An intelligent mine management system and method based on the internet of things

Abstract

This invention relates to the field of mine pollution management technology, specifically to an intelligent mine management system and method based on the Internet of Things (IoT). The system includes: a mine pollution data acquisition unit that uses mobile IoT sensors to collect real-time groundwater flow path data and pollutant concentration data; an event-driven pollution identification unit that identifies and analyzes dynamic evolution events of underground pollution in the mine based on the groundwater flow path data and pollutant concentration data; a pollution path evolution identification unit that constructs a groundwater pollution propagation evolution map covering the entire path, identifying and tracing upstream pollution source origins, downstream terminal pollution areas, and midstream potential pollution sources; and a pollution situation visualization unit that constructs a mine pollution risk level zoning map and displays it visually on terminal devices. This invention enables the identification and monitoring of post-rain pollution surge events and the traceability and management of potential pollution conditions in the upstream, midstream, and downstream of the entire mine geological structure.

Description

Technical Field

[0001] This invention relates to the field of mine pollution management technology, specifically to an intelligent mine management system and method based on the Internet of Things. Background Technology

[0002] Mining activities, especially open-pit mining, slag heaps, and ore leaching, can easily cause the spread of groundwater pollution around mining areas. Pollutants can spread rapidly to water sources through karst channels such as fissures and caves. The main types of pollution include heavy metals, acidic water, chemicals, and radioactive substances. These pollutants persist in groundwater and spread through the groundwater system, which may seriously affect regional ecology and human health. The purpose of tracking and managing pollution sources is to understand the patterns of pollution transmission and to assess the extent and impact of pollution spread, thereby providing a scientific basis for mine environmental protection, water resource management, and groundwater pollution prevention and control.

[0003] Current technologies for tracking and managing groundwater pollution generally treat rainwater as a dilution factor. After rainfall or irrigation events, some pollutants that have been trapped in the soil or rock crevices for a long time are released instantaneously by scavenging, resulting in a delayed increase in pollution concentration in groundwater. Traditional technologies may misinterpret rainfall as a dilution effect on pollution, leading to the risk of post-rain pollution surges. Actual observation data shows that pollution responses typically have a time delay of 30 minutes to 12 hours, meaning that pollution concentrations surge in the lag period after rainfall events, forming so-called post-rain pollution surge events. Furthermore, in complex hydrogeological structures, although the surface pollution concentration at some monitoring points may be within the normal range, there may be a significant discrepancy with the concentration trends of upstream, downstream, or adjacent monitoring points, indicating potential pollution sources. Therefore, it is necessary not only to monitor and manage post-rain pollution surge events in real time but also to trace the potential pollution situation upstream, midstream, and downstream of the entire mine geological structure. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent mine management system and method based on the Internet of Things to solve the problems mentioned in the background art, such as pollution surge events after rain in mines and potential pollution sources under complex hydrogeological structures.

[0005] To achieve the above objectives, the present invention aims to provide an intelligent mine management system based on the Internet of Things, comprising:

[0006] The mine pollution data acquisition unit uses mobile IoT sensing devices to collect real-time data on groundwater flow paths and pollutant concentrations.

[0007] The event-driven pollution identification unit, based on groundwater flow path data and pollutant concentration data, sequentially employs change point detection algorithm, post-rain lag pollution identification algorithm, and spatial comparison anomaly identification algorithm to identify and analyze the dynamic evolution events of underground pollution in the mine, and outputs the evolution results of various underground pollution dynamic evolution events.

[0008] The underground pollution dynamic evolution events include pollution pathway convergence and abrupt change events, post-rain delayed pollution release events, and spatially controlled pollution anomaly events;

[0009] The pollution path evolution identification unit constructs a groundwater pollution propagation evolution map covering the entire path based on the evolution results of local underground pollution dynamic evolution events, and identifies the upstream pollution source origin, downstream terminal pollution area and midstream potential pollution source.

[0010] The pollution situation visualization unit receives the underground pollution dynamic evolution event evolution results from the event-driven pollution identification unit and the upstream pollution source starting point, downstream terminal pollution area and midstream potential pollution source from the pollution path evolution identification unit. It constructs a mine pollution risk level zoning map and displays it visually on the terminal device.

[0011] Preferably, the mobile IoT sensing device is used to move along the direction of groundwater flow in the groundwater aquifer area, and collects groundwater flow path data and pollutant concentration data at its location in real time during the movement.

[0012] The flow path data includes the location coordinates of the mobile IoT sensing device, the rate of change of flow direction, and the water flow velocity information; the pollutant concentration data includes the concentration of heavy metal ions, pH value, conductivity, and redox potential.

[0013] Preferably, the event-driven pollution identification unit includes a path intersection mutation identification module; the path intersection mutation identification module is used to analyze the intersection position of water flow paths based on flow path data and pollutant concentration data, identify the pollution concentration change at the water flow path intersection position using a change point detection algorithm, and record and output the event time series, spatial location series, and pollution intensity change of the pollution path intersection mutation event.

[0014] Preferably, the event-driven pollution identification unit includes a rainfall-delayed pollution identification module; the rainfall-delayed pollution identification module identifies rainfall-delayed pollution release events in the mining area based on historical rainfall data and pollutant concentration data, using a rainfall-delayed pollution identification algorithm.

[0015] The post-rain lag pollution identification algorithm is constructed by fusing a variable-order fractional-order kernel response model with a sparse compressed sensing reconstruction method. Based on historical rainfall data and pollutant concentration data, the algorithm calculates the lag pollution release intensity time-series curve and hysteresis delay length of the post-rain heavy metal pollution release intensity. The lag pollution release intensity time-series curve represents the dynamic evolution trajectory of heavy metal pollutant concentration in groundwater over time. The hysteresis delay length of the post-rain heavy metal pollution release intensity is the time interval between increases in heavy metal pollutant concentration in groundwater after rainfall.

[0016] Preferably, the post-rain lag pollution identification algorithm calculates the lag pollution release intensity time-series curve and hysteresis delay length of the post-rain heavy metal pollution release intensity, and the specific method is as follows:

[0017] S2.2.1. Based on the time series of historical rainfall data and pollutant concentration data, establish a dynamic input-output aligned dataset;

[0018] S2.2.2 Based on the dynamic input-output aligned dataset, construct a variable-order fractional-order kernel response function for the time series of historical rainfall data;

[0019] S2.2.3. Using the sparse compressed sensing reconstruction method, the time series of pollutant concentration data is sparsely reconstructed, the main excitation components during pollutant release are extracted, and the pollution concentration abrupt change segments are identified.

[0020] S2.2.4 Calculate the dynamic curve of pollution concentration change over time, and extract the time length from the start of rain to the peak pollution concentration in combination with the pollution concentration abrupt change segment, as the hysteresis delay length.

[0021] S2.2.5. Combining the variable-order fractional-order kernel response function and the main excitation components, the change of pollution response intensity over time is constructed as a time-series function, generating a time-series curve of hysteresis pollution release intensity.

[0022] S2.2.6. The final output includes the pollution concentration abrupt change range, hysteresis delay length, and time series curve of the delayed pollution release intensity of the post-rain delayed pollution release event.

[0023] Preferably, the event-driven pollution identification unit includes a spatial contrast anomaly identification module; the spatial contrast anomaly identification module is used to identify local pollution concentration anomaly areas in the groundwater flow path of the mine, construct a dynamic pollution map of the water flow path based on the groundwater flow path data and pollutant concentration data, and use a spatial contrast anomaly identification algorithm to filter out the set of path spatial location nodes with abnormal concentration changes in the dynamic pollution map of the water flow path, and the three-dimensional area enclosed by the set of adjacent path spatial location nodes with abnormal concentration changes is regarded as the local pollution concentration anomaly area;

[0024] The spatial contrast anomaly identification algorithm is based on the fusion of graph convolutional neural network and local outlier analysis method. Specifically, it includes: using graph convolutional neural network to extract the concentration change characteristics of each spatial location node in the dynamic map of water flow path pollution; using local outlier analysis method to calculate the concentration change score of each spatial location in the path, and screening out the set of spatial location nodes of the path with abnormal concentration changes.

[0025] Preferably, the method for constructing the groundwater pollution propagation evolution map covering the entire path is as follows:

[0026] The high-slope intervals of the main excitation components are extracted as high-intensity time windows when pollutants are released. Combined with flow path data, the sources of pollution release on the flow path are identified as the starting nodes of the groundwater pollution propagation evolution map.

[0027] Based on the mobile path data of mobile IoT sensing devices, the terminal range of pollution diffusion impact is calculated, which serves as the end node of the groundwater pollution propagation evolution map.

[0028] Based on the hysteresis delay length and water flow velocity information, the spatial propagation distance of pollutants from the surface to the underground response is calculated, and the main path edge from the starting node to the ending node is constructed.

[0029] All nodes in the set of spatial location nodes of abnormal concentration change paths are used as intermediate nodes in the groundwater pollution propagation evolution map.

[0030] Based on the groundwater flow path data and spatial relationships, the spatial location nodes of the abnormal concentration change path are unidirectionally connected to the main path edge to construct the branch path edge;

[0031] Based on the concentration change score at each spatial location along the path, the concentration gradient weights are assigned to each intermediate node on the corresponding branch path edge.

[0032] The final result is a complete map of the evolution of groundwater pollution propagation.

[0033] Preferably, the groundwater pollution propagation evolution map is used to identify the upstream pollution source origin, the downstream terminal pollution area, and the midstream potential pollution source, as follows:

[0034] S3.1. Based on the groundwater pollution propagation evolution diagram, find the node where the first increase in pollutant concentration occurs in the main path from the starting node towards the water flow direction. This node is the starting point of the upstream pollution source.

[0035] S3.2 The area enclosed by the end node of the groundwater pollution propagation evolution diagram and the neighboring nodes of that node is the downstream terminal pollution area;

[0036] S3.3. Based on the groundwater pollution propagation evolution diagram, find a set of candidate nodes in the branch path where the concentration change score at the spatial location of the path is greater than the set score threshold. All candidate nodes in this set are potential midstream pollution sources.

[0037] Based on the time series curves of the lagged pollution release intensity of all candidate nodes, the slope value of each time window is extracted. If the slope values ​​of all time windows of the lagged pollution release intensity time series curve are less than the response slope threshold, the candidate node to which the lagged pollution release intensity time series curve belongs is deleted.

[0038] Calculate the ratio of the hysteresis delay length from the remaining candidate node to the node on the main path to the path distance. If the ratio is within the set groundwater flow velocity range, the candidate node is located on a reasonable pollution transmission path and is deleted.

[0039] The remaining candidate nodes were all potential sources of pollution in the midstream.

[0040] Preferably, the mine pollution risk level zoning map is based on groundwater pollution propagation evolution map data, upstream pollution source origin, downstream terminal pollution area, midstream potential pollution source and pollutant concentration data to divide the pollution area and visualize the management of mine area pollution.

[0041] On the other hand, the present invention provides an intelligent mine management method based on the Internet of Things (IoT), used in the aforementioned intelligent mine management system based on the IoT, comprising the following steps:

[0042] S10.1 Use mobile IoT sensing devices to collect groundwater flow path data and pollutant concentration data in real time;

[0043] S10.2 Based on groundwater flow path data and pollutant concentration data, the change point detection algorithm, the post-rain lag pollution identification algorithm, and the spatial comparison anomaly identification algorithm are used in sequence to identify and analyze the dynamic evolution events of underground pollution in the mine, and output the evolution results of the dynamic evolution events of underground pollution in each region.

[0044] S10.3 Based on the evolution results of groundwater pollution dynamic evolution events in various regions, construct a groundwater pollution propagation evolution map covering the entire path, and identify the upstream pollution source origin, downstream terminal pollution area and midstream potential pollution source;

[0045] S10.4 Based on the dynamic evolution results of underground pollution events, the starting point of upstream pollution sources, the downstream terminal pollution areas and potential pollution sources in the middle reaches, construct a mine pollution risk level zoning map and visualize it on the terminal equipment.

[0046] Compared with the prior art, the above-mentioned technical solution of the present invention has the following beneficial technical effects:

[0047] 1. In this invention, the post-rain delayed pollution identification algorithm can accurately identify post-rain delayed release events caused by latent pollution sources, identify and monitor post-rain pollution surge events, and improve the timeliness and accuracy of pollution channel identification and monitoring management.

[0048] 2. In this invention, by constructing a dynamic map of water flow path pollution and combining it with a spatial comparison anomaly identification algorithm, the intelligent identification of potentially abnormal pollution areas with local concentrations in the geological spatial distribution of mines is realized, thereby improving the resolution and targeting of pollution risk level zoning. Attached Figure Description

[0049] Figure 1 This is a schematic diagram of one embodiment of the present invention;

[0050] Attached reference numerals: 1. Mine pollution data acquisition unit; 2. Event-driven pollution identification unit; 21. Path intersection and abrupt change identification module; 22. Rainfall lag pollution identification module; 23. Spatial comparison anomaly identification module; 3. Pollution path evolution identification unit; 4. Pollution situation visualization unit. Detailed Implementation

[0051] Example 1, as Figure 1 As shown, an intelligent mine management system based on the Internet of Things is provided, including:

[0052] Mine pollution data acquisition unit 1 uses a mobile Internet of Things (IoT) sensor to collect groundwater flow path data and pollutant concentration data in real time.

[0053] The mobile IoT sensing device is used to move along the direction of groundwater flow in the groundwater aquifer area and collect data on the flow path of groundwater and pollutant concentration in real time during the movement.

[0054] The flow path data includes the location coordinates of the mobile IoT sensing device, the rate of change of flow direction, and the water flow velocity information; the pollutant concentration data includes the concentration of heavy metal ions, pH value, conductivity, and redox potential.

[0055] In this embodiment, the mobile IoT sensing device includes a positioning module, a flow direction sensing module, a miniature multi-parameter water quality sensor, and a communication module. When the device drifts with the direction of groundwater flow or is navigated by scheduling control, it continuously collects location information, changes in water flow direction, and flow velocity to form a flow trajectory curve. It also performs multi-parameter water quality measurements on the water body where the mobile IoT sensing device is currently located, forming concentration sampling point cloud data. The collected data is uploaded to the edge analysis node and cloud platform via a low-power wireless module.

[0056] Event-driven pollution identification unit 2, based on groundwater flow path data and pollutant concentration data, sequentially employs change point detection algorithm, post-rain lag pollution identification algorithm and spatial comparison anomaly identification algorithm to identify and analyze the dynamic evolution events of underground pollution in the mine, and outputs the evolution results of the dynamic evolution events of underground pollution in various places.

[0057] The underground pollution dynamic evolution events include pollution pathway convergence and abrupt change events, post-rain delayed pollution release events, and spatially controlled pollution anomaly events;

[0058] The event-driven pollution identification unit 2 includes a path intersection mutation identification module 21. The path intersection mutation identification module 21 is used to analyze the intersection position of water flow paths based on flow path data and pollutant concentration data, identify the pollution concentration change at the water flow path intersection position using a change point detection algorithm, and record and output the event time series, spatial location series, and pollution intensity change of the pollution path intersection mutation event.

[0059] In this embodiment, a pollution path convergence abrupt event refers to the sudden change in pollutant concentration detected by mobile IoT sensors on multiple paths when two or more groundwater flow paths converge, either at the convergence point or during continuous monitoring before and after the convergence. Pollution path convergence abrupt event is used to identify potential pollution source convergence areas or interactive pollution release areas in groundwater, and to assist in the analysis of the mutual migration and enhanced coupling behavior of pollutants between different paths.

[0060] In this embodiment, the event time series of pollution path convergence abrupt events is the set of time series of pollution concentration abrupt changes that occur at the confluence of two water flow paths within a continuous monitoring time window; the length of the time window is manually set to one minute, and the length of one time window is also the length of the monitoring cycle; the spatial location sequence of pollution path convergence abrupt events refers to the three-dimensional coordinate representation of the path convergence point corresponding to each time point in the event time series in geographic space; the pollution intensity change of pollution path convergence abrupt events is the set of changes in pollution concentration at the convergence point compared to the previous moment.

[0061] In this embodiment, the change point detection algorithm is a Bayesian online change point detection algorithm. This algorithm is a time-series data mutation detection algorithm based on the Bayesian inference principle. Furthermore, it is an online learning algorithm that can instantly determine whether a sudden change in pollution concentration occurs at the intersection of pollution paths while the monitoring data stream is continuously updated, adapting to the rapid response characteristics of groundwater flow in mining areas. Traditional change point detection algorithms often rely on manually set concentration mutation thresholds, while the Bayesian online change point detection algorithm uses Bayesian posterior arithmetic to dynamically adjust the model confidence, making mutation detection more adaptable and robust. The event time series of the pollution path intersection mutation event is a direct mapping of the structured results of the Bayesian online change point detection algorithm, allowing subsequent modules to directly rely on its results without redundant transformations or noise filtering.

[0062] In this embodiment, the event-driven pollution identification unit 2 includes a rainfall-delayed pollution identification module 22; the rainfall-delayed pollution identification module 22 identifies rainfall-delayed pollution release events in the mining area based on historical rainfall data and pollutant concentration data and using a rainfall-delayed pollution identification algorithm.

[0063] The post-rain lag pollution identification algorithm is constructed by fusing a variable-order fractional-order kernel response model with a sparse compressed sensing reconstruction method. Based on historical rainfall data and pollutant concentration data, the algorithm calculates the lag pollution release intensity time-series curve and hysteresis delay length of the post-rain heavy metal pollution release intensity. The lag pollution release intensity time-series curve represents the dynamic evolution trajectory of heavy metal pollutant concentration in groundwater over time. The hysteresis delay length of the post-rain heavy metal pollution release intensity is the time interval between increases in heavy metal pollutant concentration in groundwater after rainfall.

[0064] In this embodiment, the time-series curve of hysteresis pollution release intensity is modeled using sampled time-series data to characterize the concentration response process of heavy metal ions during leaching, infiltration, and migration, reflecting the rate change characteristics and nonlinear release mechanism of the pollution release process; the hysteresis delay length represents the time lag between surface input and pollution response, reflecting the combined effect of factors such as pollution source location, soil adsorption capacity, and groundwater response rate, and is used to assess the response delay of the pollution release process.

[0065] In this embodiment, the post-rain delayed pollution release event refers to the situation where, after a significant rainfall or surface infiltration event, the concentration of pollutants in groundwater does not rise immediately, but rather exhibits an abnormally enhanced pollution release behavior after a certain time delay. The post-rain delayed pollution release event reflects the delayed response relationship between the contaminated layer in the mining area and the groundwater system based on the time lag of the pollution response to rainfall and the dynamic characteristics of the cumulative release of pollution.

[0066] In this embodiment, the variable-order fractional kernel response model is a dynamic system modeling method based on fractional calculus theory, used to characterize the nonlinear, long-memory, and hysteresis coupling relationship between surface rainfall input and groundwater pollution concentration response; the sparse compressed sensing reconstruction method is a reconstruction technique based on sparse signal representation and linear observation matrix, used to recover the complete pollution concentration change trajectory from sparse or incomplete observation data. The sparse compressed sensing reconstruction method can prevent incomplete data collected by mobile IoT sensing devices due to geographical environment transmission, and is used to recover the complete pollution concentration change trajectory from sparse or incomplete observation data.

[0067] In this embodiment, historical rainfall data is collected in the following ways: by meteorological sensing nodes deployed in and around the mining area, such as rain gauges and remote raindrop monitoring devices; simultaneously, it is integrated with historical meteorological databases from regional water conservancy departments or third-party platforms; all historical rainfall data is uploaded to the pollution identification system database via IoT transmission protocols for subsequent modeling and analysis.

[0068] In this embodiment, the post-rain lag pollution identification algorithm calculates the lag pollution release intensity time-series curve and hysteresis delay length of the post-rain heavy metal pollution release intensity. The specific method is as follows:

[0069] S2.2.1. Based on the time series of historical rainfall data and pollutant concentration data, establish a dynamic input-output aligned dataset;

[0070] S2.2.2 Based on the dynamic input-output aligned dataset, construct a variable-order fractional-order kernel response function for the time series of historical rainfall data;

[0071] S2.2.3. Using the sparse compressed sensing reconstruction method, the time series of pollutant concentration data is sparsely reconstructed, the main excitation components during pollutant release are extracted, and the pollution concentration abrupt change segments are identified.

[0072] S2.2.4 Calculate the dynamic curve of pollution concentration change over time, and extract the time length from the start of rain to the peak pollution concentration in combination with the pollution concentration abrupt change segment, as the hysteresis delay length.

[0073] S2.2.5. Combining the variable-order fractional-order kernel response function and the main excitation components, the change of pollution response intensity over time is constructed as a time-series function, generating a time-series curve of hysteresis pollution release intensity.

[0074] S2.2.6. The final output includes the pollution concentration abrupt change range, hysteresis delay length, and time series curve of the delayed pollution release intensity of the post-rain delayed pollution release event.

[0075] In this embodiment, the variable-order fractional-order kernel response function is used to simulate the memory of rainwater input and the hysteresis characteristics of pollutant release in the soil layer, reflecting the nonlinear process of slow release, delay, and concentration amplification of pollutant migration. The main excitation component during pollutant release refers to the significant response component that plays a key driving role in the change of pollutant concentration during the pollution response process. The main excitation component represents the concentrated release or overt migration of pollutants within a specific time period and is a key influencing factor in the formation of pollution abrupt phenomena, exhibiting significant dynamic characteristics such as high amplitude, high slope, or suddenness. The pollution concentration abrupt change segment refers to the continuous time period during which the concentration of pollutants in groundwater changes drastically in a short period of time, usually manifested as a rapid increase or decrease in pollution concentration. The pollution concentration abrupt change segment is a direct reflection of pollution release, migration, or convergence, marking a key turning point in time for the pollution event. The hysteresis delay length refers to the time interval between the occurrence of rainfall or irrigation events and the significant increase in the concentration of pollutants in groundwater. The hysteresis delay length reflects the delayed characteristics of pollutant conduction from the surface or soil layer to the groundwater body. The time series curve of hysteresis pollution release intensity refers to the dynamic curve describing the evolution of pollutant release intensity in groundwater after rain over time.

[0076] In this embodiment, the event-driven pollution identification unit 2 includes a spatial contrast anomaly identification module 23; the spatial contrast anomaly identification module 23 is used to identify local pollution concentration anomaly areas in the groundwater flow path of the mine, construct a dynamic map of water flow path pollution based on groundwater flow path data and pollutant concentration data, and use a spatial contrast anomaly identification algorithm to filter out the set of path spatial location nodes with abnormal concentration changes in the dynamic map of water flow path pollution, and the three-dimensional area enclosed by the set of adjacent path spatial location nodes with abnormal concentration changes is regarded as the local pollution concentration anomaly area.

[0077] The spatial contrast anomaly identification algorithm is based on the fusion of graph convolutional neural network and local outlier analysis method. Specifically, it includes: using graph convolutional neural network to extract the concentration change characteristics of each spatial location node in the dynamic map of water flow path pollution; using local outlier analysis method to calculate the concentration change score of each spatial location in the path, and screening out the set of spatial location nodes of the path with abnormal concentration changes.

[0078] In this embodiment, the spatial control pollution anomaly event refers to a significant difference in the pollution concentration measured by the sensing device in adjacent sampling areas of groundwater or areas with similar structures, due to non-obvious factors such as discharge through hidden pipes, interference from barrier structures, or covert release from local pollution sources. In particular, it manifests as a reversal of the pollution concentration gradient direction, abnormal local peak values, or changes in the frequency of abnormal disturbances. The spatial control pollution anomaly event is used to identify spatial pollution inconsistencies caused by structural barriers, hidden pollution sources, or asymmetric diffusion mechanisms.

[0079] In this embodiment, the graph convolutional neural network is a deep learning model based on graph structure data. It learns the local dependency features between nodes by performing convolution operations on the adjacency structure of nodes in the graph. The graph convolutional neural network is used to extract the concentration change features of each spatial node in the water pollution dynamic map constructed from the groundwater flow path. The local outlier analysis method is an unsupervised anomaly detection algorithm. It identifies the degree of outlier in high-dimensional space by comparing the local density difference between a point and its neighboring points. It is used to score the pollution concentration change feature vector extracted from each spatial node in the pollution path map and identify those nodes with drastic concentration deviations compared with their spatial neighborhood, thereby marking them as possible local pollution anomaly areas.

[0080] Pollution Path Evolution Identification Unit 3: Based on the evolution results of various underground pollution dynamic evolution events, Pollution Path Evolution Identification Unit 3 constructs a groundwater pollution propagation evolution map covering the entire path, and identifies the upstream pollution source origin, downstream terminal pollution area and midstream potential pollution source.

[0081] The method for constructing the groundwater pollution propagation evolution map with full path coverage is as follows:

[0082] The high-slope intervals of the main excitation components are extracted as high-intensity time windows when pollutants are released. Combined with flow path data, the sources of pollution release on the flow path are identified as the starting nodes of the groundwater pollution propagation evolution map.

[0083] Based on the mobile path data of mobile IoT sensing devices, the terminal range of pollution diffusion impact is calculated, which serves as the end node of the groundwater pollution propagation evolution map.

[0084] Based on the hysteresis delay length and water flow velocity information, the spatial propagation distance of pollutants from the surface to the underground response is calculated, and the main path edge from the starting node to the ending node is constructed.

[0085] All nodes in the set of spatial location nodes of abnormal concentration change paths are used as intermediate nodes in the groundwater pollution propagation evolution map.

[0086] Based on the groundwater flow path data and spatial relationships, the spatial location nodes of the abnormal concentration change path are unidirectionally connected to the main path edge to construct the branch path edge;

[0087] Based on the concentration change score at each spatial location along the path, the concentration gradient weights are assigned to each intermediate node on the corresponding branch path edge.

[0088] The final result is a complete map of the evolution of groundwater pollution propagation.

[0089] In this embodiment, the terminal range of pollution diffusion impact is calculated based on the mobile IoT sensing device's movement path data, specifically as follows: A pollutant concentration safety limit threshold is set, and pollutant concentration data is collected in real time along the device's movement path. If all consecutive sampling points after a certain spatial location satisfy the condition that the pollutant concentration in the next time window of that location is less than the pollutant concentration safety limit threshold, then that location is a candidate point for the terminal boundary of the pollution response. The slope of the concentration change before the candidate point for the terminal boundary is calculated. If the concentration change tends to be stable (the slope tends to be close to zero) in the most recent path segment, then that location is determined to be the termination of pollution propagation. The terminal points on multiple movement paths are spatially registered to form the terminal range area of ​​pollution diffusion impact. This area is considered as a continuous area where the concentration is below the threshold and the change is stable.

[0090] In this embodiment, the end node in the pollution propagation diagram refers to the farthest spatial boundary that the pollutant's influence can reach, used to define the downstream terminal response area of ​​the pollution evolution. The purpose of using the terminal range of the pollution diffusion influence as the end node of the entire groundwater pollution propagation evolution diagram is to close the pollution path diagram structure, ensure path traceability, and delineate the boundary of the groundwater pollution propagation evolution diagram. Furthermore, the downstream terminal response area of ​​the end node is a range area, and the end node can be a set of boundary nodes, where the range connected by all nodes in the set can be the end node.

[0091] In this embodiment, the groundwater pollution propagation evolution map is used to identify the upstream pollution source origin, the downstream terminal pollution area, and the midstream potential pollution source, as detailed below:

[0092] S3.1. Based on the groundwater pollution propagation evolution diagram, find the node where the first increase in pollutant concentration occurs in the main path from the starting node towards the water flow direction. This node is the starting point of the upstream pollution source.

[0093] S3.2 The area enclosed by the end node of the groundwater pollution propagation evolution diagram and the neighboring nodes of that node is the downstream terminal pollution area;

[0094] S3.3. Based on the groundwater pollution propagation evolution diagram, find a set of candidate nodes in the branch path where the concentration change score at the spatial location of the path is greater than the set score threshold. All candidate nodes in this set are potential midstream pollution sources.

[0095] Based on the time series curves of the lagged pollution release intensity of all candidate nodes, the slope value of each time window is extracted. If the slope values ​​of all time windows of the lagged pollution release intensity time series curve are less than the response slope threshold, the candidate node to which the lagged pollution release intensity time series curve belongs is deleted.

[0096] Calculate the ratio of the hysteresis delay length from the remaining candidate node to the node on the main path to the path distance. If the ratio is within the set groundwater flow velocity range, the candidate node is located on a reasonable pollution transmission path and is deleted.

[0097] The remaining candidate nodes were all potential sources of pollution in the midstream.

[0098] Pollution situation visualization unit 4 is used to receive the underground pollution dynamic evolution event evolution results from event-driven pollution identification unit 2 and the upstream pollution source starting point, downstream terminal pollution area and midstream potential pollution source from pollution path evolution identification unit 3, construct a mine pollution risk level zoning map, and visualize it on the terminal device.

[0099] In this embodiment, the mine pollution risk level zoning map is based on groundwater pollution propagation evolution map data, upstream pollution source origin, downstream terminal pollution area, midstream potential pollution source, and pollutant concentration data to divide the pollution area and visualize the management of mine area pollution.

[0100] Example 2: This invention proposes an intelligent mine management method based on the Internet of Things (IoT), used in the intelligent mine management system based on the IoT described in Example 1 above, comprising the following steps:

[0101] S10.1 Use mobile IoT sensing devices to collect groundwater flow path data and pollutant concentration data in real time;

[0102] S10.2 Based on groundwater flow path data and pollutant concentration data, the change point detection algorithm, the post-rain lag pollution identification algorithm, and the spatial comparison anomaly identification algorithm are used in sequence to identify and analyze the dynamic evolution events of underground pollution in the mine, and output the evolution results of the dynamic evolution events of underground pollution in each region.

[0103] S10.3 Based on the evolution results of groundwater pollution dynamic evolution events in various regions, construct a groundwater pollution propagation evolution map covering the entire path, and identify the upstream pollution source origin, downstream terminal pollution area and midstream potential pollution source;

[0104] S10.4 Based on the dynamic evolution results of underground pollution events, the starting point of upstream pollution sources, the downstream terminal pollution areas and potential pollution sources in the middle reaches, construct a mine pollution risk level zoning map and visualize it on the terminal equipment.

[0105] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. An intelligent mine management system based on Internet of Things, characterized in that, include: Mine pollution data acquisition unit (1) uses a mobile Internet of Things sensor to collect groundwater flow path data and pollutant concentration data in real time; Event-driven pollution identification unit (2) Based on groundwater flow path data and pollutant concentration data, the event-driven pollution identification unit (2) sequentially adopts change point detection algorithm, rain-induced lag pollution identification algorithm and spatial comparison anomaly identification algorithm to identify and analyze the dynamic evolution events of underground pollution in the mine, and outputs the evolution results of the dynamic evolution events of underground pollution in various places. The underground pollution dynamic evolution events include pollution pathway convergence and abrupt change events, post-rain delayed pollution release events, and spatially controlled pollution anomaly events; The event-driven pollution identification unit (2) includes a rainfall-delayed pollution identification module (22); the rainfall-delayed pollution identification module (22) identifies rainfall-delayed pollution release events in the mining area based on historical rainfall data and pollutant concentration data and using a rainfall-delayed pollution identification algorithm. The post-rain lag pollution identification algorithm is constructed based on the fusion of a variable-order fractional-order kernel response model and a sparse compressed sensing reconstruction method. This algorithm is used to calculate the lag pollution release intensity time-series curve and hysteresis delay length of the post-rain heavy metal pollution release intensity based on historical rainfall data and pollutant concentration data. The lag pollution release intensity time-series curve represents the dynamic evolution trajectory of heavy metal pollutant concentration in groundwater over time; the hysteresis delay length represents the time interval between increases in heavy metal pollutant concentration in groundwater after rainfall. The event-driven pollution identification unit (2) includes a spatial contrast anomaly identification module (23); the spatial contrast anomaly identification module (23) is used to identify local pollution concentration anomaly areas in the flow path of mine groundwater, construct a dynamic map of water flow path pollution based on groundwater flow path data and pollutant concentration data, and use a spatial contrast anomaly identification algorithm to filter out the set of path spatial location nodes with abnormal concentration changes in the dynamic map of water flow path pollution, and the three-dimensional area enclosed by the set of adjacent path spatial location nodes with abnormal concentration changes is taken as the local pollution concentration anomaly area; The spatial anomaly identification algorithm is based on the fusion of graph convolutional neural network and local outlier analysis method. Specifically, it includes: using graph convolutional neural network to extract the concentration change characteristics of each spatial location node in the dynamic map of water flow path pollution; using local outlier analysis method to calculate the concentration change score of each spatial location in the path, and screening out the set of spatial location nodes with abnormal concentration changes. Pollution path evolution identification unit (3) Based on the evolution results of various underground pollution dynamic evolution events, the pollution path evolution identification unit (3) constructs a groundwater pollution propagation evolution map covering the entire path, and identifies the upstream pollution source starting point, downstream terminal pollution area and midstream potential pollution source; The pollution situation visualization unit (4) is used to receive the underground pollution dynamic evolution event evolution results from the event-driven pollution identification unit (2) and the upstream pollution source starting point, downstream terminal pollution area and midstream potential pollution source from the pollution path evolution identification unit (3), construct a mine pollution risk level zoning map, and visualize it on the terminal device. 2.The Internet of Things based intelligent mine management system according to claim 1, characterized in that, The mobile IoT sensing device is used to move along the direction of groundwater flow in the groundwater aquifer area and collect data on the flow path of groundwater and pollutant concentration in real time during the movement. The flow path data includes the location coordinates of the mobile IoT sensing device, the rate of change of flow direction, and the water flow velocity information; the pollutant concentration data includes the concentration of heavy metal ions, pH value, conductivity, and redox potential.

3. The intelligent mine management system based on the Internet of Things according to claim 2, characterized in that, The event-driven pollution identification unit (2) includes a path intersection mutation identification module (21); the path intersection mutation identification module (21) is used to analyze the intersection of water flow paths based on flow path data and pollutant concentration data, identify the pollution concentration change at the intersection of water flow paths using a change point detection algorithm, and record and output the event time series, spatial location series and pollution intensity change of the pollution path intersection mutation event.

4. The intelligent mine management system based on the Internet of Things according to claim 3, characterized in that, The post-rain lag pollution identification algorithm calculates the lag pollution release intensity time-series curve and hysteresis delay length of the post-rain heavy metal pollution release intensity. The specific method is as follows: S2.2.

1. Based on the time series of historical rainfall data and pollutant concentration data, establish a dynamic input-output aligned dataset; S2.2.2 Based on the dynamic input-output aligned dataset, construct a variable-order fractional-order kernel response function for the time series of historical rainfall data; S2.2.

3. Using the sparse compressed sensing reconstruction method, the time series of pollutant concentration data is sparsely reconstructed, the main excitation components during pollutant release are extracted, and the pollution concentration abrupt change segments are identified. S2.2.4 Calculate the dynamic curve of pollution concentration change over time, and extract the time length from the start of rain to the peak pollution concentration in combination with the pollution concentration abrupt change segment, as the hysteresis delay length. S2.2.

5. Combining the variable-order fractional-order kernel response function and the main excitation components, the change of pollution response intensity over time is constructed as a time-series function, generating a time-series curve of hysteresis pollution release intensity. S2.2.

6. The final output includes the pollution concentration abrupt change range, hysteresis delay length, and time series curve of the delayed pollution release intensity of the post-rain delayed pollution release event.

5. The intelligent mine management system based on the Internet of Things according to claim 4, characterized in that, The method for constructing the groundwater pollution propagation evolution map with full path coverage is as follows: The high-slope intervals of the main excitation components are extracted as high-intensity time windows when pollutants are released. Combined with flow path data, the sources of pollution release on the flow path are identified as the starting nodes of the groundwater pollution propagation evolution map. Based on the mobile path data of mobile IoT sensing devices, the terminal range of pollution diffusion impact is calculated, which serves as the end node of the groundwater pollution propagation evolution map. Based on the hysteresis delay length and water flow velocity information, the spatial propagation distance of pollutants from the surface to the underground response is calculated, and the main path edge from the starting node to the ending node is constructed. All nodes in the set of spatial location nodes of abnormal concentration change paths are used as intermediate nodes in the groundwater pollution propagation evolution map. Based on the groundwater flow path data and spatial relationships, the spatial location nodes of the abnormal concentration change path are unidirectionally connected to the main path edge to construct the branch path edge; Based on the concentration change score at each spatial location along the path, the concentration gradient weights are assigned to each intermediate node on the corresponding branch path edge. The final result is a complete map of the evolution of groundwater pollution propagation.

6. The intelligent mine management system based on the Internet of Things according to claim 5, characterized in that, The groundwater pollution propagation and evolution map is used to identify the upstream pollution source origin, the downstream terminal pollution area, and the midstream potential pollution source, as detailed below: S3.

1. Based on the groundwater pollution propagation evolution diagram, find the node where the first increase in pollutant concentration occurs in the main path from the starting node towards the water flow direction. This node is the starting point of the upstream pollution source. S3.2 The area enclosed by the end node of the groundwater pollution propagation evolution diagram and the neighboring nodes of that node is the downstream terminal pollution area; S3.

3. Based on the groundwater pollution propagation evolution diagram, find a set of candidate nodes in the branch path where the concentration change score at the spatial location of the path is greater than the set score threshold. All candidate nodes in this set are potential midstream pollution sources. Based on the time series curves of the lagged pollution release intensity of all candidate nodes, the slope value of each time window is extracted. If the slope values ​​of all time windows of the lagged pollution release intensity time series curve are less than the response slope threshold, the candidate node to which the lagged pollution release intensity time series curve belongs is deleted. Calculate the ratio of the hysteresis delay length from the remaining candidate node to the node on the main path to the path distance. If the ratio is within the set groundwater flow velocity range, the candidate node is located on a reasonable pollution transmission path and is deleted. The remaining candidate nodes were all potential sources of pollution in the midstream.

7. The intelligent mine management system based on the Internet of Things according to claim 6, characterized in that, The mine pollution risk level zoning map is based on data from the groundwater pollution propagation and evolution map, upstream pollution source origin, downstream terminal pollution area, midstream potential pollution source, and pollutant concentration data to divide the pollution area into zones and visualize the management of mine area pollution.

8. An intelligent mine management method based on the Internet of Things (IoT), used in the intelligent mine management system based on the Internet of Things as described in any one of claims 1-7, characterized in that: Includes the following steps: S10.1 Use mobile IoT sensing devices to collect groundwater flow path data and pollutant concentration data in real time; S10.2 Based on groundwater flow path data and pollutant concentration data, the change point detection algorithm, the post-rain lag pollution identification algorithm, and the spatial comparison anomaly identification algorithm are used in sequence to identify and analyze the dynamic evolution events of underground pollution in the mine, and output the evolution results of the dynamic evolution events of underground pollution in each region. S10.3 Based on the evolution results of groundwater pollution dynamic evolution events in various regions, construct a groundwater pollution propagation evolution map covering the entire path, and identify the upstream pollution source origin, downstream terminal pollution area and midstream potential pollution source; S10.4 Based on the dynamic evolution results of underground pollution events, the starting point of upstream pollution sources, the downstream terminal pollution areas, and the potential pollution sources in the middle reaches, a mine pollution risk level zoning map is constructed and visualized on the terminal equipment.

Images