Dynamic regulation and control intelligent fire preventing and extinguishing system and method for mine goaf fire

Abstract

The invention discloses a mine goaf fire dynamic regulation and control intelligent fire preventing and extinguishing system and method, and relates to the technical field of intelligent fire preventing and extinguishing, and the method comprises the steps: collecting an original monitoring data stream, transmitting the original monitoring data stream to an edge computing gateway for preprocessing, and generating a standardized data packet; the standardized data packet is input into a ground central data processing module for data fusion analysis and dynamic threshold calculation, and goaf state panoramic information and a dynamic threshold vector are generated; and inputting the goaf state panoramic information into the GBDT-TCN hybrid machine learning model for fire risk prediction, and generating a prediction instruction with a risk level and a time window. The intelligent control unit executes multi-means cooperative regulation and control and feeds back data in real time, and finally evaluates the regulation and control effect, so that full-chain closed-loop management from data acquisition, risk prediction, simulation optimization to intelligent regulation and control is realized, and the accuracy and timeliness of goaf fire prevention and control are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent fire prevention and extinguishing technology, and in particular to an intelligent fire prevention and extinguishing system and method for dynamic control of fires in mine goaf areas. Background Technology

[0002] Spontaneous combustion fire prevention in mine goaf areas is a major technical challenge in coal mine safety production. The industry's commonly used monitoring and early warning technologies are mainly based on fixed threshold alarm mechanisms. These involve deploying devices such as CO sensors and temperature sensors to collect environmental parameters, triggering an alarm signal when the monitored data exceeds a preset threshold. In recent years, with the development of Internet of Things (IoT) technology, existing technologies have enabled remote transmission and centralized display of monitoring data. Some advanced systems have also introduced time-series data analysis algorithms for short-term trend extrapolation, attempting to improve the timeliness of early warnings.

[0003] However, the above-mentioned technical methods still have significant limitations in dealing with the highly nonlinear and multi-field coupled evolution characteristics of goaf fires. Existing systems generally lack the ability to deeply integrate multi-source heterogeneous monitoring data, making it difficult to construct a panoramic view of the state reflecting the seepage-oxidation-heat transfer coupling mechanism inside the goaf, resulting in insufficient sensitivity of early warning models in identifying early oxidation signs. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides an intelligent fire prevention and extinguishing method for dynamic control of fires in mine goaf areas, which solves the problems of difficulty in constructing a panoramic view of the seepage-oxidation-heat transfer coupling state in goaf areas and insufficient sensitivity of early fire warning caused by the lack of deep fusion capability of multi-source data in existing technologies.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides an intelligent fire prevention and extinguishing method for dynamic control of fires in mine goaf areas, which includes: collecting raw monitoring data streams and transmitting the raw monitoring data streams to an edge computing gateway for preprocessing to generate standardized data packets; Standardized data packets are input into the ground central data processing module for data fusion analysis and dynamic threshold calculation to generate panoramic information on the status of the goaf and dynamic threshold vectors. The panoramic information on the status of the goaf is input into the GBDT-TCN hybrid machine learning model for fire risk prediction, generating prediction instructions with risk levels and time windows. The prediction instructions with risk levels and time windows are input into the digital twin to perform multi-plan simulation and decision optimization, and generate the optimal control strategy. The optimal control strategy is input into the intelligent control unit to execute multi-means coordinated control, generate real-time control actions, and evaluate the control effect based on the feedback data of the real-time control actions.

[0007] As a preferred embodiment of the intelligent fire prevention and extinguishing method for dynamic control of fires in mine goaf areas as described in this invention, the method includes the following steps: collecting raw monitoring data streams and transmitting the raw monitoring data streams to an edge computing gateway for preprocessing to generate standardized data packets. Based on the multi-dimensional sensor network, raw monitoring data streams of temperature, gas concentration, and vibration signals are continuously collected at a set sampling frequency. The multi-dimensional sensor network transmits the raw monitoring data streams with complete spatiotemporal markers to the edge computing gateway deployed on the side of the roadway through the intrinsically safe industrial ring network for mining. The edge computing gateway performs data cleaning, noise reduction filtering, and format standardization operations on the received raw monitoring data stream with spatiotemporal markers.

[0008] As a preferred embodiment of the intelligent fire prevention and extinguishing method for dynamic control of fires in mine goaf areas as described in this invention, the method includes the following steps: Standardized data packets are input into a ground-based central data processing module for data fusion analysis and dynamic threshold calculation to generate panoramic information on the goaf status and a dynamic threshold vector. The ground-based central data processing module receives standardized data packets from the edge computing gateway, decodes the standardized data packets, and extracts the spatiotemporal marker information and monitoring values ​​contained therein. The ground-based central data processing module performs spatiotemporal registration and coordinate unification of multi-source heterogeneous standardized data packets based on the extracted spatiotemporal marker information, and generates joint distribution data of temperature, gas and vibration in the goaf through Kalman filtering algorithm. The ground central data processing module calls the built-in dynamic threshold algorithm library. The dynamic threshold algorithm library combines the real-time mining advance speed and ventilation network parameters to calculate the temperature-gas-vibration joint distribution data of the goaf, dynamically update the concentration threshold of each area, and generate a dynamic threshold vector of multi-parameter threshold limits. The ground-based central data processing module integrates the combined distribution data of temperature, gas, and vibration in the goaf with the dynamic threshold vector and outputs it as panoramic information of the goaf status and the dynamic threshold vector.

[0009] As a preferred embodiment of the intelligent fire prevention and extinguishing method for dynamic control of fires in mine goaf areas as described in this invention, the method involves: inputting panoramic information of the goaf status into a GBDT-TCN hybrid machine learning model for fire risk prediction, and generating prediction instructions with risk levels and time windows, including the following steps: Based on the GBDT-TCN hybrid machine learning model, panoramic information of the goaf status is received. The temperature field distribution, gas concentration field distribution and vibration signal features contained in the panoramic information of the goaf status are standardized and normalized to generate standardized feature vectors. The GBDT component of the GBDT-TCN hybrid machine learning model performs deep feature selection and combination mining on the standardized feature vectors to extract key physical features from the standardized feature vectors. The TCN component based on the GBDT-TCN hybrid machine learning model receives high-dimensional feature combinations, uses its temporal convolutional network architecture to analyze the long-range temporal dependencies and nonlinear dynamic evolution trends in the high-dimensional feature combinations, and outputs prediction results for future time series. Based on the GBDT-TCN hybrid machine learning model, the prediction results of future time series output by the TCN component are probabilistically calculated and time window analyzed to generate quantitative fire risk level indicators and critical time windows.

[0010] As a preferred embodiment of the intelligent fire prevention and extinguishing method for dynamic control of fires in mine goaf areas as described in this invention, the method includes the following steps: inputting prediction instructions with risk levels and time windows into a digital twin for multi-plan simulation and decision optimization to generate the optimal control strategy. The digital twin receives prediction instructions with risk levels and time windows, analyzes the fire occurrence probability, critical time windows and potential fire source spatial location parameters contained in the prediction instructions with risk levels and time windows, and maps them into the multiphysics coupling simulation environment of the digital twin. The digital twin automatically generates multiple combinations of control plans based on the mapped fire risk parameters. These combinations of control plans include different disaster response methods and parameter configurations. The digital twin performs multi-physics coupling simulation and deduction for each plan in the combination of control plans, calculates the evolution process of the temperature field, oxygen concentration field and gas diffusion state of the goaf in the future time after the execution of each plan, and outputs the simulation result dataset of each plan. The digital twin performs a comprehensive performance evaluation and comparison of the simulation results dataset of all contingency plans through an objective function, which includes indicators such as cooling rate, resource consumption cost, and ventilation disturbance coefficient.

[0011] As a preferred embodiment of the intelligent fire prevention and extinguishing method for dynamic control of fires in mine goaf areas as described in this invention, the method involves inputting the optimal control strategy into the intelligent control unit to execute multi-means coordinated control and generate real-time control actions, including the following steps: The intelligent control unit receives the optimal control strategy generated by the digital twin and analyzes the specific execution instructions and device control parameters in the optimal control strategy; The intelligent control unit converts the parsed equipment control parameters into industrial control signals, and sends the industrial control signals to the corresponding execution equipment through the intrinsically safe control network for mining. The execution equipment includes the nitrogen injection unit frequency converter, the grouting pump motor controller, and the ventilation damper electric actuator. The intelligent control unit monitors the feedback status signal of the execution device and the real-time monitoring data stream of the multi-dimensional sensor network. It compares the feedback status signal and the real-time monitoring data stream with the expected effect curve of the optimal control strategy in real time, and dynamically adjusts the output parameters of the industrial control signal according to the deviation value. The intelligent control unit continuously outputs adjusted industrial control signals to drive the execution equipment to complete collaborative operations and generate real-time control actions that record the equipment's operating status and parameter adjustments.

[0012] As a preferred embodiment of the intelligent fire prevention and extinguishing method for dynamic control of fires in mine goaf areas as described in this invention, the method includes the following steps: Evaluating the control effect based on feedback data from real-time control actions. The effect evaluation unit receives real-time control actions generated by the intelligent control unit, extracts equipment operating status parameters and monitoring data sequences from the real-time control actions, and forms a control process dataset. The effect evaluation unit performs key indicator analysis on the control process dataset, including cooling rate, oxygen concentration decay slope, and carbon monoxide removal efficiency. The effect evaluation unit compares and analyzes the calculated key indicators with the expected effects of the optimal control strategy predicted by the digital twin, and generates a control effect evaluation report on indicator deviation and achievement degree. The effect evaluation unit serves as a reference value for determining the disaster elimination conditions in the control effect evaluation report. The disaster elimination conditions are that the carbon monoxide concentration is below 5 ppm for three consecutive hours and the temperature gradient returns to the baseline range, and the control is completed.

[0013] Secondly, the present invention provides an intelligent fire prevention and extinguishing system for dynamic control of fires in mine goaf areas, including a data acquisition module that acquires raw monitoring data streams and transmits the raw monitoring data streams to an edge computing gateway for preprocessing to generate standardized data packets. The analysis module inputs standardized data packets into the ground central data processing module for data fusion analysis and dynamic threshold calculation, generating panoramic information on the status of the goaf and dynamic threshold vectors. The prediction module inputs panoramic information on the status of the goaf into the GBDT-TCN hybrid machine learning model to predict fire risk and generate prediction instructions with risk level and time window. The control module inputs the prediction instructions with risk level and time window into the digital twin to perform multi-plan simulation and decision optimization, and generate the optimal control strategy. The evaluation module inputs the optimal control strategy into the intelligent control unit to execute multi-means coordinated control, generates real-time control actions, and evaluates the control effect based on the feedback data of the real-time control actions.

[0014] Thirdly, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, it implements any step of the intelligent fire prevention and extinguishing method for dynamic control of fire in the goaf area of ​​the mine as described in the first aspect of the present invention.

[0015] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the intelligent fire prevention and extinguishing method for dynamic control of fires in mine goaf areas as described in the first aspect of the present invention.

[0016] The beneficial effects of this invention are as follows: Raw monitoring data such as temperature, gas concentration, and vibration signals are collected through a multi-dimensional sensor network and preprocessed by an edge computing gateway to generate standardized data packets; the ground-based central data processing module fuses and analyzes the data and calculates dynamic thresholds to generate panoramic information and dynamic threshold vectors of the goaf state; a GBDT-TCN hybrid machine learning model is used for fire risk prediction, outputting prediction instructions with risk levels and time windows; multiple contingency plans are simulated and optimized based on a digital twin to generate the optimal control strategy; and the intelligent control unit executes multi-method collaborative control and provides real-time data feedback, ultimately evaluating the control effect. This achieves a closed-loop management system covering the entire chain from data collection, risk prediction, simulation optimization to intelligent control, significantly improving the accuracy and timeliness of goaf fire prevention and control. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart for a dynamic control and intelligent fire prevention and extinguishing method for fires in mine goaf areas.

[0019] Figure 2 This is a schematic diagram of an intelligent fire prevention and extinguishing system for dynamic control of fires in mine goaf areas. Detailed Implementation

[0020] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0021] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0022] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0023] Reference Figure 1 and Figure 2 As one embodiment of the present invention, this embodiment provides an intelligent fire prevention and extinguishing method for dynamic control of fires in mine goaf areas, comprising the following steps: S1. Collect the raw monitoring data stream and transmit it to the edge computing gateway for preprocessing to generate standardized data packets.

[0024] S1.1 Based on a multi-dimensional sensor network, raw monitoring data streams of temperature, gas concentration, and vibration signals are continuously collected at a set sampling frequency. The multi-dimensional sensor network transmits the raw monitoring data streams with complete spatiotemporal markers to the edge computing gateway deployed on the side of the roadway through an intrinsically safe industrial ring network for mining. Furthermore, the multi-dimensional sensor network continuously collects monitoring data streams of temperature, gas concentration, and vibration signals at a preset sampling frequency, with the monitoring data streams accompanied by precise spatiotemporal markers. The multi-dimensional sensor network transmits the monitoring data streams with spatiotemporal markers to the edge computing gateway deployed on the side of the roadway through an intrinsically safe industrial ring network for mining, ensuring data integrity and transmission reliability.

[0025] S1.2 The edge computing gateway performs data cleaning, noise reduction filtering and format standardization operations on the received raw monitoring data stream with spatiotemporal markers.

[0026] Furthermore, the edge computing gateway performs data cleaning operations on the received monitoring data stream with spatiotemporal markers to remove outliers and invalid records; applies noise reduction filtering algorithms to eliminate high-frequency interference and random fluctuations; and finally performs format standardization processing to unify the data structure and encoding method, generating standardized data packets that meet the requirements of subsequent processing.

[0027] S2. Input the standardized data packet into the ground central data processing module for data fusion analysis and dynamic threshold calculation to generate panoramic information on the status of the goaf and dynamic threshold vector.

[0028] S2.1 The ground central data processing module receives standardized data packets from the edge computing gateway, decodes the standardized data packets, and extracts the spatiotemporal marker information and monitoring values ​​contained therein.

[0029] Furthermore, the ground-based central data processing module receives standardized data packets from the edge computing gateway, performs binary decoding on the standardized data packets, and extracts the spatiotemporal marker information and monitoring values ​​embedded in the data packets. The spatiotemporal marker information includes the acquisition timestamp and three-dimensional spatial coordinates, and the monitoring values ​​include temperature, gas concentration, and vibration signal measurements.

[0030] S2.2 The ground central data processing module performs spatiotemporal registration and coordinate unification on the extracted spatiotemporal marker information, and generates joint distribution data of temperature, gas and vibration in the goaf by fusing the data through the Kalman filter algorithm.

[0031] Furthermore, the ground-based central data processing module performs spatiotemporal registration of the multi-source heterogeneous standardized data packets based on the extracted spatiotemporal marker information. It then uses a unified coordinate system transformation method to map all data to the global coordinate system of the goaf area. Finally, it uses a Kalman filter algorithm to fuse the registered multi-source data and generate joint distribution data of temperature, gas, and vibration in the goaf area.

[0032] S2.3 The ground central data processing module calls the built-in dynamic threshold algorithm library. The dynamic threshold algorithm library combines the real-time mining advance speed and ventilation network parameters to calculate the joint distribution data of temperature-gas-vibration in the goaf, dynamically update the concentration threshold of each area, and generate a dynamic threshold vector of multi-parameter threshold limits.

[0033] Specifically, the expression is: ; in, For the first Class monitoring parameters in time The dynamic threshold, For the first The baseline threshold for the monitoring parameters. This is the impact coefficient of mining. To ensure real-time mining site advancement speed, This represents the total air volume of the real-time ventilation network. The coefficient of thermal inertia. The change in temperature For the time change, The vibration coupling coefficient is... The standard deviation of the vibration signal. For time variables, For monitoring parameters.

[0034] Furthermore, the ground-based central data processing module calls the built-in dynamic threshold algorithm library. The dynamic threshold algorithm library combines the real-time mining advance speed and ventilation network parameters to calculate the joint distribution data of temperature-gas-vibration in the goaf. It adjusts the mining influence coefficient according to the real-time mining advance speed, updates the thermal inertia coefficient according to the total air volume of the ventilation network, calculates the vibration coupling coefficient through the standard deviation of the vibration signal, dynamically updates the concentration threshold of each area, and generates a dynamic threshold vector with multiple parameter threshold limits.

[0035] S2.4 The ground central data processing module integrates the combined distribution data of temperature, gas and vibration in the goaf with the dynamic threshold vector and outputs it as panoramic information of the goaf status and dynamic threshold vector.

[0036] Furthermore, the ground-based central data processing module structurally integrates the combined distribution data of temperature, gas, and vibration in the goaf with the dynamic threshold vector. It uses data encapsulation technology to package the two types of information into output data in a unified format, ultimately generating panoramic information of the goaf status and dynamic threshold vector containing complete monitoring data and corresponding threshold standards.

[0037] S3. Input the panoramic information of the goaf status into the GBDT-TCN hybrid machine learning model to predict fire risk and generate prediction instructions with risk level and time window.

[0038] S3.1. Based on the GBDT-TCN hybrid machine learning model, receive panoramic information on the status of the goaf, and perform data format standardization and normalization on the temperature field distribution, gas concentration field distribution and vibration signal characteristics contained in the panoramic information on the status of the goaf to generate standardized feature vectors.

[0039] Furthermore, the GBDT-TCN hybrid machine learning model receives panoramic information on the goaf status, performs data format standardization processing on the temperature field distribution, gas concentration field distribution, and vibration signal features contained in the panoramic information on the goaf status, uses the Z-score method to eliminate dimensional differences, and maps each feature value to the zero-one interval through Min-Max normalization to generate a standardized feature vector that meets the requirements of machine learning input.

[0040] S3.2. The GBDT component of the GBDT-TCN hybrid machine learning model performs deep feature selection and combination mining on the standardized feature vectors to extract key physical features from the standardized feature vectors.

[0041] Furthermore, the GBDT component of the GBDT-TCN hybrid machine learning model performs deep feature selection and combination mining on the standardized feature vectors. It calculates the feature importance score through the gradient boosting decision tree algorithm, and selects the temperature gradient, gas concentration change rate and vibration energy features that are most correlated with fire risk, generating high-dimensional feature combinations with clear physical meaning.

[0042] S3.3. The TCN component based on the GBDT-TCN hybrid machine learning model receives high-dimensional feature combinations, uses its temporal convolutional network architecture to analyze the long-range temporal dependencies and nonlinear dynamic evolution trends in the high-dimensional feature combinations, and outputs the prediction results of future time series.

[0043] Furthermore, the TCN component of the GBDT-TCN hybrid machine learning model receives high-dimensional feature combinations, uses the dilated causal convolutional layers in its temporal convolutional network architecture to capture multi-scale temporal patterns, maintains gradient flow through residual connections, analyzes the long-range temporal dependencies and nonlinear dynamic evolution trends in the high-dimensional feature combinations, and outputs predictions of temperature, gas concentration, and vibration state for multiple future time periods.

[0044] S3.4. Based on the GBDT-TCN hybrid machine learning model, perform probabilistic calculations and time window analysis on the prediction results of future time series output by the TCN component to generate quantitative fire risk level indicators and critical time windows.

[0045] Furthermore, the GBDT-TCN hybrid machine learning model performs probabilistic predictions on the future time series output by the TCN component, uses kernel density estimation to generate the probability distribution of fire occurrence, and analyzes the critical time window through an inflection point detection algorithm, ultimately outputting a quantified fire risk level index and a critical time window prediction instruction accurate to the minute.

[0046] S4. Input the forecast instructions with risk level and time window into the digital twin to perform multi-plan simulation and decision optimization, and generate the optimal control strategy.

[0047] S4.1 The digital twin receives prediction instructions with risk levels and time windows, parses the fire occurrence probability, critical time window and potential fire source spatial location parameters contained in the prediction instructions with risk levels and time windows, and maps them to the multiphysics coupling simulation environment of the digital twin.

[0048] Furthermore, the digital twin receives prediction commands with risk levels and time windows, analyzes the fire occurrence probability values, critical time window ranges, and three-dimensional coordinate parameters of potential fire sources contained in the prediction commands, and uses a spatial mapping algorithm to accurately map these parameters into the multiphysics coupled simulation environment of the digital twin, ensuring that the fire risk parameters and the spatial topology of the simulation environment are perfectly matched.

[0049] S4.2 The digital twin automatically generates multiple control plan combinations based on the mapped fire risk parameters. The control plan combinations include different disaster response methods and parameter configurations.

[0050] Furthermore, the digital twin automatically generates multiple control plan combinations based on the mapped fire risk parameters. These control plan combinations include disaster response methods and parameter configurations such as nitrogen injection flow gradient adjustment, grouting concentration ratio adjustment, and ventilation damper opening combinations. Each plan corresponds to a different set of equipment control parameters.

[0051] S4.3 The digital twin performs multi-physics coupling simulation on each plan in the control plan combination, calculates the evolution of the temperature field, oxygen concentration field and gas diffusion state of the goaf in future time after the execution of each plan, and outputs the simulation result dataset of each plan.

[0052] Specifically, the expression is: ; in, For the first Dataset of multiphysics simulation results corresponding to each control plan For temperature field, It is a spatial position vector. For time variables, For oxygen concentration field, For carbon monoxide concentration field, For the gas velocity field, The end time of the simulation. This serves as an index for the contingency plan.

[0053] Furthermore, the digital twin performs multi-physics coupling simulations on each plan in the control plan combination. By solving the temperature field conduction equation, oxygen concentration diffusion equation, and gas convection equation through a computational fluid dynamics model, it simulates the dynamic evolution of the temperature field distribution, oxygen concentration field gradient, and carbon monoxide gas diffusion state in the goaf over a future time period after each plan is executed, and outputs a dataset of simulation results containing spatiotemporal evolution data.

[0054] S4.4 The digital twin performs a comprehensive performance evaluation and comparison of the simulation results dataset of all contingency plans through an objective function, which includes indicators such as cooling rate, resource consumption cost, and ventilation disturbance coefficient.

[0055] Furthermore, the digital twin performs a comprehensive performance evaluation and comparison of the simulation results dataset of all contingency plans through an objective function. The objective function quantifies and calculates the cooling rate efficiency index, resource consumption cost index, and ventilation disturbance coefficient index. A weighted scoring method is used to comprehensively rank the various indicators, and finally the control strategy with the best comprehensive performance is selected.

[0056] S5. Input the optimal control strategy into the intelligent control unit to execute multi-means coordinated control and generate real-time control actions.

[0057] S5.1 The intelligent control unit receives the optimal control strategy generated by the digital twin and analyzes the specific execution instructions and device control parameters in the optimal control strategy.

[0058] Furthermore, the intelligent control unit receives the optimal control strategy generated by the digital twin, performs structured analysis on the optimal control strategy, extracts the specific execution instructions and equipment control parameters contained therein, such as nitrogen injection flow rate setpoint, grouting pressure parameters, and damper opening instructions, and converts these parameters into executable equipment control commands.

[0059] S5.2 The intelligent control unit converts the parsed equipment control parameters into industrial control signals, and sends the industrial control signals to the corresponding execution equipment through the intrinsically safe control network for mining. The execution equipment includes the nitrogen injection unit frequency converter, the grouting pump motor controller, and the ventilation damper electric actuator.

[0060] Furthermore, the intelligent control unit converts the parsed equipment control parameters into standard industrial control signals, including 4-20mA analog signals and Modbus RTU digital signals. These industrial control signals are then sent to the corresponding execution devices via an intrinsically safe mining control network. The execution devices include the nitrogen injection unit frequency converter, the grouting pump motor controller, and the ventilation damper electric actuator, ensuring that control commands are accurately transmitted to each execution terminal.

[0061] S5.3 The intelligent control unit monitors the feedback status signal of the execution device and the real-time monitoring data stream of the multi-dimensional sensor network. It compares the feedback status signal and the real-time monitoring data stream with the expected effect curve of the optimal control strategy in real time, and dynamically adjusts the output parameters of the industrial control signal according to the deviation value.

[0062] Furthermore, the intelligent control unit monitors the feedback status signal of the execution device and the real-time monitoring data stream of the multi-dimensional sensor network. It uses a real-time data comparison algorithm to compare the feedback status signal and the real-time monitoring data stream with the expected effect curve of the optimal control strategy in real time. It calculates the control error based on parameters such as temperature deviation and concentration deviation, and dynamically adjusts the output parameters of the industrial control signal based on the error value.

[0063] S5.4 The intelligent control unit continuously outputs adjusted industrial control signals to drive the execution equipment to complete collaborative operations and generate real-time control actions that record the equipment's operating status and parameter adjustments.

[0064] Furthermore, the intelligent control unit continuously outputs adjusted industrial control signals, and drives the nitrogen injection unit frequency converter, grouting pump motor controller and ventilation damper electric actuator to complete coordinated operations through PID control algorithm. It records the equipment operating status and parameter adjustment process in real time, and generates real-time control action records including timestamps, equipment status and control parameters.

[0065] S6. Evaluate the control effect based on feedback data from real-time control actions.

[0066] S6.1 The effect evaluation unit receives the real-time control actions generated by the intelligent control unit, extracts the equipment operating status parameters and the monitoring data sequence of the multi-dimensional sensor network from the real-time control actions, and forms a control process dataset.

[0067] Furthermore, the effect evaluation unit receives real-time control actions generated by the intelligent control unit, and extracts equipment operating status parameters such as nitrogen injection unit frequency, grouting pump pressure, and damper opening from the real-time control actions through data parsing technology. At the same time, it acquires monitoring data such as temperature sequence, oxygen concentration sequence, and carbon monoxide concentration sequence collected by multi-dimensional sensor network, and integrates these data by aligning them according to timestamps to form a structured control process dataset.

[0068] S6.2 The effect evaluation unit performs key indicator analysis on the control process dataset, including cooling rate, oxygen concentration decay slope, and carbon monoxide removal efficiency.

[0069] Furthermore, we effectively employ linear regression analysis to calculate the cooling rate index of temperature decrease per unit time, calculate the decay slope index of oxygen concentration decrease trend through exponential fitting, and use the mass conservation formula to calculate the ratio of carbon monoxide concentration reduction to initial amount as a removal efficiency index.

[0070] S6.3 The effect evaluation unit compares and analyzes the calculated key indicators with the expected effects of the optimal control strategy predicted by the digital twin, and generates a control effect evaluation report on the indicator deviation value and achievement degree.

[0071] Furthermore, a comparative analysis is conducted between key indicators such as cooling rate, oxygen concentration decay slope, and carbon monoxide removal efficiency and the expected effect curves of the optimal control strategy predicted by the digital twin. The absolute deviation value and relative achievement percentage are obtained, and a control effect evaluation report containing an indicator comparison table and achievement rating is generated.

[0072] S6.4 The effect evaluation unit serves as a reference value for determining the disaster elimination conditions in the control effect evaluation report. When the disaster elimination conditions are that the carbon monoxide concentration is below 5 ppm for three consecutive hours and the temperature gradient returns to the baseline range, the control completion command is output.

[0073] Furthermore, by continuously monitoring whether the carbon monoxide concentration data remains below the 5ppm concentration threshold for three consecutive hours, and simultaneously detecting whether the temperature gradient distribution has returned to the historical baseline fluctuation range, the system automatically outputs a control completion command to terminate the control process when both conditions are met.

[0074] This embodiment also provides an intelligent fire prevention and extinguishing system for dynamic control of fires in mine goaf areas, including: a data acquisition module, which acquires raw monitoring data streams and transmits the raw monitoring data streams to an edge computing gateway for preprocessing to generate standardized data packets; The analysis module inputs standardized data packets into the ground central data processing module for data fusion analysis and dynamic threshold calculation, generating panoramic information on the status of the goaf and dynamic threshold vectors. The prediction module inputs panoramic information on the status of the goaf into the GBDT-TCN hybrid machine learning model to predict fire risk and generate prediction instructions with risk level and time window. The control module inputs the prediction instructions with risk level and time window into the digital twin to perform multi-plan simulation and decision optimization, and generate the optimal control strategy. The evaluation module inputs the optimal control strategy into the intelligent control unit to execute multi-means coordinated control, generates real-time control actions, and evaluates the control effect based on the feedback data of the real-time control actions.

[0075] This embodiment also provides a computer device applicable to the intelligent fire prevention and extinguishing method for dynamic control of fires in mine goaf areas, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the intelligent fire prevention and extinguishing method for dynamic control of fires in mine goaf areas as proposed in the above embodiment.

[0076] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0077] This embodiment also provides a storage medium storing a computer program. When executed by a processor, the program implements the intelligent fire prevention and extinguishing method for dynamic control of fires in mine goaf areas as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0078] In summary, this invention utilizes a multi-dimensional sensor network to collect raw monitoring data such as temperature, gas concentration, and vibration signals. This data is then preprocessed by an edge computing gateway to generate standardized data packets. A central ground-based data processing module fuses and analyzes the data, calculating dynamic thresholds to generate panoramic information and dynamic threshold vectors for the goaf area. A GBDT-TCN hybrid machine learning model is employed for fire risk prediction, outputting prediction instructions with risk levels and time windows. Multi-plan simulation and decision optimization are performed based on a digital twin to generate the optimal control strategy. An intelligent control unit executes multi-method collaborative control and provides real-time data feedback, ultimately evaluating the control effect. This achieves a closed-loop management system across the entire chain, from data collection and risk prediction to simulation optimization and intelligent control, significantly improving the accuracy and timeliness of fire prevention and control in goaf areas.

[0079] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for dynamic control and intelligent fire prevention and extinguishing of fires in mine goaf areas, characterized in that: This includes collecting raw monitoring data streams and transmitting them to an edge computing gateway for preprocessing to generate standardized data packets; Standardized data packets are input into the ground central data processing module for data fusion analysis and dynamic threshold calculation to generate panoramic information on the status of the goaf and dynamic threshold vectors. The panoramic information on the status of the goaf is input into the GBDT-TCN hybrid machine learning model for fire risk prediction, generating prediction instructions with risk levels and time windows. The prediction instructions with risk levels and time windows are input into the digital twin to perform multi-plan simulation and decision optimization, and generate the optimal control strategy. The optimal control strategy is input into the intelligent control unit to execute multi-means coordinated control, generate real-time control actions, and evaluate the control effect based on the feedback data of the real-time control actions.

2. The intelligent fire prevention and extinguishing method for dynamic control of fires in mine goaf areas as described in claim 1, characterized in that: The process involves collecting raw monitoring data streams and transmitting them to an edge computing gateway for preprocessing to generate standardized data packets. This includes the following steps: Based on the multi-dimensional sensor network, raw monitoring data streams of temperature, gas concentration, and vibration signals are continuously collected at a set sampling frequency. The multi-dimensional sensor network transmits the raw monitoring data streams with complete spatiotemporal markers to the edge computing gateway deployed on the side of the roadway through the intrinsically safe industrial ring network for mining. The edge computing gateway performs data cleaning, noise reduction filtering, and format standardization operations on the received raw monitoring data stream with spatiotemporal markers.

3. The intelligent fire prevention and extinguishing method for dynamic control of fires in mine goaf areas as described in claim 2, characterized in that: The standardized data packets are input into the ground-based central data processing module for data fusion analysis and dynamic threshold calculation, generating panoramic information on the status of the goaf and dynamic threshold vectors, including the following steps: The ground-based central data processing module receives standardized data packets from the edge computing gateway, decodes the standardized data packets, and extracts the spatiotemporal marker information and monitoring values ​​contained therein. The ground-based central data processing module performs spatiotemporal registration and coordinate unification of multi-source heterogeneous standardized data packets based on the extracted spatiotemporal marker information, and generates joint distribution data of temperature, gas and vibration in the goaf through Kalman filtering algorithm. The ground central data processing module calls the built-in dynamic threshold algorithm library. The dynamic threshold algorithm library combines the real-time mining advance speed and ventilation network parameters to calculate the temperature-gas-vibration joint distribution data of the goaf, dynamically update the concentration threshold of each area, and generate a dynamic threshold vector of multi-parameter threshold limits. The ground-based central data processing module integrates the combined distribution data of temperature, gas, and vibration in the goaf with the dynamic threshold vector and outputs it as panoramic information of the goaf status and the dynamic threshold vector.

4. The intelligent fire prevention and extinguishing method for dynamic control of fires in mine goaf areas as described in claim 3, characterized in that: The panoramic information on the status of the goaf is input into the GBDT-TCN hybrid machine learning model for fire risk prediction, generating prediction instructions with risk levels and time windows, including the following steps: Based on the GBDT-TCN hybrid machine learning model, panoramic information of the goaf status is received. The temperature field distribution, gas concentration field distribution and vibration signal features contained in the panoramic information of the goaf status are standardized and normalized to generate standardized feature vectors. The GBDT component of the GBDT-TCN hybrid machine learning model performs deep feature selection and combination mining on the standardized feature vectors to extract key physical features from the standardized feature vectors. The TCN component based on the GBDT-TCN hybrid machine learning model receives high-dimensional feature combinations, uses its temporal convolutional network architecture to analyze the long-range temporal dependencies and nonlinear dynamic evolution trends in the high-dimensional feature combinations, and outputs prediction results for future time series. Based on the GBDT-TCN hybrid machine learning model, the prediction results of future time series output by the TCN component are probabilistically calculated and time window analyzed to generate quantitative fire risk level indicators and critical time windows.

5. The intelligent fire prevention and extinguishing method for dynamic control of fires in mine goaf areas as described in claim 4, characterized in that: The prediction instructions with risk levels and time windows are input into the digital twin for multi-plan simulation and decision optimization to generate the optimal control strategy, including the following steps: The digital twin receives prediction instructions with risk levels and time windows, analyzes the fire occurrence probability, critical time windows and potential fire source spatial location parameters contained in the prediction instructions with risk levels and time windows, and maps them into the multiphysics coupling simulation environment of the digital twin. The digital twin automatically generates multiple combinations of control plans based on the mapped fire risk parameters. These combinations of control plans include different disaster response methods and parameter configurations. The digital twin performs multi-physics coupling simulation and deduction for each plan in the combination of control plans, calculates the evolution process of the temperature field, oxygen concentration field and gas diffusion state of the goaf in the future time after the execution of each plan, and outputs the simulation result dataset of each plan. The digital twin performs a comprehensive performance evaluation and comparison of the simulation results dataset of all contingency plans through an objective function, which includes indicators such as cooling rate, resource consumption cost, and ventilation disturbance coefficient.

6. The intelligent fire prevention and extinguishing method for dynamic control of fires in mine goaf areas as described in claim 5, characterized in that: The optimal control strategy is input into the intelligent control unit to execute multi-method coordinated control and generate real-time control actions, including the following steps: The intelligent control unit receives the optimal control strategy generated by the digital twin and analyzes the specific execution instructions and device control parameters in the optimal control strategy; The intelligent control unit converts the parsed equipment control parameters into industrial control signals, and sends the industrial control signals to the corresponding execution equipment through the intrinsically safe control network for mining. The execution equipment includes the nitrogen injection unit frequency converter, the grouting pump motor controller, and the ventilation damper electric actuator. The intelligent control unit monitors the feedback status signal of the execution device and the real-time monitoring data stream of the multi-dimensional sensor network. It compares the feedback status signal and the real-time monitoring data stream with the expected effect curve of the optimal control strategy in real time, and dynamically adjusts the output parameters of the industrial control signal according to the deviation value. The intelligent control unit continuously outputs adjusted industrial control signals to drive the execution equipment to complete collaborative operations and generate real-time control actions that record the equipment's operating status and parameter adjustments.

7. The intelligent fire prevention and extinguishing method for dynamic control of fires in mine goaf areas as described in claim 6, characterized in that: Based on feedback data from real-time control actions, the control effect is evaluated, including the following steps: The effect evaluation unit receives real-time control actions generated by the intelligent control unit, extracts equipment operating status parameters and monitoring data sequences from the real-time control actions, and forms a control process dataset. The effect evaluation unit performs key indicator analysis on the control process dataset, including cooling rate, oxygen concentration decay slope, and carbon monoxide removal efficiency. The effect evaluation unit compares and analyzes the calculated key indicators with the expected effects of the optimal control strategy predicted by the digital twin, and generates a control effect evaluation report on indicator deviation and achievement degree. The effect evaluation unit serves as a reference value for determining the disaster elimination conditions in the control effect evaluation report. The disaster elimination conditions are that the carbon monoxide concentration is below 5 ppm for three consecutive hours and the temperature gradient returns to the baseline range, and the control is completed.

8. A dynamic control intelligent fire prevention and extinguishing system for mine goaf fires, based on the dynamic control intelligent fire prevention and extinguishing method for mine goaf fires according to any one of claims 1 to 7, characterized in that: This includes a data acquisition module that collects raw monitoring data streams and transmits them to an edge computing gateway for preprocessing to generate standardized data packets. The analysis module inputs standardized data packets into the ground central data processing module for data fusion analysis and dynamic threshold calculation, generating panoramic information on the status of the goaf and dynamic threshold vectors. The prediction module inputs panoramic information on the status of the goaf into the GBDT-TCN hybrid machine learning model to predict fire risk and generate prediction instructions with risk level and time window. The control module inputs the prediction instructions with risk level and time window into the digital twin to perform multi-plan simulation and decision optimization, and generate the optimal control strategy. The evaluation module inputs the optimal control strategy into the intelligent control unit to execute multi-means coordinated control, generates real-time control actions, and evaluates the control effect based on the feedback data of the real-time control actions.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the intelligent fire prevention and extinguishing method for dynamic control of fires in mine goaf areas as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the intelligent fire prevention and extinguishing method for dynamic control of fires in the goaf of a mine as described in any one of claims 1 to 7.

Images