Mine high temperature heat disaster monitoring system based on multi-modal internet of things data

Abstract

The application relates to the technical field of mine safety monitoring, and discloses a mine high-temperature heat disaster monitoring system based on multi-modal Internet of Things data, which comprises a prediction control integrated system architecture module, a thermal field flow field joint optimizer module and a self-adaptive anti-interference control mechanism module.

Description

Technical Field

[0001] This invention relates to the field of mine safety monitoring technology, and more specifically, to a mine high-temperature heat hazard monitoring system based on multimodal Internet of Things data. Background Technology

[0002] With the increasing depth of mining and the growing complexity of mine structures, the problem of high-temperature heat hazards in mines is becoming increasingly prominent, seriously threatening miners' health and production safety. Therefore, mine heat hazard monitoring and control technology has become an important research direction in the field of mine safety production. Currently, mine heat hazard monitoring systems generally adopt a "prediction-control" separation architecture, that is, environmental parameters are collected through a sensing layer, the development trend of heat hazards is analyzed using a prediction system, and then the control system formulates control measures such as ventilation based on the prediction results.

[0003] However, existing technologies still have many shortcomings in practical applications, mainly in the following aspects:

[0004] The existing system suffers from low information flow efficiency between the prediction system and the control system. Because they belong to different functional modules, delays and losses easily occur during information transmission, causing control decisions to lag behind the development of thermal hazards and making it difficult to respond promptly to rapid changes in the mine environment. There is a contradiction between prediction accuracy and control timeliness. High-precision thermal hazard prediction typically relies on complex models and substantial computational resources, resulting in a lengthy prediction process. Conversely, the control system requires rapid response to mitigate thermal hazard risks. This contradiction makes it difficult for the system to balance prediction accuracy and real-time control. Furthermore, prediction results in existing systems often cannot directly guide control decisions. The thermal hazard development trend output by the prediction system usually requires manual analysis and decision transformation to formulate specific control strategies. This process is not only time-consuming but also... Furthermore, subjective judgment errors can easily be introduced, affecting control effectiveness. Existing technologies lack adaptive coordination mechanisms, with the predictive and control systems optimizing independently without a unified optimization objective or collaborative adjustment mechanism. This prevents adaptive optimization based on the overall system performance, impacting overall system performance. Traditional mine thermal hazard monitoring systems have complex structures, typically comprising a five-layer architecture of "sensing-analysis-prediction-decision-control." Too many layers lead to high system complexity and maintenance costs, and frequent interface compatibility issues between layers affect system stability and scalability. Existing systems also have poor anti-interference capabilities. When faced with emergencies such as roadway collapses or ventilation equipment failures, the system lacks rapid response and automatic adjustment capabilities, easily leading to thermal hazard monitoring failure and posing significant safety hazards.

[0005] In summary, existing mine heat hazard monitoring and control technologies have problems that urgently need improvement in terms of information flow, prediction and control coordination, adaptive optimization, system structure simplification, and anti-interference capabilities. A new type of mine heat hazard monitoring system that can achieve deep coupling between heat hazard prediction and ventilation control and has efficient coordination and adaptive capabilities is needed. Summary of the Invention

[0006] This invention provides a mine high-temperature heat hazard monitoring system based on multimodal Internet of Things data, which solves the technical problems of delayed heat hazard response, complex system architecture, and poor anti-interference ability caused by the separation of prediction and control in related technologies.

[0007] This invention provides a mine high-temperature heat hazard monitoring system based on multimodal Internet of Things data, comprising:

[0008] Predictive control integrated system architecture module: Construct a predictive control integrated system architecture and establish a three-layer PCIS architecture consisting of a perception layer, a prediction layer, and a control layer;

[0009] The thermal and flow field joint optimizer module constructs a thermal and flow field joint optimizer based on the multimodal data output from the sensing layer, realizing the direct coupling between the prediction model and the control algorithm;

[0010] The adaptive disturbance rejection control mechanism module implements an adaptive disturbance rejection control mechanism based on the thermal field prediction results generated by the prediction layer and the feedback information during the execution process of the control layer. It achieves system fault tolerance through a fault mode library and dynamic reconfiguration.

[0011] The performance self-optimization module implements system performance self-optimization based on historical data and real-time performance feedback accumulated during system operation, and continuously improves system performance through incremental learning and knowledge transfer.

[0012] Furthermore, the construction of the integrated predictive control system architecture includes:

[0013] A multimodal data sensing layer is constructed, and temperature sensor arrays, airflow sensors, air pressure difference sensors, humidity sensors and gas composition sensors are deployed. Edge computing units are used for data preprocessing and anomaly detection, and convergence processing units are used to realize spatiotemporal alignment and quality assessment of multimodal data.

[0014] Construct a thermal hazard prediction layer and establish a mine thermal hazard prediction model based on a spatiotemporal graph convolutional network;

[0015] Construct a ventilation control layer, optimize ventilation parameters based on prediction results, and implement control strategies.

[0016] Furthermore, the construction of the joint thermal and flow field optimizer includes:

[0017] A thermal field-flow field mapping model is constructed, and a two-way mapping relationship between temperature distribution and airflow distribution is established based on the principles of computational fluid dynamics.

[0018] Construct a response matrix to describe the degree of influence of changes in control variables on the thermal and flow fields;

[0019] Construct a unified predictive control framework to directly map thermal field prediction results into ventilation control parameters;

[0020] It enables the generation of control strategies across multiple time scales, generating short-term, medium-term, and long-term control strategies respectively.

[0021] Furthermore, the implementation of the adaptive disturbance rejection control mechanism includes:

[0022] Build a fault mode library to store the characteristic patterns of common fault types and predefined coping strategies;

[0023] Implement a dynamic control reconfiguration algorithm to quickly reconfigure the control strategy when a system anomaly is detected;

[0024] A self-calibration mechanism for the prediction model is constructed to automatically adjust the model parameters by comparing the prediction results with the actual measurement data.

[0025] Implement an incremental learning module to continuously learn from runtime data and optimize system performance.

[0026] Furthermore, the implementation of system performance self-optimization includes:

[0027] Construct a performance evaluation index system, including prediction accuracy index, control effect index, and system stability index;

[0028] To achieve incremental learning and parameter optimization, the prediction model and control algorithm are updated regularly based on newly collected data;

[0029] Build a knowledge accumulation and transfer learning mechanism to quickly adapt to new environments and tasks by leveraging historical experience.

[0030] Furthermore, the spatiotemporal graph convolutional network of the heat damage prediction layer includes:

[0031] The spatial graph convolution module processes the spatial topology of the mine and extracts spatial features; the temporal convolution module uses causal convolution to process time-series data and preserve temporal relationships; the spatiotemporal attention mechanism focuses on important spatial regions and key time points respectively; and the multi-scale feature extraction module captures multi-time-scale patterns through different convolution kernel sizes.

[0032] Furthermore, the thermal field and flow field mapping model is based on a simplified computational fluid dynamics model, which discretizes the mine space into a node network. The thermal field and flow field are represented as node temperature values ​​and edge flow values, respectively, and mass conservation constraints, energy conservation constraints, and physical boundary constraints are established.

[0033] Furthermore, the predictive control unified framework employs a differential dynamic programming algorithm to construct a multi-objective optimization function, comprehensively considering heat hazard reduction effect, energy consumption, and ventilation balance to generate the optimal ventilation control strategy.

[0034] Furthermore, the dynamic control reconfiguration algorithm includes:

[0035] Fault detection and diagnosis: Real-time monitoring of system operating status and detection of potential faults;

[0036] Control constraint reconfiguration: Reconfigure control constraint conditions based on fault conditions;

[0037] The control strategy is reconfigured in real time, and the control strategy is re-optimized based on the updated system model and constraints.

[0038] The present invention provides a computer storage medium for executing the above-mentioned mine high temperature heat hazard monitoring system based on multimodal Internet of Things data, including a memory and one or more processors. The memory stores executable code, and when the one or more processors execute the executable code, they implement the above-mentioned mine high temperature heat hazard monitoring system based on multimodal Internet of Things data.

[0039] The beneficial effects of this invention are as follows: by using multimodal data fusion and spatiotemporal graph convolutional networks, the accuracy of heat hazard prediction is improved and the prediction time range is expanded, providing more sufficient response time for heat hazard prevention and control.

[0040] The integrated predictive control architecture shortens the entire process time from data acquisition to control execution, reducing system response time from minutes to seconds in traditional systems, enabling rapid response to heat hazards;

[0041] The system hierarchy has been simplified from the traditional five layers of "sensing-analysis-prediction-decision-control" to three layers of "sensing-prediction-control". The calculation process has been changed from serial to parallel, which has improved the system's operating efficiency and reduced maintenance costs.

[0042] When faced with emergencies (such as tunnel collapse, ventilation equipment failure, etc.), the system can quickly and automatically reconfigure the control strategy, maintain most of the system stability, and outperform the performance of traditional systems in fault conditions.

[0043] This system can predict heat disasters in advance and take preventive measures, which improves the efficiency of heat disaster reduction and reduces the incidence and impact of heat disaster accidents.

[0044] Through multi-objective optimization and predictive control, the system can allocate ventilation resources more rationally and reduce the energy consumption of the ventilation system while ensuring safety, thus achieving the dual goals of safety and energy saving.

[0045] Through incremental learning and self-optimization mechanisms, the system continuously accumulates experience during operation, and the prediction model and control strategy are continuously optimized. The system performance improves as the running time increases, and the overall system performance is improved compared to the initial state after three months. Attached Figure Description

[0046] Figure 1 This is a flowchart of the mine high-temperature heat hazard monitoring system based on multimodal Internet of Things data in this invention;

[0047] Figure 2 This is a bar chart comparing the prediction accuracy and response time of the system of this invention with those of the traditional system;

[0048] Figure 3 This is a line graph showing the expansion of the prediction time range of the system of this invention compared to traditional systems;

[0049] Figure 4 This is a pie chart showing the distribution of the system optimization effects of this invention;

[0050] Figure 5 This is a radar chart comparing the performance of the system of this invention with that of a traditional system in five key performance indicators. Detailed Implementation

[0051] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, some features described in the examples may be combined in other examples.

[0052] At least one embodiment of the present invention discloses a mine high-temperature heat hazard monitoring system based on multimodal Internet of Things data, such as... Figure 1 As shown, it includes:

[0053] Predictive control integrated system architecture module: Construct a predictive control integrated system architecture and establish a three-layer PCIS architecture consisting of a perception layer, a prediction layer, and a control layer;

[0054] According to embodiments of this application, a three-layer predictive control integrated architecture is constructed, including a sensing layer, a prediction layer, and a control layer, to achieve deep coupling between heat hazard prediction and ventilation control. Specifically, it includes:

[0055] Step 1.1: Construct a multimodal data perception layer;

[0056] A multimodal sensor network is deployed in the mining environment to collect environmental parameter data such as temperature, humidity, airflow velocity, air pressure, and gas composition. The sensors include:

[0057] Temperature sensor array: Temperature sensors are deployed at key nodes in the mine working face, roadway and ventilation system to collect spatial temperature distribution data;

[0058] Airflow sensors: Install airflow speed and direction sensors at key locations along the ventilation path to monitor the operating status of the ventilation system;

[0059] Differential pressure sensor: Install differential pressure sensors between nodes in the ventilation network to monitor changes in the resistance of the ventilation system;

[0060] Humidity sensor: Monitors air humidity in mines, serving as an important parameter for assessing heat hazards;

[0061] Gas composition sensor: monitors the concentration of gases such as CO, CO2, and CH4, and is used for heat hazard identification and safety monitoring.

[0062] Optionally, in some implementations, the temperature sensor can employ a distributed fiber optic temperature sensing system to achieve continuous temperature monitoring of long-distance tunnels. Furthermore, in high-risk areas, infrared thermal imagers can be added to provide more intuitive visualization of heat distribution.

[0063] The perception layer adopts a hierarchical data processing architecture, including edge computing units and aggregation processing units:

[0064] Edge computing unit: Located on sensor node or sensor gateway, it is responsible for data preprocessing, anomaly detection and preliminary feature extraction to reduce data transmission volume;

[0065] Convergence processing unit: Integrates multimodal data, performs spatiotemporal alignment, data completion, and quality assessment, providing high-quality input data for the prediction layer.

[0066] It should be understood that in some implementations, the edge computing unit may be implemented using a low-power ARM architecture processor or a dedicated ASIC chip to meet the low power consumption requirements and explosion-proof requirements of mines.

[0067] Step 1.2: Construct a heat hazard prediction layer;

[0068] According to embodiments of this application, a mine thermal hazard prediction model is constructed based on a spatiotemporal graph convolutional network to achieve accurate prediction of the spatiotemporal dynamic changes of the thermal field.

[0069] like Figure 2As shown, this system improves prediction accuracy by 40% compared to traditional systems, and reduces response time from minutes (approximately 180 seconds) to seconds (approximately 5 seconds), achieving rapid response throughout the entire process from data acquisition to control execution. This performance improvement is mainly attributed to the innovative design of the integrated predictive control architecture, which eliminates the information transmission delay between the prediction and control modules in traditional systems. The comparative data in the figure demonstrates that by employing a spatiotemporal graph convolutional network and a deeply coupled predictive control mechanism, the system achieves a 36-fold increase in response speed while maintaining high prediction accuracy, providing strong technical support for the timely prevention and control of mine thermal hazards.

[0070] like Figure 3 As shown, the prediction time range is expanded. Traditional systems predict within 10–15 minutes (average 12.5 minutes), while this system's prediction time range extends to 30–45 minutes (average 37.5 minutes), providing more sufficient response time for thermal hazard prevention and control. This three-fold expansion of the prediction time range is attributed to the deep learning capabilities of the spatiotemporal graph convolutional network for complex spatiotemporal patterns, as well as the information gain from multimodal data fusion. The longer prediction time window enables the system to achieve true preventative control, rather than a passive, reactive response, fundamentally improving the proactive protection capabilities of mine thermal hazard monitoring.

[0071] Specifically, it includes:

[0072] Mine spatial modeling: Constructing a mine spatial topology map based on 3D mine terrain data.

[0073] ;

[0074] in Represents the spatial topology of the mine; Represents a set of nodes, which represents spatial sampling points; This represents an edge set, indicating the physical connections between nodes.

[0075] Spatiotemporal graph convolutional network structure: A deep neural network containing temporal and spatial convolutional modules is constructed to capture the spatiotemporal patterns of thermal field evolution. The network structure includes:

[0076] Spatial graph convolutional layers: process spatial topological relationships and extract spatial features;

[0077] Temporal convolutional layers: process temporal data and capture patterns of temporal evolution;

[0078] Attention mechanism: Highlighting the impact of key areas and time points;

[0079] Multi-scale feature fusion: Integrating features from different spatiotemporal scales to improve prediction accuracy.

[0080] The spatiotemporal graph convolutional network in this application is a deep learning model specifically designed for the task of predicting thermal hazards in mines. Its specific implementation is as follows:

[0081] Input layer: Receives multimodal sensor data in the form of a time-series data structure.

[0082] ;

[0083] in Represents the input data tensor; Represents the real number field; This represents the number of sensor nodes; For feature dimensions (such as temperature, humidity, airflow, etc.); The length of the historical time window;

[0084] Spatial Graph Convolution Module: This module uses graph convolution algorithms to process the spatial topology of the mine. The calculation formula is as follows:

[0085] The spatial graph convolution module in this application is used to handle information transfer in the spatial topology of a mine. The module's processing is as follows: First, the system obtains the node feature matrix of the current layer, which contains the feature information of each spatial sampling point. Then, the system constructs an enhanced adjacency matrix, which ensures that node information is not lost during transfer by adding self-connections (i.e., connections between nodes and themselves). Next, the system calculates the normalized form of the adjacency matrix by multiplying the adjacency matrix by the inverse of the square root of the angle matrix. This step ensures that information is not subject to scale bias due to different numbers of node connections during transfer. Subsequently, the system multiplies the normalized adjacency matrix with the node feature matrix of the current layer, realizing feature transfer across the spatial topology. Finally, the system multiplies the transferred feature matrix with a learnable weight matrix and processes it through an activation function to obtain the node feature representation of the next layer. This process achieves effective transfer and fusion of spatial information across the mine topology.

[0086] Temporal convolution module: Uses causal convolution to process temporal data, avoiding information leakage and preserving temporal relationships. The calculation formula is:

[0087] The temporal convolution module in this application uses a causal convolution method to process time-series data, ensuring that the model does not use information from future time points during prediction. The module's processing procedure is as follows: For each time point in the input sequence... The system calculates the output value at that time point, specifically by taking the time point... and its predecessor 1 time point (of which) The input values ​​(of kernel size) are multiplied by the corresponding convolutional filter parameters and then summed. This processing ensures that the output at each time point depends only on the current and past inputs, without using future information, thus maintaining the causality of time series predictions. By selecting convolutional kernels of different sizes, the system can capture patterns at different time scales, enhancing its ability to extract time series features.

[0088] Spatiotemporal attention mechanism: Introducing a dual attention mechanism, focusing on important spatial regions and key time points respectively:

[0089] The spatiotemporal attention mechanism in this application comprises two parts: spatial attention and temporal attention.

[0090] Spatial attention mechanisms are used to evaluate the importance relationships between different spatial nodes. The specific implementation process is as follows: First, the system calculates the relevance score between each pair of nodes, which is achieved by inputting the node feature vectors into a relevance evaluation function; then, the system applies a softmax function to normalize all relevance scores, ensuring that the sum of all attention weights is 1, thus obtaining the node... For nodes Attention weights are applied. This mechanism enables the system to automatically identify and focus on spatially influential node pairs, improving the accuracy of heat hazard propagation path prediction.

[0091] The time attention mechanism is used to identify critical moments in a time series. The specific implementation process is as follows: First, the system inputs the feature vector of each time point into a time importance evaluation function to obtain the importance score for that time point; then, the system applies a softmax function to normalize the importance scores of all time points to obtain the attention weight for each time point. This mechanism enables the system to automatically identify and highlight key change points in a time series, improving the early warning capability for sudden heat disasters.

[0092] By combining attention mechanisms in both spatial and temporal dimensions, the system can accurately capture the spatiotemporal patterns of thermal hazard evolution in complex mine environments, enabling more accurate predictions.

[0093] Multi-scale feature extraction: By using parallel convolutional kernels of different sizes (e.g., 3×1, 5×1, 7×1), patterns at different time scales are captured, and features from different layers are fused through skip connections to enhance the model's ability to perceive multi-scale spatiotemporal patterns; these different sizes of convolutional kernels correspond to different k_c values;

[0094] Output layer: Generating the future The thermal field distribution prediction for each time step is as follows:

[0095] ;

[0096] in , , These represent the first, second, and third time steps in the future, respectively. Predicted values ​​of thermal field distribution at each sampling point in the mine space at each time step. This represents the predicted number of future time steps.

[0097] In the scenario of monitoring thermal hazards in mines, a specific application example of this spatiotemporal graph convolutional network is as follows:

[0098] Early identification of hotspot areas: The system can identify areas with abnormally rising temperature trends by analyzing historical data from a temperature sensor network. For example, in one application, the model successfully predicted that the temperature of a working surface would exceed a safe threshold in 35 minutes, triggering cooling measures in advance.

[0099] Heat hazard propagation path prediction: Based on spatial topology and airflow data, the system can predict possible propagation paths of heat hazards. In a coal seam spontaneous combustion event, the system accurately predicted the direction of heat flow diffusion, guiding the ventilation system to adjust in a timely manner and preventing the heat flow from spreading to densely populated areas.

[0100] Dynamic operating condition adaptation: During the advancement of the mining face, the system can automatically adjust the prediction model according to changes in the topology. For example, when a new roadway is completed, the system completes the learning of the thermal flow field distribution characteristics of the new ventilation network within a few hours, maintaining prediction accuracy.

[0101] Optionally, in some implementations, the spatiotemporal graph convolutional network can adopt a hierarchical training strategy, first pre-training with global data and then fine-tuning with historical data from a specific mine, in order to improve the prediction accuracy of the model in a specific environment.

[0102] Step 1.3, construct the ventilation control layer;

[0103] According to embodiments of this application, a ventilation optimization control system is constructed based on thermal field prediction results to achieve precise control of ventilation parameters. Specifically, this includes:

[0104] Ventilation network modeling: Establish a mine ventilation network model, including the characteristic parameters and constraints of components such as fans, air doors, and roadways;

[0105] Ventilation optimization objective function: Construct a multi-objective optimization function that comprehensively considers heat hazard reduction, energy consumption, and ventilation balance.

[0106] The ventilation optimization objective in this application is to optimize ventilation within the predicted timeframe (from the current moment). To the future time, The function that minimizes the total cost (representing the predicted number of future time steps) The total cost function consists of three parts:

[0107] First, for each prediction time (from 0 to ... ), calculate the thermal damage state deviation cost, which is the weighted deviation between the thermal damage state vector obtained after inputting the predicted temperature and control parameters into the deviation function and the ideal state;

[0108] Secondly, calculate the control cost, which is the weighted norm of the control vector (including parameters such as fan speed and damper opening), reflecting energy consumption;

[0109] Finally, the cost of control change is calculated, which is the weighted norm of the control vector change between adjacent time steps, reflecting the control stationarity.

[0110] The total cost function is obtained by multiplying each of the three cost components by the corresponding weight matrix and then summing the results. The system achieves optimal ventilation control by finding the control sequence that minimizes this total cost.

[0111] The predicted temperature is represented by Always looking towards the future Temperature prediction at any time, among which For the prediction time step index, the value range is: arrive The control vector includes parameters such as fan speed and damper opening, and the weight matrix... , and These are respectively used to balance the importance of three objectives: heat damage reduction, energy consumption, and control stability. (Function) The function representing the deviation between temperature and control parameters is implemented as follows: First, the difference between the predicted temperature and the target safe temperature is calculated. Then, the heat hazard reduction effect is evaluated in conjunction with the control parameters, and a heat hazard state vector is output. This function maps the temperature prediction results and control parameters to the heat hazard state space, enabling the system to directly optimize the heat hazard reduction effect.

[0112] Control strategy generation: Differential Dynamic Programming (DDP) is used to solve the above optimization problem to generate a ventilation control strategy.

[0113] The control strategy generation in this application employs a differential dynamic programming algorithm, which obtains the control increment through iterative optimization calculation. Specifically, at each control moment, the system first obtains the deviation between the current state and the target state. Then multiply the deviation by the feedback gain matrix. The system obtains the state-dependent control adjustment; simultaneously, it also calculates a state-independent feedforward control term. This is used to compensate for the known dynamic characteristics of the system. Finally, the two control variables are added together to obtain the final control increment, which is used to adjust ventilation control parameters such as fan speed and damper opening. The feedforward term is mainly pre-calculated based on the system model and predicted trajectory, while the feedback gain matrix is ​​obtained by solving the locally linearized model of the system. This allows for real-time correction of state deviations and improves the system's robustness to disturbances.

[0114] Optionally, in some implementations, control policy generation can employ Model Predictive Control (MPC) instead of differential dynamic programming, especially when deployed on computationally limited edge devices, where MPC offers better real-time performance. MPC is based on the principle of rolling time-domain optimization. In each control cycle, the system uses the current state as initial conditions to solve for the optimal control sequence within a finite time domain, but only executes the first control action, repeating this process in the next control cycle. This method has low computational complexity and can explicitly handle control constraints, making it suitable for applications in resource-constrained environments.

[0115] Control execution unit: The optimized control parameters are sent to the fan frequency converter, automatic damper and other execution equipment to realize the automatic control of the ventilation system.

[0116] The thermal and flow field joint optimizer module constructs a thermal and flow field joint optimizer based on the multimodal data output from the sensing layer, realizing the direct coupling between the prediction model and the control algorithm;

[0117] According to the embodiments of this application, a thermal field-flow field joint optimizer is constructed, which directly maps thermal field prediction to ventilation control strategy, realizing deep coupling between prediction and control. This is the core innovative module of this system.

[0118] like Figure 4 As shown, the distribution of system optimization effects demonstrates the comprehensive advantages of the proposed technical solution: a 40% improvement in heat hazard reduction, a 35% increase in ventilation efficiency, and a 25% reduction in energy consumption, comprehensively enhancing the overall performance of the mine heat hazard monitoring system. This multi-dimensional performance improvement fully reflects the core value of the thermal-flow field co-optimizer, achieving a comprehensive leap in system performance by unifying physical field prediction and control strategy generation within the same optimization framework. The data in the figure indicates that this co-optimizer not only improves heat hazard control but also achieves energy-saving operation of the ventilation system, providing an economical and efficient technical solution for safe mine production.

[0119] Step 2.1: Construct a mapping model between the thermal field and the flow field;

[0120] Based on the principles of Computational Fluid Dynamics (CFD), a two-way mapping model between the thermal field and the flow field is constructed to achieve a unified representation of the two physical fields. Specifically, this includes:

[0121] Heat transfer equation: describes the propagation of heat in the mine space, considering three heat transfer modes: convection, conduction and radiation;

[0122] Fluid dynamics equations: describe the flow of air in the mine space, taking into account viscosity, pressure gradient and boundary conditions;

[0123] Thermo-fluid coupling model: Establishes the interaction between the thermal field and the flow field, including the effect of temperature on fluid density and the effect of airflow on heat transfer.

[0124] The joint thermal-flow field optimizer in this application is the core innovative part of the system, and its specific implementation is as follows:

[0125] Simplified CFD Model: To achieve real-time computation, a simplified computational fluid dynamics model is adopted, reducing the complex set of partial differential equations to a node network model. For the heat transfer equation, the following is used:

[0126] The heat transfer model in this application, based on the principle of energy conservation, describes the propagation process of heat in a mine space. The model considers four key factors: first, the rate of change of temperature over time, representing the temperature change at a point per unit time; second, convective heat transfer, describing the phenomenon of heat transfer with airflow, represented by the product of air density, specific heat capacity, airflow velocity vector, and temperature gradient; third, thermal conduction, describing the diffusion of heat from high-temperature regions to low-temperature regions, represented by the thermal conductivity coefficient and the divergence of the temperature gradient; and finally, the heat source term, representing the areas in the space that generate or absorb heat, such as equipment heating or spontaneous combustion of coal seams. By comprehensively considering these four factors, the system can accurately simulate the propagation law of heat in the complex environment of a mine, providing a physical basis for heat hazard prediction.

[0127] Networked representation: The mine space is discretized into a network of nodes, where each node represents a region in the space, and the connections between nodes represent physical connectivity. The thermal and flow fields are represented by the node temperatures in the network representation. and the flow value on the edge .

[0128] Response matrix construction: Construct the control response matrix through offline simulation or online learning. This describes the degree of influence of changes in control variables (such as fan speed, damper opening, etc.) on the thermal and flow fields, in the form of:

[0129] ;

[0130] in, This represents a vector representing the temperature change at each node. This represents the vector of changes in flow on each side. This represents the control response matrix, which describes the coefficients that describe the impact of changes in control variables on the system state. This represents the adjustment vector of control variables (such as fan speed, damper opening, etc.). This formula establishes a linear mapping relationship between control adjustment and system response, and is the mathematical basis for joint optimization of the thermal field and the flow field.

[0131] The response matrix construction in this application aims to establish a mapping relationship between control variable adjustments and system state changes. The system first defines a matrix where rows represent system state variables (including temperature changes at each node and flow rate changes at each edge), and columns represent control variables (such as fan speed and damper opening). Each element in the matrix represents the magnitude of the impact of a unit change in the corresponding control variable on the corresponding state variable. Through this matrix representation, the system can predict the combined impact of any combination of control adjustments on the entire thermal and flow fields, thereby supporting optimal decision-making for control strategies. This response matrix can be obtained through offline numerical simulation or continuously updated through online learning during system operation to adapt to dynamic changes in the mine environment.

[0132] Thermal-fluid coupling constraints: Establishing coupling constraints between the thermal field and the flow field to ensure the consistency of the two physical fields. This mainly includes:

[0133] Mass conservation constraint: For each non-source / sink node (i.e., a node that is neither an air inlet nor an air outlet) in a mine ventilation network, the algebraic sum of the flow rates on all adjacent connections of that node must equal 0. This constraint ensures that no gas is created or destroyed in the system, reflecting the fundamental physical law of mass conservation. In practical applications, this constraint is used to ensure the rationality of airflow distribution in the ventilation network and is a basic condition for ventilation control optimization.

[0134] Energy conservation constraint: For each node in a mine ventilation network, the total energy entering that node (including heat flowing in from adjacent nodes and the node's own heat source) must equal the total energy flowing out of that node. Specifically, the system calculates the sum of the heat flowing in from each adjacent node (flow rate multiplied by specific heat capacity and then by the adjacent node's temperature) and the node's own heat source; this value should equal the sum of the heat flowing out (flow rate multiplied by specific heat capacity and then by the node's own temperature). This constraint ensures energy conservation in the system and is the physical basis for predicting thermal field evolution.

[0135] Physical boundary constraints: Physical quantities in the system, such as temperature, flow rate, and pressure, must meet certain boundary conditions. For example, the temperature cannot be lower than the minimum ambient temperature, the flow rate cannot exceed the maximum allowable value of the equipment, and the pressure must be within the range that the equipment can withstand. These constraints reflect the actual limitations of the physical system and ensure the feasibility of the control strategy.

[0136] The following is a specific application example of this thermal field-flow field co-optimizer in the scenario of monitoring thermal hazards in mines:

[0137] Localized heat hazard mitigation: When an abnormal temperature rise is detected at a working face, the system can quickly calculate the optimal local ventilation adjustment scheme. For example, in a coal mining machine overheating event, the system automatically increased the air volume in that area and adjusted the opening of the dampers in the upper and lower airways, resulting in a 3.5°C reduction in the working face temperature within 8 minutes, thus avoiding production shutdown.

[0138] Global ventilation optimization: Based on predicted heat hazard development trends, the system can optimize the configuration of the global ventilation network. For example, during the high-temperature period in summer, the system automatically plans a "differentiated day and night ventilation scheme," increasing airflow in key areas during the high-temperature period of the day and reducing airflow in non-working areas at night, reducing energy consumption by 25% while ensuring safety.

[0139] Emergency ventilation plan generation: When an abnormal heat source (such as the spontaneous combustion point of coal seams) is detected, the system can generate an emergency ventilation plan in real time. In a simulation test, the system calculated the optimal damper control plan that isolates the heat source and protects the evacuation route in just 2.3 seconds, which is more than 80% faster than manual decision-making.

[0140] Optionally, in some implementations, the thermal-flow field co-optimizer can integrate an expert knowledge base, combining data-driven methods with domain expert experience to improve the reliability of the system's decision-making in special situations.

[0141] Step 2.2, construct a unified prediction-control framework;

[0142] According to embodiments of this application, a framework is constructed to unify thermal field prediction and ventilation control under a single optimization objective, achieving seamless integration of prediction results and control decisions. Specifically, this includes:

[0143] State-space representation: The thermal field prediction results and ventilation control parameters They are uniformly represented as variables in the state space;

[0144] Predictive-control joint optimization: Based on the predicted future thermal field state, ventilation control parameters are directly optimized, skipping the traditional intermediate decision-making process;

[0145] Feedback correction mechanism: Automatically adjust the prediction model and control strategy based on the deviation between the actual control effect and the expected effect.

[0146] Step 2.3: Generate multi-timescale control strategies;

[0147] According to embodiments of this application, a hierarchical control strategy is generated based on prediction results at different time scales, achieving a unification of short-term precise control and long-term optimization planning. Specifically, this includes:

[0148] Short-term control strategy (1-5 minutes): Based on recent thermal field predictions, optimize local ventilation parameters to achieve rapid response to heat hazards;

[0149] Mid-term control strategy (5-30 minutes): Based on the mid-term thermal field evolution trend, adjust the regional ventilation layout to achieve preventive control of heat hazards;

[0150] Long-term control strategy (30 minutes or more): Based on the long-term heat hazard development trend, plan the overall ventilation resource allocation to maximize energy efficiency.

[0151] The adaptive disturbance rejection control mechanism module implements an adaptive disturbance rejection control mechanism based on the thermal field prediction results generated by the prediction layer and the feedback information during the execution process of the control layer. It achieves system fault tolerance through a fault mode library and dynamic reconfiguration.

[0152] According to embodiments of this application, an adaptive disturbance rejection control mechanism is constructed to improve the robustness and responsiveness of the system in the face of sudden events.

[0153] Step 3.1: Construct a failure mode library and corresponding response strategies;

[0154] Based on historical data and expert knowledge, a database of common failure modes and corresponding coping strategies is constructed to provide a knowledge base for responding to emergencies. Specifically, this includes:

[0155] Fault mode recognition: including feature patterns of common fault types such as sensor failure, fan failure, damper failure, and tunnel collapse;

[0156] Predefined response strategies: Basic response strategies for different types of faults, such as sensor backup switching and ventilation path reconstruction.

[0157] Optionally, in some implementations, the fault mode library can be constructed in the form of a knowledge graph, containing multi-dimensional information such as fault type, characteristic manifestation, severity, typical scenarios, and countermeasures, to improve the accuracy of fault diagnosis and the pertinence of countermeasures.

[0158] Step 3.2: Implement the dynamic control reconfiguration algorithm;

[0159] According to embodiments of this application, a dynamic control reconfiguration algorithm is employed to quickly reconfigure the control strategy when a system anomaly or fault is detected, ensuring continuous system operation. Specifically, this includes:

[0160] Fault detection and diagnosis: Real-time monitoring of system operating status, detection of potential faults and location diagnosis;

[0161] Control constraint reconfiguration: Based on the fault conditions, reconfigure the control constraints to avoid the faulty component;

[0162] Real-time reconfiguration of control strategy: Based on the updated system model and constraints, the control strategy is re-optimized.

[0163] Step 3.3: Construct a self-calibration mechanism for the prediction model;

[0164] According to embodiments of this application, a self-calibration mechanism for the prediction model is employed to continuously optimize the accuracy of the prediction model by comparing the prediction results with actual measurement data. Specifically, this includes:

[0165] Prediction error analysis: Calculate the deviation between the predicted results and the actual measured values, and analyze the error distribution characteristics;

[0166] Adaptive adjustment of model parameters: Based on error analysis results, the parameters of the prediction model are automatically adjusted;

[0167] Dynamic optimization of model structure: Adjust the model structure as needed to adapt to changes in the environment.

[0168] Optionally, in some implementations, the predictive model self-calibration can employ a Bayesian learning framework, which not only updates the point estimates of the model parameters but also updates the probability distribution of the parameters, thereby improving the robustness of the model in uncertain environments.

[0169] The performance self-optimization module implements system performance self-optimization based on historical data and real-time performance feedback accumulated during system operation, and continuously improves system performance through incremental learning and knowledge transfer;

[0170] According to the embodiments of this application, a system performance self-optimization mechanism is constructed to continuously improve the overall system performance through continuous learning and feedback adjustments.

[0171] Step 4.1, construct a performance evaluation index system;

[0172] Establish a comprehensive performance evaluation index system to quantify the system's performance in heat hazard prediction and control. Specifically, this includes:

[0173] Prediction accuracy metrics include: mean absolute error (MAE), root mean square error (RMSE), etc.

[0174] Control performance indicators: such as temperature control accuracy, response time, energy efficiency, etc.

[0175] System stability indicators: such as anti-interference capability, fault recovery time, etc.

[0176] Step 4.2: Implement incremental learning and parameter optimization;

[0177] According to embodiments of this application, an incremental learning mechanism is employed, enabling the system to continuously learn from operational data and constantly optimize model parameters and control strategies. Specifically, this includes:

[0178] Online data collection: Continuously collect system operation data, including sensor data, control parameters, performance indicators, etc.

[0179] Incremental model update: Based on newly collected data, the predictive model and control algorithm are updated periodically;

[0180] Automatic parameter tuning: Using algorithms such as Bayesian optimization, the system's key parameters are automatically tuned.

[0181] Optionally, in some implementations, the incremental learning process can adopt a federated learning framework, using edge computing nodes in various areas of the mine as learning clients and a central server as a model aggregation node, thereby achieving mine-wide model optimization while protecting data privacy.

[0182] Step 4.3: Construct a knowledge accumulation and transfer learning mechanism;

[0183] According to embodiments of this application, a knowledge accumulation and transfer learning mechanism is employed, enabling the system to leverage historical experience and quickly adapt to new environments or tasks. Specifically, this includes:

[0184] Knowledge base construction: Storing the experiences and patterns acquired during system operation into the knowledge base;

[0185] Transfer learning algorithms: leverage knowledge accumulated in similar environments to accelerate model training and optimization in new environments;

[0186] Model generalization enhancement: Improve the model's adaptability and generalization ability in different mining environments.

[0187] like Figure 5As shown, this system outperforms traditional systems in key performance indicators such as prediction accuracy, response time, energy efficiency, prediction time range, and system stability, demonstrating the technological advantages of the integrated predictive control architecture. The radar chart clearly illustrates the comprehensive superiority of this system across five core dimensions, with the most significant improvements in prediction accuracy and response time, reaching 40% and 36 times the improvement, respectively. This comprehensive performance enhancement is not simply the result of parameter optimization, but stems from fundamental innovation at the system architecture level. By deeply integrating prediction and control functions, it achieves a synergistic effect of 1+1>2, setting a new benchmark for the development of mine thermal hazard monitoring technology.

[0188] A computer storage medium for executing the above-described mine high-temperature heat hazard monitoring system based on multimodal Internet of Things (IoT) data includes a memory and one or more processors. The memory stores executable code, and when the one or more processors execute the executable code, they implement the above-described mine high-temperature heat hazard monitoring system based on multimodal IoT data.

[0189] The embodiments of the present invention have been described above. However, the embodiments are not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make more equivalent embodiments under the guidance of the present embodiments, and all of them are within the protection scope of the present embodiments.

Claims

1. A mine high temperature heat damage monitoring system based on multi-modal Internet of Things data, characterized in that, The method comprises the following steps: a prediction control integrated system architecture module is constructed to build a prediction control integrated system architecture, and a three-layer PCIS architecture of a perception layer, a prediction layer and a control layer is established; the construction of the prediction control integrated system architecture comprises: a multi-modal data perception layer is constructed, temperature sensor arrays, air flow sensors, air pressure difference sensors, humidity sensors and gas component sensors are deployed, edge computing units are used for data preprocessing and anomaly detection, and a convergence processing unit is used to realize the spatio-temporal alignment and quality evaluation of multi-modal data; a heat disaster prediction layer is constructed, and a mine heat disaster prediction model is established based on a spatio-temporal graph convolution network; a ventilation control layer is constructed, and ventilation parameters are optimized and control strategies are executed according to the prediction results; the spatio-temporal graph convolution network of the heat disaster prediction layer comprises: a spatial graph convolution module for processing mine spatial topological relations and extracting spatial features; a time convolution module for processing time series data and maintaining time sequence relations by using causal convolution; a spatio-temporal attention mechanism for focusing on important spatial regions and key time points; and a multi-scale feature extraction module for capturing multi-time scale patterns through different convolution kernel sizes; a thermal field and flow field joint optimizer module is constructed based on the multi-modal data output by the perception layer to build a thermal field and flow field joint optimizer, so as to directly couple the prediction model and the control algorithm; an adaptive anti-disturbance control mechanism module is constructed based on the thermal field prediction results generated by the prediction layer and the feedback information in the execution process of the control layer, so as to implement an adaptive anti-disturbance control mechanism, realize system fault tolerance through a fault mode library and dynamic reconstruction, and continuously improve system performance through incremental learning and knowledge transfer. a performance self-optimization module is constructed based on historical data accumulated during system operation and real-time performance feedback, so as to implement system performance self-optimization and continuously improve system performance through incremental learning and knowledge transfer.

2. The mine high temperature heat disaster monitoring system based on multi-modal Internet of Things data according to claim 1, characterized in that, The construction of the thermal field and flow field joint optimizer comprises: a thermal field and flow field mapping model is constructed to establish a bidirectional mapping relationship between temperature distribution and air flow distribution based on computational fluid dynamics principles; a response matrix is constructed to describe the influence degree of control variable changes on the thermal field and the flow field; a prediction control unified framework is constructed to directly map the thermal field prediction results to ventilation control parameters; multi-time scale control strategy generation is realized to generate short-term, medium-term and long-term control strategies respectively.

3. The mine high temperature heat hazard monitoring system based on multi-modal Internet of Things data according to claim 1, characterized in that, The adaptive anti-disturbance control mechanism comprises: a fault mode library is constructed to store feature modes and pre-defined response strategies of common fault types; a dynamic control reconstruction algorithm is implemented to quickly reconstruct the control strategy when system anomalies are detected; a prediction model self-calibration mechanism is constructed to automatically adjust model parameters by comparing prediction results with actual measurement data; an incremental learning module is implemented to continuously learn and optimize system performance from operation data.

4. The mine high temperature heat harm monitoring system based on multi-modal Internet of Things data according to claim 1, characterized in that, The implementation of system performance self-optimization comprises: a performance evaluation index system is constructed, including prediction accuracy indicators, control effect indicators and system stability indicators; incremental learning and parameter optimization are realized to regularly update the prediction model and the control algorithm based on newly collected data; a knowledge accumulation and transfer learning mechanism is constructed to quickly adapt to new environments and new tasks using historical experience.

5. The mine high temperature heat hazard monitoring system based on multi-modal Internet of Things data according to claim 2, characterized in that, The heat field and flow field mapping model discretizes the mine space into a network of nodes based on a simplified computational fluid dynamics model, and represents the heat field and flow field as node temperature values and flow values on edges, respectively, and establishes mass conservation constraints, energy conservation constraints, and physical boundary constraints.

6. The mine high temperature heat harm monitoring system based on multi-modal Internet of Things data according to claim 2, characterized in that, The predictive control unified framework adopts a differential dynamic programming algorithm, constructs a multi-objective optimization function, comprehensively considers heat damage reduction effect, energy consumption, and ventilation balance, and generates an optimal ventilation control strategy.

7. The mine high temperature heat harm monitoring system based on multi-modal Internet of Things data according to claim 3, characterized in that, The dynamic control reconstruction algorithm includes: Fault detection and diagnosis, real-time monitoring of system operation state and detection of potential faults; Control constraint reconfiguration, reconfiguration of control constraint conditions according to fault conditions; Control strategy real-time reconstruction, re-optimization of the control strategy based on the updated system model and constraint conditions.

8. A computer storage medium, characterized in that, The mine high-temperature heat damage monitoring system based on multi-modal Internet of Things data comprises a memory and one or more processors, the memory stores executable code, and the one or more processors execute the executable code to implement the mine high-temperature heat damage monitoring system based on multi-modal Internet of Things data according to any one of claims 1-7.

Images