A mine safety management system

Abstract

This invention belongs to the field of mining engineering and addresses the problems of large data transmission volumes, incomplete risk assessments, and delayed early warnings in traditional mine safety management systems. It proposes a new mine safety management system. The data acquisition module collects multi-dimensional raw data, including mine environmental parameters and geomechanical data, through sensors. The edge computing module deploys nodes in the acquisition area to locally clean, denoise, and extract features from the raw data, generating secondary data which is then transmitted via the information transmission module. The data processing module incorporates a risk emergence prediction component using multimodal fusion and dynamic Bayesian networks. It integrates multimodal features and dynamically simulates risk evolution to calculate risk levels. The data response module pushes early warning information to personnel at different levels through multiple channels based on the risk level. This system reduces data transmission volume, improves the comprehensiveness of risk assessment, and enables proactive risk prediction, meeting the needs of real-time mine safety monitoring and efficient management.

Description

Technical Field

[0001] This invention belongs to the field of mining engineering and relates to a mine safety management system. Background Technology

[0002] Mining production is characterized by high risks and significant management challenges. Scientific and effective management of mine-related information is a crucial element in achieving safe production. Since the latter half of the 20th century, developed countries in Europe and America have vigorously promoted mining automation and informatization, implementing production control and safety monitoring at all stages of production and safety, both on the surface and underground. They have also adopted computer network technology for comprehensive control of various subsystems, thereby improving production efficiency and the level of mine safety management. Mine safety data comes from complex sources, such as environmental parameters, geomechanics, equipment status, personnel behavior, and video images. Traditional management systems often transmit all raw data directly to remote cloud / central servers for processing, resulting in large data transmission volumes and high latency. This is especially problematic in mines, which are often located in remote areas with limited network bandwidth. Furthermore, the raw data contains a lot of noise and redundant information and has not undergone local preprocessing, further slowing down subsequent analysis and failing to meet the needs of real-time mine safety monitoring. Traditional solutions often employ independent analysis models for different types of data, such as numerical gas concentration, visual equipment malfunction footage, and behavioral violations by personnel. For example, they may monitor gas concentration or view equipment videos separately without establishing a logical association between the multimodal data. For instance, they may only detect excessive gas concentrations without considering factors such as whether personnel are in hazardous areas or whether ventilation equipment is malfunctioning. This results in a one-sided risk assessment that may overlook complex safety hazards. Mine safety risks often have the characteristics of multiple factors emerging simultaneously, such as geological subsidence, equipment vibration, and aging of support structures, which gradually accumulate into collapse risks. Traditional systems mostly rely on threshold-triggered early warnings, which are post-event responses. Moreover, they do not use dynamic models to capture the evolution of risks over time, making it impossible to predict in advance how risks will accumulate from minor to major hidden dangers, resulting in delayed early warnings and missed intervention opportunities. Summary of the Invention

[0003] To address the problems existing in the background technology, this application proposes a mine safety management system. To achieve the above objectives, the technical solution adopted in this application is as follows: A mine safety management system, comprising: The data acquisition module acquires raw data from different areas of the mine through sensors; The edge computing module deploys edge computing nodes in each data acquisition area to perform preliminary cleaning, noise reduction, and feature extraction on the original data to obtain the second data. The data processing module has a built-in risk emergence prediction component that combines multimodal fusion and dynamic Bayesian networks to calculate the risk level within the mine based on the second data. The data response module pushes data to managers and on-site workers at different levels through various channels, including PC, App, broadcast, SMS, and audible and visual alarms, based on the risk level. Furthermore, the raw data includes environmental parameters, geomechanics, equipment status, personnel behavior, and videos and images of the mining area within the mine. Furthermore, the raw data obtained by the data processing module is directly submitted to the edge computing module for preprocessing to obtain the second data. Furthermore, the risk emergence prediction component with built-in multimodal fusion and dynamic Bayesian network includes: Multimodal feature extraction and fusion components; Dynamic Bayesian network components; Early warning generation component. Furthermore, the dynamic Bayesian network construction and risk simulation include: Nodes are defined as risk state variables, and nodes are connected by directed edges to represent causal relationships. The influence of the state at time t on the state at time t+1 is inferred through a Bayesian network containing time slices. Furthermore, the second data from the edge computing module is transmitted to the data processing module via the information transmission module. Compared with the prior art, this application has the following beneficial effects: This application deploys nodes in the data acquisition area using an edge computing module to perform localized cleaning, noise reduction, and feature extraction on the raw data, significantly reducing the amount of data transmission, avoiding the network bandwidth limitations in remote mines, and avoiding the efficiency loss of remote servers processing redundant data, thus ensuring real-time response to subsequent risk analysis. This application utilizes a multimodal feature extraction and fusion component in the data processing module to convert different types of secondary data into a unified feature vector and establish correlation logic. Compared with traditional independent analysis, multimodal fusion can capture complex risks that cannot be reflected by single data, making risk assessment more comprehensive and avoiding the omission of security risks due to data fragmentation. This application sets up a dynamic Bayesian network component that can simulate the evolution of risks over time based on the multimodal characteristics of time series. Compared with traditional threshold alarms, this component realizes the transformation from post-event response to pre-event prediction, and can provide early warning of major hidden dangers several hours or even days in advance, leaving sufficient time for mine intervention. Attached Figure Description

[0004] Figure 1This is a system architecture diagram of this application. The mine safety management system shown in the diagram mainly consists of five core modules: a data acquisition module, an edge computing module, a data processing module, a data response module, and an information transmission module. Detailed Implementation

[0005] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application. Example 1: like Figure 1 As shown, this application provides a mine safety management system, including: The data acquisition module uses sensors to acquire raw data from different areas of the mine. The edge computing module deploys edge computing nodes in each data acquisition area to perform preliminary cleaning, noise reduction, and feature extraction on the original data to obtain the second data. The data processing module has a built-in risk emergence prediction component that combines multimodal fusion and dynamic Bayesian networks to calculate the risk level within the mine based on the second data. The data response module pushes data to managers and on-site workers at different levels through various channels, including PC, App, broadcast, SMS, and audible and visual alarms, based on the risk level. The data acquisition module collects raw data from the mine through sensors deployed in different areas, covering five core dimensions: environmental parameters, such as underground gas concentration, dust content, temperature, humidity, and ventilation volume; geomechanical data, such as roadway surrounding rock stress, strata displacement, and surface subsidence; equipment status data, such as the operating temperature, vibration frequency, and fault codes of mining equipment (drilling rigs, crushers), and the speed and load of transportation equipment (mine cars); personnel behavior data, such as the location of workers, the wearing status of safety helmets, and whether they have entered dangerous areas without authorization; and video and image data, acquiring real-time video and static images of roadways, working faces, and shaft entrances through monitoring cameras within the mine. This raw data coverage of five scenarios—environment, geology, equipment, personnel, and vision—avoids the limitations of single data dimensions and provides a complete data foundation for subsequent risk assessment. Simultaneously, the sensors collect data in real time, ensuring timely detection of dynamic changes within the mine, such as sudden increases in gas concentration or abnormal equipment vibration, allowing time for risk warnings. The edge computing module deploys edge computing nodes in each data acquisition area, such as the underground working face, wellhead machine room, and transportation roadway, to receive raw data transmitted from the corresponding area's sensors and perform preprocessing operations. Outliers, such as extreme values ​​caused by sensor malfunctions and missing values, are removed from the raw data to ensure accuracy. Environmental parameters, such as gas concentration fluctuations, and data containing interference from equipment vibration are filtered to retain valid signals. Key features, such as peak frequencies of equipment vibration and characteristic values ​​of personnel violations, are extracted from the preprocessed data to generate structured secondary data. Edge nodes perform data preprocessing locally, transmitting only structured secondary data rather than massive amounts of raw data. This reduces bandwidth consumption within the mine and avoids data congestion. Local preprocessing shortens the time lag between data acquisition and analysis, making it particularly suitable for real-time-critical scenarios such as underground mines, preventing missed risk assessments due to data transmission delays. Cleaning and noise reduction remove invalid interference, and feature extraction focuses on key information, providing high-quality input for subsequent risk calculations and reducing the probability of misjudgments. After preprocessing, the second data is transmitted to the data processing module through the information transmission module, such as through mining-specific industrial Ethernet, 5G, or other means. The data processing module incorporates a risk emergence prediction component using multimodal fusion and a dynamic Bayesian network. This component comprises three sub-modules, calculating the risk level within the mine based on secondary data. The multimodal feature extraction and fusion component further integrates features from different modalities in the secondary data, such as associating environmental features of excessive methane concentration with behavioral features of personnel illegally entering methane-affected areas. Through weighted fusion using an attention mechanism, a comprehensive feature vector reflecting the synergistic influence of multiple factors is generated. The dynamic Bayesian network component uses the comprehensive feature vector as input to construct a dynamic Bayesian network model. This model can capture the temporal patterns of risk changes, such as inferring increasing roadway collapse risk through continuously rising surrounding rock stress, and calculating real-time risk levels for different areas of the mine (e.g., working faces, explosives depots) through probabilistic reasoning. The early warning generation component generates differentiated early warning information based on the threshold corresponding to the risk level and transmits this information to the data response module. The construction and risk inference of the dynamic Bayesian network includes defining nodes as risk state variables. Nodes are not simple sensor data, but abstract risk state variables. Examples include: "roof stability," "gas outburst trend," "ventilation effectiveness," "equipment overheating risk," and "personnel gathering risk." Each node is a hidden state, and its value (e.g., "low," "medium," "high") needs to be inferred from observed data (i.e., S(t)). Nodes are connected by directed edges, representing causal relationships. This combines domain knowledge (such as "roof instability leads to abnormal gas outbursts") with data-driven approaches (verifying and quantifying the strength of these causal relationships using historical data). For example: Roof stability (t) -> Gas outburst trend (t+Δt). The essence of a DBN is a Bayesian network that includes time slices, where the state at time t affects the state at time t+1. This simulates the dynamic evolution of risk. Multimodal fusion avoids misjudgments of risk caused by single data points, such as relying solely on gas concentration for risk reasoning. It can identify complex risks resulting from the combined effects of environment, personnel, and equipment, such as the risk generated by insufficient ventilation due to equipment failure, gas accumulation, and personnel violations. Dynamic Bayesian networks consider the evolution of risks over time and can predict risk trends in advance, such as predicting that gas concentration will exceed the standard 1-2 hours in advance, rather than simply reacting passively to anomalies that have already occurred. Based on the risk level output by the data processing module, the data response module pushes early warning information to personnel at different levels through multiple channels to achieve precise risk prediction. This ensures that all personnel are aware of the emergency situation; medium-risk information is pushed to team leaders and safety administrators; low-risk information is only pushed to safety administrators to avoid information overload. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A mine safety management system, characterized in that, include: The data acquisition module is used to acquire raw data from different areas of the mine through sensors; An edge computing module is deployed at edge computing nodes within each data acquisition area to perform preliminary cleaning, noise reduction, and feature extraction on the original data to generate second data. The data processing module has a built-in risk emergence prediction component based on multimodal fusion and dynamic Bayesian network, which is used to calculate the risk level within the mine based on the second data; The data response module is used to push early warning information to managers and on-site workers at different levels through multiple communication channels based on the risk level.

2. The mine safety management system according to claim 1, characterized in that, The raw data includes at least one of the following: environmental parameters within the mine, geomechanical data, equipment status data, personnel behavior data, video and image data.

3. The mine safety management system according to claim 1, characterized in that, The edge computing module preprocesses the raw data locally, including outlier removal, filtering and noise reduction, feature extraction, and spatiotemporal alignment of multimodal data.

4. The mine safety management system according to claim 1, characterized in that, The risk emergence prediction component includes: A multimodal feature extraction and fusion component is used to convert different types of secondary data into a unified feature vector and fuse them. Dynamic Bayesian network components are used to simulate risk evolution processes based on time series data. The early warning generation component is used to generate differentiated early warning information based on the risk level.

5. The mine safety management system according to claim 4, characterized in that, In the dynamic Bayesian network component, nodes are defined as risk state variables, and causal relationships are represented by directed edges between nodes. The influence of the state at time t on the state at time t+1 is inferred through a Bayesian network containing time slices.

6. The mine safety management system according to claim 1, characterized in that, The second data generated by the edge computing module is transmitted to the data processing module through the information transmission module, which includes an industrial Ethernet or 5G communication network.

7. The mine safety management system according to claim 1, characterized in that, The communication channels of the data response module include at least one of PC, mobile App, broadcast, SMS and sound and light alarm, and are pushed to the corresponding level of personnel according to the risk level.

Images