Coal mine underground environment information monitoring system based on Internet of Things converged communication

Abstract

The invention discloses a coal mine underground environment information monitoring system based on Internet of Things converged communication, which is applied to the field of environment information monitoring and aims to solve the problem that in the prior art, underground environment information is inconvenient to intuitively know due to the complex underground environment of a coal mine. According to the method, three-dimensional modeling is carried out on an underground environment, and underground information is displayed in a three-dimensional modeling image through a GIS (Geographic Information System); and meanwhile, underground data are collected through a sensor, anomaly detection is carried out on the underground data, and early warning is carried out on the abnormal condition.

Description

Technical Field

[0001] The invention belongs to the field of environmental information monitoring, and in particular relates to an underground coal mine environmental information monitoring system. Background Art

[0002] A coal mine is a place or facility where coal resources are extracted. Coal is a fossil fuel formed primarily from plants under pressure and heat underground over millions of years, and a coal mine is the site of mining activities to extract this fuel.

[0003] In the prior art, since most coal mines are often distributed in underground coal layers, people need to dig wellheads and then mine along the coal layers when excavating coal mines.

[0004] In the actual excavation process of coal mines, carrying out environmental information monitoring has great practical significance:

[0005] 1. Ensure the safety of workers' lives: The working environment of coal mines is special and risky, especially underground operations, which are often extremely dangerous. In some coal mines, there are a lot of harmful gases including methane, carbon monoxide, and carbon dioxide, which may lead to suffocation and poisoning risks. At the same time, due to the high temperature and high humidity inside some coal mines, miners will suffer from heat stroke and dehydration. Some coal mines even have great underground pressure due to poor geological conditions, which may cause coal mine collapse. With the help of environmental information monitoring, various environmental data can be tracked and monitored in real time, abnormal conditions can be discovered and handled in time, the accident rate can be reduced, and the personal safety of operators can be ensured.

[0006] 2. Effectively improve production efficiency: Through environmental information monitoring, the equipment operation status can be fully monitored, such as equipment operating status, temperature, current, voltage, etc., to achieve remote control of equipment and reduce equipment failure rate. At the same time, underground personnel can also transmit relevant information back in time through system communication means, thereby helping to improve coal mine production efficiency.

[0007] 3. Enhance environmental awareness: Coal mining and use may cause a variety of environmental problems, such as air pollution, water and land resource damage, and noise pollution. With the help of environmental information monitoring, it is easy to grasp the pollution situation and take timely response measures to achieve a virtuous interaction between economic development and environmental protection.

[0008] In general, environmental information monitoring plays a key role in the underground environment of coal mines. It can not only ensure the life safety of operators and improve production efficiency, but also enhance environmental awareness.

[0009] At present, there are tools for environmental information monitoring in coal mines, but there are a series of problems. First, the working area of ​​coal mines, especially the underground terrain, is complex and changeable, and the layout of existing monitoring tools is relatively scattered, covering only the environmental data of specific locations or part of the equipment data, making the collected data not detailed and comprehensive. At the same time, it is also limited by the information transmission method, which leads to the inability of these tools to accurately reflect the overall terrain of the mine and the structure of the channel, and to fully monitor the environmental conditions of various areas in the coal mine, affecting the overall control of the mine area. Secondly, the existing monitoring tools are relatively simple in function, usually only providing basic data entry, query and viewing functions, lacking intelligent analysis and linkage effects, resulting in low monitoring efficiency. In many cases, manual processing is required, resulting in low real-time and accuracy, and it is difficult to effectively respond to emergencies. Third, the underground working environment is harsh, the space is narrow, the geological conditions are complex, and coupled with problems such as electromagnetic interference and signal attenuation, traditional radio wave communication equipment often cannot work effectively underground. In addition, the underground workers in coal mines are highly mobile and their working positions are not fixed. All of these factors have caused serious restrictions on underground communication methods, making it difficult to meet the communication needs of mine workers. Summary of the invention

[0010] In order to improve the problem in the prior art that it is not convenient to intuitively understand the underground environmental information due to the complex underground environment of the coal mine, the present invention provides a coal mine underground environmental information monitoring system based on Internet of Things fusion communication.

[0011] The technical solution adopted by the present invention is: a coal mine underground environment information monitoring system based on Internet of Things fusion communication, including: a sensor receiving module, a data processing module, a central control module and a fault warning module;

[0012] The sensor receiving module is used to receive data detected by various sensors installed downhole under the control of the central control module, and transmit the received data to the data processing module;

[0013] The data processing module is used to pre-process the data transmitted by the sensor receiving module under the control of the central control module;

[0014] The fault warning module, under the control of the central control module, detects the pre-processed data output by the data processing module and issues a warning for detected abnormal situations.

[0015] The data received by the sensor receiving module include gas concentration, temperature and humidity, wind speed, wind direction, dust concentration, pressure change, water level and seepage, and vibration.

[0016] The fault warning module specifically includes an abnormality detection unit and an abnormality matching unit;

[0017] The anomaly detection unit performs anomaly detection on the data preprocessed by the data processing module according to the threshold value corresponding to each data;

[0018] The abnormal matching unit first extracts key feature values ​​from the abnormal data output by the abnormal detection unit to form an abnormal feature vector to be matched; calculates the similarity between the abnormal feature vector to be matched and the abnormal feature vectors corresponding to each abnormal state in the historical record; and uses the abnormal state with the greatest similarity as the abnormal situation of the abnormal feature vector to be matched.

[0019] Furthermore, an explosion-proof wireless communication module is used for communication between the sensor and the sensor receiving module in the mine.

[0020] It also includes: a three-dimensional modeling module and a GIS module; the three-dimensional modeling module generates three-dimensional modeling graphics according to the development drawings of the mine under the control of the central control module; the GIS module is used to divide the three-dimensional modeling graphics into regions and mark various information in the mine under the control of the central control module, and the marked information includes the specific location of the equipment, the location of the electrical lines and the location of the personnel.

[0021] Beneficial effects of the present invention: Compared with the prior art, in the present invention, through the fault warning module, it is possible to automatically provide feedback on continuously generated abnormal conditions, thereby reducing the amount of false alarms caused by occasional abnormal data. At the same time, through the analysis and matching of abnormal information, it is possible to automatically assign corresponding management personnel to the abnormal condition to rush to the site for maintenance. When the management personnel rush to the abnormal area, they can also formulate corresponding maintenance plans based on the abnormal information fed back by the mobile terminal. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 This is the program flowchart of the coal mine underground environmental information monitoring system based on the Internet of Things integrated communication. DETAILED DESCRIPTION

[0023] To facilitate those skilled in the art to understand the technical content of the present invention, the present invention is further explained below with reference to the accompanying drawings.

[0024] The coal mine underground environment information monitoring system based on the Internet of Things integrated communication of the present invention includes: a sensor receiving module, a communication module, a data processing module, a three-dimensional modeling module, a GIS module, a central control module, a fault warning module, an energy management module, and a remote management module; the specific implementation of each module is described in detail below:

[0025] 1. Sensor receiving module

[0026] It is used to sense and detect the environmental data in the coal mine. The sensor receiving module includes a sensor, a data identification module and a data transmission module. The sensors are widely distributed in various areas in the coal mine and are used to monitor environmental parameters such as temperature, humidity, gas concentration, etc. in real time. The data acquisition module sorts the collected signals for subsequent processing. The data identification module is used to identify the identities of these sensors according to the communication connection port numbers of different sensors, and assign specific identifiers to the received sensor information data according to the functions of each group of sensors. The data transmission module is used to transmit the data identified by the data identification module to the data processing module through a communication connection. During the transmission process, the data transmission module can use wired or wireless methods to exchange data, ensuring the flexibility and efficiency of information flow.

[0027] 2. Communication module

[0028] The communication module is used to perform mutual communication between data and voice and video communication of underground workers. It includes a network construction module, a protocol data intercommunication module, and an image and voice playback module. The network construction module is the foundation of the entire IoT communication network and is responsible for building data transmission channels between sensors and communication links between the remote management module and each group of sensor devices. In this module, we use communication technologies that adapt to the special environment of mines, including explosion-proof wireless communication modules and low-power wide area network technology to ensure the stability and security of the network.

[0029] The protocol data intercommunication module is used to process the data protocol and data format conversion between different sensors, and to realize the update of the control detection program by signal transmission overwriting. Since there are many types of sensors in coal mines, and the data protocols and formats generated are different, the protocol data intercommunication module adopts flexible data processing and protocol conversion technology to ensure that various sensor data can be smoothly transmitted in the Internet of Things communication network. In addition, the module also realizes the update of the control detection program by signal transmission overwriting to meet the real-time processing requirements of sensor data in the mine production process.

[0030] The image and voice playback module is used to realize voice and video communication of underground workers. The module adopts high-performance voice recognition and video compression technology to ensure that workers can obtain clear voice and real-time image information in complex mine environments. In emergency situations, the image and voice playback module can provide real-time on-site images to help rescuers quickly understand the situation at the accident site, thereby improving rescue efficiency.

[0031] In the present invention, through the communication module, real-time message voice and video communication can be realized between the ground and underground personnel in the coal mine, which can not only enable the rapid transmission of instructions and feedback to improve work efficiency, but also improve the ability to respond to emergencies and provide timely rescue in emergency situations, thereby providing strong protection for coal mine safety production.

[0032] 3. Data processing module

[0033] It is used to collect data transmitted from the sensor and process the data. The data processing module includes a data collection module, a data cleaning module and a data classification module.

[0034] 31. The data acquisition module is used to collect the sensor detection signals sent by the sensor receiving module, and transmit the sensor signals to the data cleaning module in order of generation time.

[0035] 32. The data cleaning module is used to filter and denoise the received sensor signal data, and standardize it according to the type of sensor signal data received. The purpose of data filtering is to eliminate obviously erroneous, abnormal or invalid sensor signal data to ensure high quality of subsequent processed data. Noise reduction is to reduce random noise in sensor signals and enhance the availability of valid signals. Standardization is used to convert sensor signal data into a unified scale for further processing and analysis. For the data cleaning process, a combination algorithm can be used to combine adaptive filtering, noise reduction and standardization to achieve efficient cleaning and optimization of sensor signal data. Data cleaning specifically includes the following steps:

[0036] 321. Adaptive Filtering:

[0037] Assume that the raw data collected by the sensor is x(t), where t represents the time point. First, adaptive filtering is applied to remove obvious outliers in the data. Adaptive filtering determines whether a data point is an outlier by setting a dynamic threshold. The threshold can be determined based on the mean μ(t) and standard deviation σ(t) of the real-time data. If |x(t)-μ(t)|>k·σ(t)=>x(t) is considered an anomaly, it is processed according to the following rules:

[0038]

[0039] Here, k is a user-defined parameter, usually ranging from 2 to 3, which is used to indicate the tolerated deviation.

[0040] 322. Noise reduction processing:

[0041] After filtering out obvious anomalies, the data is then subjected to denoising. Here, wavelet transform can be used to decompose the data into different frequency bands to remove high-frequency noise. Assume that the data after wavelet transform is represented as W[x filtered (t)], each wavelet coefficient can be denoised by the adaptive threshold method. The denoising formula is:

[0042]

[0043] in, is an adaptive threshold calculated according to the data noise level, and λ(t) represents the dynamic adjustment factor of the threshold function, which is used to adjust the threshold in different segments of the data.

[0044] 323. Standardization processing:

[0045] The denoised data is then standardized to eliminate the scale differences between different data features. The result after standardization is expressed as y(t), and its calculation formula is:

[0046]

[0047] Here μ and σ are the mean and standard deviation of the denoised data, respectively. Through standardization, the processed data can have the characteristics of zero mean and unit variance, thus adapting to subsequent analysis and modeling.

[0048] 33. The data classification module is used to classify the received data. The data input from the data acquisition module can be divided into parameter categories that reflect environmental characteristics, such as gas concentration, temperature and humidity, wind speed and direction, dust concentration, pressure change, water level and seepage, vibration, etc., and the feedback signals of the same type of sensors are summarized based on these categories. This not only reduces redundant data, but also facilitates subsequent data analysis and decision-making.

[0049] These characteristic parameters are then further divided into real-time data, historical data, and predicted data. Real-time data refers to the key environmental parameters collected from the sensor at the current moment. These data need to be monitored in real time, reflecting the immediate status of the underground environmental characteristics, so as to detect abnormalities in time and take emergency measures. Historical data refers to the storage of real-time data collected by sensors over a long period of time to form a historical database. For example, historical records of gas concentrations over the past few days, weeks, and months. Historical data is used to analyze environmental change patterns and equipment operating conditions, which helps to identify long-term trends and periodic changes and provide a basis for management decisions.

[0050] Then, based on real-time data and historical data, forecast data can be generated by following the steps below:

[0051] a. Data preprocessing: Check and process missing values, outliers, and noise in existing real-time or historical data, and use interpolation, filling, or deletion methods to handle missing data.

[0052] b. Feature engineering: Extract key features from the data, such as peak value, average value, standard deviation, rate of change, etc., and extract time series features based on the data collection time point to provide corresponding input for the subsequent model.

[0053] c. Build the model: Create a model based on a multi-layer LSTM neural network, organize the processed feature vectors according to time series features, and pass them into the model as input.

[0054] d. Model training and evaluation: Start training the model based on the Adam optimizer and the MSE loss function, and use the early stopping method to prevent overfitting.

[0055] e. System deployment: After the model training is completed, a prediction processing pipeline is established based on the above steps, including extracting real-time and historical data, data preprocessing, feature extraction and model inference, and finally obtaining the predicted data.

[0056] The prediction data is generated entirely based on the real-time and historical data obtained from sensor detection. Without real-time and historical data, the prediction model cannot run.

[0057] After the prediction data is generated, the prediction result is stored in the prediction database, and the record content includes the predicted future time point, environmental parameters and their values, and model version. When the real time reaches the future time point corresponding to the prediction, the system will receive the current real-time data feedback. At this time, the data at both ends can be used to complete the optimization cycle through reinforcement learning to improve the monitoring ability of the entire system. The specific implementation steps are:

[0058] a. Data matching: The data matching process requires both temporal and spatial alignment. Temporal alignment means matching the predicted results with the actual real-time data according to the timestamp to ensure that the comparison is made at the same time point. Spatial alignment means that if the sensor has geographic location information, the compared data is ensured to be from the same location.

[0059] b. Error calculation: the model effect is evaluated by selecting the MSE (mean square error) error evaluation indicator.

[0060] c. Optimization based on reinforcement learning:

[0061] a) Environment definition: The coal mine underground environment monitoring system is used as the environment for reinforcement learning, and the data collected by the sensor is used as the environment state.

[0062] b) Designing an agent: The goal of the agent is to select the optimal action (such as adjusting sensor sampling frequency, model parameters) to minimize prediction error and monitoring cost.

[0063] c) Define the state: The state is the current sensor data and prediction error, as well as the sensor configuration

[0064] (position, sampling frequency, etc.).

[0065] d) Designed actions: The designed actions include adjusting the sensor sampling frequency, adjusting the sensor deployment location, etc.

[0066] e) Define the reward function: Define the reward function based on the prediction error. The reward can be set to the negative prediction error. The smaller the error, the higher the reward, that is:

[0067] R=-error value

[0068] f) DQN is selected as the reinforcement learning algorithm because the coal mine environmental parameters are generally continuous data rather than discrete values, and the algorithm is suitable for continuous state and action space, using neural networks to approximate value functions or policy functions. When constructing the model structure, the state is designed as input and the output is the value or policy of the action.

[0069] g) Complete the training process. At each time step, the agent selects an action based on the current state. After executing the action, the environment feedbacks the new state and reward. Continue iterative training, use the continuously generated environmental feedback as a new round of training data, and continuously optimize the model performance.

[0070] d. Optimize the sampling frequency and deployment location of sensors based on the action results output by the model. Optimizing the deployment location of sensors can solve the problem of missing key data due to the lack of sensors in key areas, the inability of the model to accurately predict environmental changes in the corresponding areas, and the problem of repeated collection of similar data, which reduces the value of data; and

[0071] It can also avoid monitoring blind spots that affect the comprehensiveness of the prediction.

[0072] Similarly, deploying sensors at key locations (such as areas with sensitive changes in environmental parameters, high risks, and high safety requirements) can more accurately obtain key location data and improve the model's monitoring capabilities for important areas; the appropriate spacing between sensors ensures the continuity of the monitoring range; and avoids monitoring blind spots to improve prediction accuracy.

[0073] e. Form an optimization closed loop. After adjusting the actual operating status of the sensor, the optimization result will be fed back to the sensor receiving module, forming a closed-loop control system from sensor receiving, data classification, model prediction, reinforcement learning to sensor receiving. Based on this process, the system can autonomously adjust parameters according to environmental changes and model performance to achieve adaptive optimization. Through continuous cycles and iterations, the system's monitoring capabilities and prediction accuracy are continuously improved.

[0074] 4. 3D modeling module

[0075] It is used to perform three-dimensional modeling of underground environmental information of coal mines and to mark corresponding characteristic information in the mines. The three-dimensional modeling module includes a drawing import module, a model generation module and a sensor import module.

[0076] 41. The drawing importing module reads the development design drawings of the mine cave through an external carrier, imports the development drawings of the mine cave, unifies and converts the imported coal mine cave drawing data, and transmits the converted data to the model generation module.

[0077] The main contents read from external carriers include geometric data, material and geological information, structure and layout, annotations and tags:

[0078] 1) For geometric data, this is the core part of the drawing, including the geometry, coordinate points, specific dimensions and three-dimensional spatial relationships of the mine outline, tunnels, lanes, etc. These data are precisely defined by the coordinate system and are usually provided in two-dimensional or three-dimensional CAD format.

[0079] 2) For material and geological information, the drawings often indicate the geological composition around the mine, such as the thickness of the coal seam,

[0080] The hardness and water content of the rock formation, etc. This information plays an important role in subsequent 3D modeling and simulation.

[0081] 3) For structure and layout, including the specific location, type and functional layout of the equipment in the mine. For example, the specific distribution of lifting equipment, ventilation equipment, drainage system, transportation equipment, etc. in space.

[0082] 4) For notes and markings, these are special instructions or annotation information in the drawings, such as safety signs at certain locations,

[0083] Maintenance reminders or detection points, etc., these marks usually also need to be marked accordingly in the 3D model.

[0084] Since the data in the mine drawings may come from different systems or standards, the units are not uniform. Therefore, it is necessary to use relevant technical means to convert and standardize these data into uniform physical measurement units, and then pass the sorted data to the model generation module for subsequent work. The specific conversion technologies that can be used are:

[0085] 1) CAD data parsing tool: Use APIs such as AutoCAD or other similar CAD parsing libraries to read geometric data in drawings and convert the format, for example, from DXF, DWG or IFC format to a standardized data format that the system can handle, such as JSON, XML.

[0086] 2) Physical unit conversion algorithm: The unit standards used in different regions and systems may be inconsistent. Through a custom unit conversion algorithm, the system can convert different physical quantities (such as length, area, volume, mass, etc.)

[0087] Convert to the required units uniformly. For example, convert imperial units (such as feet, inches) to metric units (meters, centimeters).

[0088] 3) Spatial coordinate conversion: Mine development drawings may use different coordinate systems. We need to use geographic coordinate conversion tools (such as GIS tools) to uniformly convert different coordinate systems (such as WGS84 and UTM coordinate systems) to ensure geometric accuracy in 3D modeling.

[0089] 42. The model generation module is used to construct a vivid and accurate three-dimensional mine model in a vivid form according to the large amount of original data uploaded in the drawing import module, making full use of modern scientific and technological means, adopting a variety of algorithms, and strictly following the standards and principles of digital three-dimensional modeling. The model model contains precise and detailed information such as the mine's topographic features, geological conditions, equipment layout, etc.

[0090] During this process, the generated 3D model also needs to be refined until it reaches the highest accuracy requirements. The implementation steps can be mainly divided into the following stages:

[0091] 1) Data preprocessing: First, the drawing import module analyzes the geometry, size, material and other data imported from the external carrier. The conversion and standardization of physical units have been completed before.

[0092] 2) Geometric modeling: Using the imported geometric data, the initial 3D model of the mine is first generated. This step usually uses mesh generation algorithms in computer graphics (such as Marching Cubes or Delaunay triangulation) to convert 2D or 3D point cloud data into a 3D surface model. The purpose of geometric modeling is to accurately depict the structures such as tunnels and shafts in the mine.

[0093] 3) Embedding of geological information: After the geometric model is generated, the system combines the geological information of the mine (such as rock layers, coal seams, groundwater, etc.) with the geometric model. This process requires the use of geological model algorithms to volumetrically model the underground geological data and superimpose it with the mine structure.

[0094] 4) Equipment and layout insertion: The equipment layout information in the mine will also be imported into the 3D model. The model generation module will place the lifting equipment, ventilation equipment and other facilities in the model according to the real location based on the size and location data of the equipment. This process can be achieved through the equipment library, which contains the 3D models and installation specifications of various types of equipment.

[0095] 5) Model repair and optimization: The initially generated 3D model may have some inaccuracies in details, especially due to terrain complexity or data errors. Therefore, the model generation module needs to be repaired later. This step can be done with the help of algorithms (such as automatic mesh optimization, geometry correction algorithms, etc.) to automatically repair the unreasonable geometric structure in the model and improve the overall accuracy of the model.

[0096] 6) Sensor labeling and area coloring: Import the data of various sensors into the model, and label the location of each sensor and its corresponding detection range on the 3D model. By color-coding the areas within the sensor detection range, the safety status of different areas in the mine can be displayed in real time. For example, green indicates normal, and orange or red indicates abnormal detection.

[0097] 7) Model rendering and interaction: Finally, the repaired model needs to be visualized by a graphics rendering engine to generate high-quality 3D model images and support user interaction with the model. This can be achieved through a 3D rendering engine (such as Unity or Unreal Engine), allowing users to view the details of the mine model by rotating, zooming, etc.

[0098] 43. The sensor import module is used to add each group of sensors to the model generation module according to their installation positions, and to mark the sensor detection range of the group of sensor devices accordingly. At the same time, the sensor detection results of the sensor detection range area are marked by adding colors. If the detection area data is normal, the sensor detection area will be marked in green. If intermittent detection abnormalities occur, the group of sensor detection areas will be marked in orange or red in the three-dimensional graph.

[0099] In the present invention, by importing the design drawings of the mine through the three-dimensional modeling module, the internal environment of the mine can be modeled, and through the import division of the sensor import module, the position and detection area of ​​each group of sensors can be intuitively displayed, so that people can coordinate the management of the overall situation of the mine through the control terminal.

[0100] 5. GIS module

[0101] Used to display various information of the coal mine in the generated three-dimensional modeling image, the GIS module includes an information annotation module, a region division module and a miner positioning module;

[0102] The information annotation module is used to annotate various information in the mine through text labels in the 3D modeling image using GIS positioning technology, including the specific location of equipment and electrical line location. Through this module, workers can quickly and accurately find the required equipment, thereby improving work efficiency. At the same time, the module can also update the equipment location information in real time, so that managers can grasp the dynamic changes in the mine.

[0103] The area division module is used to mark the mine model established in the 3D modeling according to the actual area division, including the power supply area, liquid supply area, water storage area, mining area and area to be excavated. Through this module, the staff can clearly understand the division of each area in the mine, which is convenient for targeted work. In addition, the module can also adjust the area according to the actual situation to meet the needs of mine production.

[0104] The miner positioning module is used to locate the position of each operating employee entering the coal mine through GIS positioning technology, and feed back the corresponding position information to the three-dimensional modeling image in the form of highlighted light spots. Through GIS fusion positioning technology, the module can obtain the location information of each employee in real time and feed this information back to the three-dimensional modeling image. In this way, managers can keep track of the location of employees at any time to ensure the safety of mine production. At the same time, the location information is displayed in the form of highlighted light spots, making the position of employees in the mine more eye-catching in the three-dimensional map. The specific functions of this positioning module are as follows:

[0105] 51. Wireless positioning technology

[0106] Using wireless signals (including WiFi, Bluetooth, RFID, etc.) to establish location base stations in coal mines, each employee wears a tag equipped with a wireless positioning module. By collecting signal strength and delay information between the employee and the base station, the employee's location can be estimated.

[0107] Using the signal strength and delay information obtained by wireless positioning technology, the location of the employee can be estimated through the multi-lateration positioning algorithm. Assume that there are n base stations and the distance between the i-th base station and the employee is d i , then the employee's position (x, y) can be calculated by the following formula:

[0108]

[0109] Where (x i ,y i ) is the coordinate of the i-th base station.

[0110] 52. Inertial Navigation Sensor

[0111] The positioning tags worn by employees have built-in inertial sensors such as accelerometers and gyroscopes to measure the employee's movement status and direction changes. These data can be combined with the location information provided by wireless positioning technology to track the employee's movement trajectory in real time.

[0112] Combined with inertial navigation sensor data, a motion model can be used to predict the employee’s next location. Assume that the employee’s current location is (x t ,y t ), the speed is (v x , v y ), the acceleration is (a x , a y ), the time interval is Δt, then the next step position (x t+1 ,y t+1 ) can be predicted by the following formula:

[0113]

[0114]

[0115] 53. Display in 3D map

[0116] The 3D modeling and GIS modules are used together to create a 3D map of the coal mine, including information such as tunnels, work areas, equipment locations, etc. These map data can be combined with real-time positioning information to provide a more intuitive display of personnel locations.

[0117] By matching the real-time positioning information with the underground map data, the employee's location can be mapped to the map so that it can be displayed in the 3D modeling image. The matching process can be achieved through the following algorithms:

[0118] Assume that the coordinates of the real-time point are (x real ,y real ), the coordinates of the i-th known point on the map are (x i ,y i), then the distance d between the real-time point and the i-th map point i It can be calculated by the following formula:

[0119]

[0120] The map point closest to the real-time point is found as the nearest neighbor point, thus realizing the mapping of employee locations.

[0121] In the present invention, through the established GIS module, various equipment and partitions inside the mine can be accurately located, and the partitions can be marked so that people can plan the subsequent development area of ​​the mine through the positioning information in the GIS module, and it is also convenient to optimize the division of the internal area of ​​the mine.

[0122] 6. Central control module

[0123] Used to store and analyze data in transmission or processing, the central control module includes a data storage module, a real-time monitoring module and a data analysis module.

[0124] The data storage module is used to store real-time monitoring data, historical monitoring data and analysis results after deep mining in a database to facilitate subsequent query and analysis;

[0125] The real-time monitoring module is used to monitor the environmental parameters fed back by each group of sensors in the coal mine in real time, and perform data processing and data display through the central control module;

[0126] The data analysis module is used to perform statistics, analysis and prediction on the stored data to improve the intelligence level of the monitoring system. It also realizes effective prediction, thereby effectively improving the intelligence level of the entire monitoring system. The specific processing process includes the following steps:

[0127] 1) Data integration and normalization

[0128] First, in order to ensure the accuracy of data analysis, data from different sources need to be integrated and normalized. This includes but is not limited to environmental parameters such as temperature, humidity, and gas concentration, as well as equipment status data such as current and voltage. Data normalization is the prerequisite for ensuring that these data of different magnitudes and units can be compared equally in the same analysis model.

[0129] 2) Feature Engineering

[0130] Next, the process of feature engineering involves selecting features that are closely related to coal mine safety and efficiency, such as the methane concentration at a specific location in the mine or the temperature trend of a piece of equipment. The optimization of feature selection can be automatically performed through machine learning algorithms such as decision trees and random forests to determine which features are most relevant to the prediction target.

[0131] 3) Model construction

[0132] In order to analyze data and make predictions, it is necessary to build appropriate statistical and machine learning models. This can be matched to the corresponding model based on a variety of factors including prediction accuracy, model complexity, computational cost, and data characteristics.

[0133] 4) Real-time data stream analysis

[0134] Due to the real-time requirements of monitoring, the data analysis module needs to be able to process real-time data streams. Here, the Apache Flink framework is selected to process streaming data in order to achieve high throughput and low latency data processing effects.

[0135] 5) Forecasting and trend analysis

[0136] In addition to real-time monitoring, future predictions are also critical. Time series analysis methods can be used to predict the changing trends of environmental parameters over a period of time. These predictions can help mine managers make adjustments in advance and optimize production plans.

[0137] In the design of the time series analysis method, the system constructs a graph representation of environmental parameters, where nodes represent individual environmental parameters and edges represent the mutual influence between parameters. The network uses graph neural networks to capture and learn the complex relationships between these parameters, and dynamically adjusts the learning process through an adaptive mechanism to cope with real-time changes in the environment. Its core components include:

[0138] a. Graph construction: Build a graph based on the physical or statistical relationships between environmental parameters. Nodes are individual parameters, and edges are the connection strengths between parameters calculated based on correlation or mutual information.

[0139] Let x i is the i-th environment parameter, constructing a graph G = (V, E), where v is the node set and E is the edge set. The edge weight w ij By parameter x i and x j The mutual information I(x i ;x j )Decide:

[0140]

[0141] b. Graph neural network: used to process graph structure data, which can effectively transmit and integrate information between nodes. The network updates the node status through the message passing mechanism, thereby capturing the dependencies between environmental parameters.

[0142] The state update of each node v is determined by the following formula, where is the state of node v at layer k, and N(v) is the neighbor node of v:

[0143]

[0144] c. Adaptive mechanism: Incorporate environmental feedback to adjust network parameters. For example, if the prediction results deviate significantly from the actual monitoring data, the adaptive mechanism will adjust the learning rate or optimize the network weights to improve the model's predictive ability.

[0145] Let L be the loss function, t be the time step, α t is the learning rate of the tth step, and the adjustment method is:

[0146] α t+1 =α t exp(-γL t )

[0147] Among them, L t is the prediction error at step t, and γ is the adjustment coefficient.

[0148] 6) Feedback Mechanism

[0149] Finally, establish a feedback mechanism to continuously optimize the data analysis module. This can be done by collecting user feedback, monitoring the model's prediction error, and adjusting the model parameters and feature engineering operations based on this information.

[0150] 7. Fault warning module

[0151] Used to detect the alarm information in the coal mine according to the underground information feedback and sensor data, and the alarm warning information includes an abnormality detection module, an abnormality matching module and an abnormality dispatching module;

[0152] 71. The anomaly detection module is used to detect anomalies in environmental parameters underground in coal mines.

[0153] The underground environmental parameters come from the data processing module, which mainly include gas concentration, temperature and humidity, wind speed and direction, dust concentration, pressure change, water level and seepage, vibration and other types of data. The sensor will periodically collect various environmental data according to the configured frequency, and refer to the corresponding warning thresholds of various parameters set according to safety standards in the system.

[0154] In this embodiment, the gas concentration warning threshold is set, including: when the methane concentration is detected to be higher than 1.0%, the methane concentration is abnormal; when the carbon monoxide concentration is higher than 30ppm, the carbon monoxide concentration is abnormal; when the carbon dioxide concentration is higher than 0.5%, the carbon dioxide concentration is abnormal; when the oxygen concentration is lower than 18%, the oxygen concentration is abnormal.

[0155] In this embodiment, the temperature and humidity warning thresholds are set as follows: when the temperature is higher than 30°C or lower than 0°C, the temperature is abnormal; when the humidity is higher than 75% or lower than 25%, the humidity is abnormal;

[0156] In this embodiment, when the wind speed is higher than 5 m / s, or lower than 0.25 m / s, the wind speed is abnormal;

[0157] In this embodiment, when the dust concentration is higher than 4 mg / m 3 , then the dust concentration is abnormal;

[0158] In this embodiment, if the pressure change rate during the initial pressure is higher than 1MPa / hour, or higher than 0.3MPa / hour during the periodic pressure, the pressure change is abnormal;

[0159] In this embodiment, when the depth of accumulated water is greater than 30 cm, the water level is abnormal;

[0160] In this embodiment, when the PGA (peak ground acceleration) is higher than 40 gal, the vibration is abnormal.

[0161] If the parameter value exceeds the set threshold for N consecutive times (depending on the actual system configuration), the system will determine it as a persistent abnormality. If a single detection value exceeds the threshold but then returns to normal, it will be recorded as an occasional abnormality. The system will issue an alarm in a timely manner when a persistent abnormality is found.

[0162] In this embodiment, for a feature whose security standard is level 1, N is taken as 2 times; for a feature whose security standard is level 2, N is taken as 3 times; and for a feature whose security standard is level 3, N is taken as 5 times.

[0163] In the system, the above settings for warning thresholds and continuous monitoring times support dynamic configuration, and mines can make adjustments based on local safety supervision requirements and their own specific needs.

[0164] 72. The abnormal matching module is used to match possible abnormal situations according to the abnormal value fed back through system matching.

[0165] The basis for system matching mainly comes from the historical abnormal event library that contains various abnormal events that occurred in the past and their characteristic parameters, as well as the public expert knowledge base that records the abnormal patterns and corresponding processing methods summarized by mine safety experts.

[0166] Abnormal events and their corresponding system problems can be collected and obtained through the following ways:

[0167] (1) The mine directly provides abnormal matching data table information applicable to the current mine environment;

[0168] (2) If the mine cannot provide the data directly, it is necessary to summarize and establish an abnormal information table based on the mine's historical processing records and the experience of field staff before implementing the system;

[0169] After the abnormal information table is compiled, it is entered into the system as the initial configuration so that the system can begin to identify abnormalities according to the preset conditions.

[0170] For example, the abnormal information provided by a mine is as follows:

[0171] 1. Abnormal gas concentration accompanied by a sudden drop in wind speed is considered a ventilation system problem;

[0172] 2. If the humidity, water seepage and pressure changes are abnormal at the same time, it is determined to be a flood risk;

[0173] 3. When the temperature exceeds the limit and the vibration frequency of the equipment is abnormal, it is determined that the problem is caused by the wear of the mechanical parts of the equipment, rather than a simple heat dissipation failure;

[0174] In the subsequent actual use of the system, feedback from mine workers is crucial. They need to confirm and feedback the abnormal conditions reported by the system, thereby overwriting inaccurate historical results. In this way, the system can continuously improve the accuracy of analysis of various abnormal conditions in the mine environment, thereby improving the safety and work efficiency of the entire mine.

[0175] When performing actual matching, a set of key feature values, such as concentration, temperature, etc., are first extracted from the current anomaly data and combined into a set of anomaly feature vectors. Then, a list of similarity values ​​between the current anomaly and similar historical events is calculated in percentage.

[0176] Given that there are many feature parameters in the coal mine environment, the similarity calculation of anomaly vectors can be performed by introducing adaptive weights, feature importance scaling, and oscillation behavior for each feature:

[0177]

[0178] D(x,y) is used to represent the specific similarity value between two abnormal vectors x and y; i and i Respectively represent the specific characteristic values ​​in the two abnormal vectors, such as concentration, temperature, etc.; w iis an adaptive weight assigned to a feature type, reflecting the importance of the feature deviation. The reason is that not every environmental parameter is equally important. For example, gas concentration or sudden pressure change may be more critical than wind direction.

[0179] Weight w i The rules for determining specific values ​​are described below with examples:

[0180] (1) Based on the weighted baseline values ​​set by safety standards, abnormal characteristic parameters are divided into three levels according to the degree of danger:

[0181] a) Level 1 (fatal danger): gas parameters such as methane concentration and carbon monoxide concentration, with a base weight of 0.4

[0182] b) Level 2 (important warning): temperature, pressure, dust concentration and other parameters, with a base weight of 0.3

[0183] c) Level 3 (conventional monitoring): humidity, wind speed and other parameters, with a base weight of 0.2

[0184] Abnormal characteristic parameters that are closely related to dangerous events will receive higher baseline weights.

[0185] (2) Introduce the position weighting coefficient and set the position weighting coefficient according to the importance of the sensor installation location:

[0186] a) Coal mining face: 1.4

[0187] b) Return air channel: 1.2

[0188] c) Air inlet: 1.1

[0189] d) Other areas: 0.9

[0190] Intermediate weight = reference weight × position weighting coefficient

[0191] (3) Add the historical data distribution law, analyze the historical data of the past year, count the percentage of the parameter value exceeding the critical threshold, and add the calculated percentage result to the calculation of the actual weight:

[0192] Final weight = the result of the intermediate weight calculation in the previous step × (1 + abnormal proportion × 0.3)

[0193] After calculating through the above steps, the final initial weight w of each parameter can be determined iAt the same time, it is also necessary to add a weight update mechanism, set a weight update cycle (for example, 60 days), and regularly adjust the weight according to actual conditions. For example, if the safety standard is updated, the baseline weight corresponding to the feature will change; or if a sensor is added to a new location, the location weight coefficient of the corresponding feature will change; and historical data will be iteratively updated over time, so that the abnormal proportion of the feature will change.

[0194] M i The maximum possible value of a feature used to indicate normalization. By normalizing the values ​​to ensure that parameters of different ranges are scaled appropriately, this ensures that no single feature dominates the similarity calculation simply because it has a larger value range.

[0195] A reasonable M needs to be preset for each feature. i value to avoid the fluctuation of parameter value causing M i The value changes frequently, which affects the stability of the weight and the accuracy of the similarity calculation. i The values ​​can generally be set as follows:

[0196] Methane concentration: 3%

[0197] Carbon monoxide concentration: 150ppm

[0198] Carbon dioxide concentration: 5%

[0199] Oxygen concentration: 30%

[0200] Temperature: 50℃

[0201] Humidity: 100%

[0202] Wind speed: 10m / s

[0203] Dust concentration: 10mg / m 3

[0204] Pressure change: The pressure change rate during the initial pressure is 3MPa / hour, and the pressure change rate during the periodic pressure is 1MPa / hour

[0205] Water level (water depth): 100cm

[0206] Vibration (PGA): 80gal

[0207] If the actual value of a parameter exceeds the upper limit set above, it indicates that serious problems have occurred in the mine environment and there are serious safety hazards. The role of the specific value of the parameter has been weakened. At this time, the normalized result value can be directly set to 1, thereby marking the parameter as the highest risk and prompting the system to respond quickly.

[0208] An oscillation factor is introduced to account for parameters that have natural oscillations (e.g., temperature cycles between day and night). The sinusoidal curve ensures that these periodic changes are not interpreted as anomalies. The specific amplitude of the oscillation is given by a i This parameter allows us to adjust how much these cyclic changes affect the final similarity calculation. For parameters with larger amplitudes and higher vibration frequencies (such as temperature), they will have a higher a i The value is used to reduce the impact of periodic changes on the similarity calculation, while the parameters that do not exhibit this behavior (such as gas concentration) have a i It is approximately 0. Table 1 shows the a of each characteristic parameter i The value suggestion of .

[0209] Table 1 a of each characteristic parameter i Suggested value of

[0210] Characteristic parameters <![CDATA[a i Value range]]> illustrate temperature 0.3-0.5 The temperature has a distinct diurnal cycle. humidity 0.1-0.2 The diurnal cycle of humidity is relatively mild. Gas concentration 0.0 There is no obvious periodic change in gas concentration. pressure 0.1-0.3 The pressure may fluctuate slightly during operation. Wind speed 0.1 Wind speed will be affected by work activities, but not significantly. Dust concentration 0.2-0.4 The dust concentration fluctuates with the operation.

[0211] Then, the abnormal state with the highest similarity in the list is taken as the final matching result and sent to the control terminal for display.

[0212] 73. The abnormality dispatching module is used to match the relevant maintenance personnel according to the confirmation results of the abnormal state at the numerical control terminal, and send the abnormal condition location and abnormality type to the control terminal location of the maintenance personnel. In this embodiment, the abnormal state with the highest similarity percentage is sent to the control terminal for display, and after confirmation by the relevant personnel, it is recorded in the historical abnormal event library.

[0213] 8. Energy management module

[0214] Used to detect the electricity and water source information in the coal mine and ensure the stable supply of the two, the energy management module includes a power supply equipment module, a liquid supply equipment module and a data summary module; the power supply equipment module is used to detect the power supply situation in the coal mine to ensure the normal operation of the power supply equipment, the liquid supply equipment module is used to detect the operation of the liquid supply equipment in the coal mine to ensure the normal production of the mine, and the data summary module is used to summarize and process the data fed back by the power supply equipment module and the liquid supply equipment module, and feed it back to the control terminal to facilitate management personnel to understand the use of mine energy.

[0215] 9. Remote management module

[0216] Allows remote monitoring and control of the system so that timely actions can be taken when necessary. The remote management module includes a remote authority module, a remote monitoring module, a remote control module, and an operation log module. The remote authority module is used to authorize users to access the coal mine underground environmental information monitoring system through the Internet, the remote monitoring module is used to read relevant detection data and display it, the remote control module is used to achieve remote control of coal mine underground environmental equipment to improve mine safety and production efficiency, and the operation log module is used to record user operation behaviors to ensure the safety and controllability of the system.

[0217] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the scope of protection of the present invention is not limited to such specific statements and embodiments. For those skilled in the art, the present invention may have various changes and variations. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of the claims of the present invention.

Claims

1. A coal mine underground environment information monitoring system based on Internet of Things fusion communication, characterized in that: include: Sensor receiving module, data processing module, central control module and fault warning module; The sensor receiving module is used to receive data detected by various sensors installed downhole under the control of the central control module, and transmit the received data to the data processing module; The data processing module is used to pre-process the data transmitted by the sensor receiving module under the control of the central control module; The fault warning module, under the control of the central control module, detects the pre-processed data output by the data processing module and issues a warning for detected abnormal situations.

2. According to claim 1, a coal mine underground environment information monitoring system based on Internet of Things fusion communication is characterized in that: The fault warning module specifically includes an abnormality detection unit and an abnormality matching unit; The anomaly detection unit performs anomaly detection on the data preprocessed by the data processing module according to the threshold value corresponding to each data; The abnormal matching unit first extracts feature values ​​from the abnormal data output by the abnormal detection unit to form an abnormal feature vector to be matched; calculates the similarity between the abnormal feature vector to be matched and the abnormal feature vectors corresponding to each abnormal state in the historical record; and uses the abnormal state with the greatest similarity as the abnormal situation of the abnormal feature vector to be matched.

3. According to claim 2, a coal mine underground environment information monitoring system based on Internet of Things fusion communication is characterized in that: The data received by the sensor receiving module include gas concentration, temperature and humidity, wind speed, wind direction, dust concentration, pressure change, water level and water seepage, and vibration; The elements in the abnormal feature vector to be matched and the abnormal feature vector corresponding to each abnormal state in the historical record include: gas concentration, temperature and humidity, wind speed, wind direction, dust concentration, pressure change, water level and water seepage, and vibration.

4. The coal mine underground environment information monitoring system based on Internet of Things fusion communication according to claim 3 is characterized in that: The similarity calculation formula is: Where D(x,y) represents the similarity value between two abnormal vectors x and y; i and i Respectively represent the value of the i-th feature in the two abnormal vectors; w i is the adaptive weight assigned to the i-th feature; M i The maximum possible value of a feature used to indicate normalization.

5. The coal mine underground environment information monitoring system based on Internet of Things fusion communication according to claim 4 is characterized in that: w i The calculation formula is: w i = Middle weight × (1 + abnormal proportion × 0.3) Among them, the intermediate weight = reference weight × position weighting coefficient; the reference weight is set according to the security level corresponding to each feature, and the position weighting coefficient is set according to the sensor installation position corresponding to each feature; the anomaly ratio is the percentage of anomalies of the feature in historical data.

6. The coal mine underground environment information monitoring system based on Internet of Things fusion communication according to claim 5 is characterized in that: Gas concentration is a primary safety feature, and the corresponding baseline weight is 0.4; Temperature, dust concentration, pressure change, water level and seepage, and vibration are secondary safety features, and the corresponding benchmark weight is 0.3; Humidity, wind speed, and wind direction are three-level safety features, and the corresponding baseline weight is 0.

2.

7. The coal mine underground environment information monitoring system based on Internet of Things fusion communication according to claim 6 is characterized in that: The sensor installation locations are divided into: coal mining face, return air lane, air intake lane, and other areas; the weighted coefficient of the position corresponding to the coal mining face is 1.4, the weighted coefficient of the position corresponding to the return air lane is 1.2, the weighted coefficient of the position corresponding to the air intake lane is 1.1, and the weighted coefficient of the position corresponding to other areas is 0.9.

Images