reverse flow vector calculation, path backtracking, and node analysis. Potential water inflow point generation involves a probability weight allocation model (i.e., calculating the probability weight of branch reverse paths according to flow ratio) and the generation of probability heatmaps for candidate water inflow point regions. The verification module uses forward hydrodynamic simulation to verify the back-calculated results, thereby correcting errors and forming a closed-loop optimization. This module specifically includes two parts: forward hydrodynamic simulation and error calculation. The forward hydrodynamic simulation calculates simulated water level and flow velocity values ​​by numerically solving the Saint-Venant equations. The error calculation involves iterative calculations based on error calculation and probability updates, followed by probability normalization and convergence checks to generate the final possible water inrush points. This system is applicable to single-source and multi-source water inrush scenarios, providing scientific basis and decision support for mine water control and emergency rescue.

[0140] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for intelligent tracing of water sources in mine disasters, characterized in that, include: Based on the real-time hydrological data collected by sensors at various observation points in the mine roadway, a hydraulic network diagram is constructed, and then the global flow field of the mine is reconstructed in the hydraulic network diagram. Based on the global flow field of the mine, starting from each observation point, the path is traced back along the direction of the reverse flow to generate multiple reverse tracing paths; Based on the probability weights of each reverse backtracking path, the cumulative probability weights corresponding to each location in the mine are calculated, and a probability heat map is drawn using the cumulative probability weights of each location. The probability heat map is used to indicate the location of one or more candidate water inrush points. Using candidate water inrush points as input, the final mine water inrush probability distribution map is obtained through forward hydrodynamic simulation and probability update.

2. The method according to claim 1, characterized in that, The construction of the hydraulic network diagram includes: The mine roadway system is discretized into a topological structure consisting of nodes and edges, where nodes represent roadway intersections and key locations, and edges represent roadway segments connecting nodes. Assign geometric properties and hydraulic parameters to each edge to obtain a hydraulic network diagram.

3. The method according to claim 1, characterized in that, Reconstructing the global flow field of the mine in the hydraulic network diagram includes: By using the hydrological data collected in real time by sensors at various observation points and combining it with the continuity equation, the flow rate of each tunnel segment in the hydraulic network diagram is determined. By setting boundary conditions, flow direction data for unobserved areas are generated using interpolation methods to obtain a continuous flow field covering the entire mine.

4. The method according to any one of claims 1 to 3, characterized in that, Based on the global flow field of the mine, starting from each observation point, path backtracking is performed along the direction of reverse water flow to generate multiple reverse backtracking paths, including: Calculate the unit reverse flow vector at each observation point; Based on the global flow field of the mine, starting from each observation point, the reverse backtracking path is generated by iteratively advancing along the direction of the unit reverse flow vector with a preset step size. The expression for the unit reverse flow vector is as follows: ， In the formula, Represents the unit reverse flow vector. Indicates the position of any observation point The velocity vector at that point express The model.

5. The method according to claim 4, characterized in that, The generation of the reverse backtracking path further includes: When the reverse backtracking is advanced to the intersection of the roadways, the velocity vector and flow rate of each connected branch of the intersection are obtained; The inflow branch set is selected based on the dot product of the velocity vectors of each connected branch and the unit direction vectors pointing outward from the intersection node. For the set of inflow branches, the probability weight of the reverse path corresponding to each inflow branch is calculated according to the flow ratio, and then the probability weight of multiple reverse backtracking paths corresponding to the intersection node is generated.

6. The method according to claim 5, characterized in that, Based on the probability weights of each reverse backtracking path, the cumulative probability weights corresponding to each location in the mine are calculated, and a probability heat map is drawn using the cumulative probability weights of each location, including: The frequency of each location in the mine roadway being traversed by the reverse backtracking path is counted, and the probability weight of the corresponding reverse backtracking path is accumulated and superimposed on the grid cell where each location is located to obtain the accumulated probability weight. A probability heat map is generated based on the cumulative probability weight, and the regions in the probability heat map with a cumulative probability weight greater than a preset weight threshold are taken as the locations of candidate water inrush points. The grid cells are obtained by discretizing the space of the mine roadway.

7. The method according to any one of claims 1 to 6, characterized in that, Using candidate water inrush points as input, the final mine water inrush probability distribution map is obtained through forward hydrodynamic simulation and probability update, including: The flow rate values ​​of candidate inrush points in the probabilistic heat map are used as source terms input into the positive hydrodynamic model; A forward hydrodynamic simulation was performed using the Saint-Venant equation combined with the Manning formula to calculate the simulated values ​​of water level and flow velocity at each observation point. The simulated values ​​are compared with the actual observed values ​​to calculate the comprehensive error of the candidate water inrush points. The probability distribution of each candidate water inrush point is then iteratively updated based on the magnitude of the comprehensive error to obtain the final mine water inrush probability distribution map.

8. The method according to claim 7, characterized in that, The step of updating the probability distribution of each candidate water inrush point based on the magnitude of the error includes: Based on the probability of the candidate water inrush point location generated in the previous iteration, a new probability is generated through a probability decay mechanism. At the same time, an error adjustment factor is introduced to control the probability decay rate, and probability normalization is performed after each decay to obtain the normalized new probability. The probability heat map is redrawn based on the new normalized probability, and new candidate water inrush points are obtained based on the new probability heat map. Then, the forward simulation and probability update process is repeated until the change in probability distribution between two adjacent iterations is less than the preset probability threshold, at which point the iteration stops.

9. The method according to claim 1, characterized in that, The sensors are deployed in the goaf, fault fracture zone, roadway intersection and drainage pipeline node of the mine roadway, and are locally densified in key areas. The hydrological data is uploaded to the main control center via a wireless network and, after data cleaning and missing data interpolation, is used to reconstruct the global flow field of the mine.

10. A mine disaster water source intelligent tracing system, the system being used to perform the method as described in any one of claims 1 to 9, comprising: The flow field construction unit is configured to construct a hydraulic network diagram based on the hydrological data collected in real time by sensors at various observation points in the mine roadway, and then reconstruct the global flow field of the mine in the hydraulic network diagram. The backtracking unit is used to backtrack along the direction of the reverse water flow from each observation point based on the global flow field of the mine, generating multiple reverse backtracking paths; The candidate water inflow point generation unit is configured to calculate the cumulative probability weight corresponding to each location in the mine based on the probability weight of each reverse backtracking path, and draw a probability heat map using the cumulative probability weight of each location. The probability heat map is used to indicate the location of one or more candidate water inflow points. The iterative update unit is configured to take candidate water inrush points as input and obtain the final mine water inrush probability distribution map through forward hydrodynamic simulation and probability update.

Images