Dust-gas double-channel inhibition and early warning method for end slope mining chamber and related device

Abstract

The invention discloses a dust-gas double-channel inhibition and early warning method for an end slope mining chamber and a related device, and relates to the technical field of mine safety and environmental protection, and the method comprises the following steps: obtaining original monitoring data in the end slope mining chamber, methane concentration, temperature, humidity and strain data acquired by a distributed optical fiber sensing module and fog drop current data of an electrostatic dry fog generator are included; inverting the dust concentration based on the fog drop current data to obtain a dust concentration time sequence; inputting the data into a pre-trained Transform-LSTM model, predicting the peak concentration of dust and the peak concentration of methane in the next five minutes, and performing early warning; and inputting the prediction peak value into a model prediction control algorithm to obtain an adjustment instruction of the fog drop flow, the directional air curtain spray angle and the wind speed, and executing a suppression action. Continuous monitoring is achieved through the distributed optical fibers, and methane early warning is 3-4 minutes earlier than that of traditional spot measurement; the static dry fog low water consumption avoids humidity rise and electric conduction risks in the cavern, and the average daily concentration of dust is reduced by 65%; model prediction is accurate, the number of times of gas over-limit shutdown is reduced by 40%, and dust and gas double-channel accurate suppression is effectively achieved.

Description

Technical Field

[0001] This application relates to the field of mine safety and environmental protection technology, and in particular to a method and device for dust-gas dual-channel suppression and early warning in end-face mining tunnels. Background Technology

[0002] End-face mining is an important technical means for recovering residual coal in open-pit coal mines, but its core equipment, the end-face mining machine, generates a large amount of tunneling dust during excavation, with PM2.5 concentrations reaching peak levels of 800 mg·m³. -3 Conventional water curtain or airflow control methods are ineffective at capturing ultrafine dust, which not only endangers the health of workers but also obstructs the field of view of the coal mining machine's camera, affecting the accuracy of remote operation. Furthermore, end-face mining tunnels are narrow, high-pressure spaces prone to sudden localized gas accumulation. Traditional monitoring methods rely on portable or airborne point sensors, which suffer from low monitoring density and slow response, failing to capture the spatial distribution and dynamic changes in gas concentration in a timely manner, thus failing to meet safety early warning requirements.

[0003] In existing technologies, dry fog dust collection, foam dust suppression, and air curtain isolation technologies have been applied in tunnel or longwall mining, but they have not yet achieved synergistic integration for the "high dust + sudden gas" composite scenario of end-face mining. Dry fog alone will increase the humidity and electrical conductivity risk inside the tunnel, and air curtain alone cannot efficiently capture ultrafine dust. Moreover, there is a lack of adaptive control strategies based on multi-source data, making it difficult to achieve precise dual-channel suppression of dust and gas. Summary of the Invention

[0004] The purpose of this application is to provide a method and related device for dust-gas dual-channel suppression and early warning in end-face mining tunnels, which can solve the problems of existing technologies being unable to simultaneously reduce the concentration of ultrafine dust and sudden gas, monitoring lag, and easy increase in environmental risks.

[0005] To achieve the above objectives, this application provides the following solution: In a first aspect, this application provides a method for dust-gas dual-channel suppression and early warning in end-face mining tunnels, comprising the following steps: Acquire raw monitoring data within the end-mounted sampling chamber; the raw monitoring data includes methane concentration, temperature and humidity data, strain data collected by the distributed fiber optic sensing module, and droplet current data from the electrostatic dry fog generator.

[0006] Dust concentration was retrieved by inverting the droplet current data from the electrostatic dry fog generator, resulting in a dust concentration time series.

[0007] Methane concentration, temperature and humidity data, strain data, and dust concentration time series are input into a pre-trained Transformer-LSTM model to predict the peak dust concentration and peak methane concentration within the next 5 minutes and issue an early warning.

[0008] The predicted peak dust concentration and peak methane concentration for the next 5 minutes are input into the model predictive control algorithm to obtain adjustment commands for droplet flow rate, directional air curtain spray angle, and wind speed.

[0009] Based on the adjustment commands of droplet flow rate, directional air curtain spray angle and wind speed, the electrostatic dry fog generator and the directional steerable air curtain are controlled to perform dust-gas suppression actions.

[0010] Optionally, the distributed optical fiber sensing module adopts a "Λ"-shaped main line arrangement of "top plate - right side - loop"; the main optical fiber is arranged close to the center line of the top plate of the end-side mining tunnel, and dense monitoring points are set near the head and tail of the end-side coal mining machine; the distributed optical fiber sensing module also includes branch optical fibers that branch to the left and right sides, a temperature and humidity MEMS chip arranged every 10-15m, and 10-20m sections of DSS strain optical fiber are arranged near the head of the end-side coal mining machine and in easily deformable areas downstream of the air curtain.

[0011] The distributed fiber optic sensing module is based on Raman spectroscopy distributed methane sensing technology. The methane concentration monitoring resolution is 0.05%CH4, the spatial monitoring spacing is 1m, and the maximum monitoring length is 300m.

[0012] The droplet current data of the electrostatic dry fog generator is collected by the electrodes and collecting electrodes of the electrostatic dry fog generator; the electrostatic dry fog generator is arranged on both sides of the head of the end-face coal mining machine, the nozzle particle size is 6μm, it is equipped with a 30kV negative high-voltage electrode, and the water consumption is ≤0.3L・min. -1 Furthermore, the 30kV high-voltage power supply and the metal structure maintain an air gap of ≥150mm.

[0013] Optionally, before retrieving the dust concentration from the droplet current data of the electrostatic dry fog generator to obtain the dust concentration time series, the method further includes the following steps: The droplet current data is filtered to remove electromagnetic interference noise.

[0014] By combining the droplet flux of the electrostatic dry fog generator and the wind field data in the end-side mining tunnel, a mapping relationship between droplet current data and dust concentration was established; the dust concentration includes the mass concentrations of PM2.5 and PM10.

[0015] The established mapping relationship is verified and the inversion error is corrected by using measured data collected by an intrinsically safe dust sensor.

[0016] Optionally, before inputting methane concentration, temperature and humidity data, strain data, and dust concentration time series into a pre-trained Transformer-LSTM model to predict and issue an early warning for peak dust and methane concentrations within the next 5 minutes, the method further includes the following steps: Historical monitoring data is collected, including methane concentration, temperature and humidity data, strain data, droplet current data, and historical control parameters collected by the distributed fiber optic sensing module. Historical control parameters include droplet flow rate, directional air curtain spray angle, and wind speed.

[0017] Using historical monitoring data from the past 300 seconds as the input time window, and the measured peak concentrations of dust, methane, and the time of their occurrence in the next 300 seconds as labels, a training sample set is constructed.

[0018] The AdamW optimizer and layer normalization technique are used, with mean squared error as the loss function. The Transformer-LSTM model is trained based on the training sample set to obtain the pre-trained Transformer-LSTM model. During model training, monotonicity constraints are introduced, and the model is optimized through regularization terms. A sliding window incremental training method is adopted, and the model is updated online based on edge GPU nodes to adapt to changes in coalbed methane content and humidity.

[0019] Optionally, the objective function of the model predictive control algorithm is: .

[0020] in, J The objective function value, w d , w g These are the weighting factors for dust and gas, respectively. and These are the predicted peak dust concentration and peak methane concentration for the next 5 minutes, respectively. The dust concentration limit is... For methane safety limits, r Q , r θ , r V These are the penalty coefficients for controlling actions. D Q , Dth , ΔV These represent the changes in droplet flow rate, directional air curtain spray angle, and wind speed within adjacent control cycles. Q , i , V These are droplet flow rate, directional air curtain spray angle, and wind speed, respectively.

[0021] The constraints of the model predictive control algorithm include: 0 ≤ Q ≤ Q max , Q max This represents the maximum droplet flow rate of the electrostatic dry fog generator; 0°≤ i ≤45°; 0≤ V ≤ V max , V max The maximum wind speed of the ventilation system; and the humidity increase inside the end-side mining tunnel is ≤3%.

[0022] The control cycle of the model predictive control algorithm is 10-20 seconds. It solves the nonlinear programming once per cycle and outputs adjustment commands.

[0023] Optionally, warnings can be issued according to the following process: When the predicted peak methane concentration is ≥1% or the peak dust concentration is >50 mg·m³ within the next 5 minutes... -3 When this occurs, a yellow alert is triggered, and an audio-visual warning is issued.

[0024] When the predicted peak methane concentration is ≥1.5% or the peak dust concentration is >80 mg·m³ within the next 5 minutes. -3 When this occurs, an orange alert is triggered, and the adjustment range of fog droplet flow and wind speed is increased.

[0025] When the predicted peak methane concentration is ≥2% or the peak dust concentration is >120 mg·m³ within the next 5 minutes. -3 At that time, a red alert was triggered, forcing a shutdown and cutting off the power supply to the mining tunnel.

[0026] Secondly, this application provides a dust-gas dual-channel suppression and early warning system for end-face mining tunnels, including the following functional modules: The raw monitoring data acquisition unit is used to acquire raw monitoring data within the end-side sampling chamber. The raw monitoring data includes methane concentration, temperature and humidity data, strain data, and droplet current data from the electrostatic dry fog generator, all collected by the distributed fiber optic sensing module.

[0027] The dust concentration sequence inversion unit is used to invert dust concentration based on the droplet current data of the electrostatic dry fog generator to obtain the dust concentration time series.

[0028] The future concentration prediction and early warning unit is used to input methane concentration, temperature and humidity data, strain data and dust concentration time series into a pre-trained Transformer-LSTM model to predict the peak dust concentration and peak methane concentration within the next 5 minutes and issue an early warning.

[0029] The system control parameter determination unit is used to input the predicted peak dust concentration and peak methane concentration within the next 5 minutes into the model predictive control algorithm to obtain adjustment commands for droplet flow rate, directional air curtain spray angle and wind speed.

[0030] The dust-gas combined suppression unit is used to control the electrostatic dry fog generator and the directional steerable air curtain to perform dust-gas suppression actions according to the adjustment commands of droplet flow rate, directional air curtain spray angle and wind speed.

[0031] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the aforementioned method for dust-gas dual-channel suppression and early warning in end-face mining tunnels.

[0032] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the aforementioned method for dust-gas dual-channel suppression and early warning in end-face mining tunnels.

[0033] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the aforementioned method for dust-gas dual-channel suppression and early warning in end-face mining tunnels.

[0034] According to the specific embodiments provided in this application, the following technical effects are disclosed: This application provides a method and related device for dust-gas dual-channel suppression and early warning in end-face mining tunnels. In this method, a distributed optical fiber sensing module enables continuous monitoring within the end-face mining tunnel at the 100-meter level. Compared to traditional portable point sensors, it can capture the spatial distribution and dynamic changes of gas concentration earlier, laying the foundation for early methane warning. It acquires methane concentration, temperature and humidity data, strain data, and droplet current data from the electrostatic dry fog generator within the end-face mining tunnel. Subsequently, by inverting the mapping relationship between droplet current and dust concentration, and combining it with measured data from an intrinsically safe dust sensor for verification and correction, accurate dust concentration data can be obtained, providing precise data support for subsequent dust peak prediction and suppression. Then, the above data is input into a pre-trained Transformer-LSTM model, and through Transformer-LSTM... The MERS encoder captures the spatial dependencies between different monitoring points (such as the impact of upstream gas accumulation on downstream). Combined with the LSTM decoder to learn the temporal dynamic characteristics of concentration changes, it can accurately predict the concentration peak in the next 5 minutes and issue warnings based on the predicted values. Finally, based on the model predictive control algorithm, with the goal of "dust concentration close to the safety limit, gas concentration close to the safety limit, and smooth control action", it obtains adjustment commands for droplet flow rate, directional air curtain spray angle and wind speed and implements corresponding control, realizing dynamic adaptation of dust and gas collaborative control. This solves the problem that existing technologies lack adaptive control strategies based on multi-source data. The entire suppression process relies on low water consumption design and adaptive spray angle adjustment to ensure that the humidity increase in the chamber is ≤3%, further avoiding the risk of electrical conductivity, and ultimately achieving precise dual-channel suppression of dust and gas. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0036] Figure 1 The flowchart illustrates a method for dust-gas dual-channel suppression and early warning in end-face mining tunnels, as provided in one embodiment of this application.

[0037] Figure 2 This is a schematic diagram of the functional units of a dust-gas dual-channel suppression and early warning system for end-face mining tunnels, provided in an embodiment of this application.

[0038] Figure 3 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0039] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0040] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0041] This application provides a method for dust-gas dual-channel suppression and early warning in end-face mining tunnels. In one exemplary embodiment, such as... Figure 1 As shown, it includes the following steps: A1. Obtain raw monitoring data from the end-mounted sampling chamber; the raw monitoring data includes methane concentration, temperature and humidity data, strain data, and droplet current data from the electrostatic dry fog generator collected by the distributed fiber optic sensing module.

[0042] Specifically, the distributed fiber optic sensing module adopts a "Λ"-shaped main line arrangement of "top plate - right side - loop"; the main fiber is arranged close to the center line of the top plate of the end-side mining tunnel, with dense monitoring points near the head and tail of the end-side coal mining machine; the distributed fiber optic sensing module also includes branch fibers branching to the left and right sides, a temperature and humidity MEMS chip arranged every 10-15m, and 10-20m sections of DSS strain fiber arranged near the head of the end-side coal mining machine and in easily deformable areas downstream of the gas curtain. The distributed fiber optic sensing module is based on Raman spectroscopy distributed methane sensing technology, with a methane concentration monitoring resolution of 0.05%CH4, a spatial monitoring spacing of 1m, and a maximum monitoring length of 300m.

[0043] Compared to portable gas detectors or handheld CH4 detectors commonly used underground, the distributed fiber optic sensing module used in this embodiment is characterized by point measurement and sparse spatial coverage, which may result in a lag in the response to early local enrichment in narrow mining tunnels. Traditional portable and airborne sensors have low point measurement density and lag. This solution uses distributed Raman fiber to achieve continuous scanning at the hundred-meter level, thus enabling earlier detection of the evolution of "local enrichment - migration towards the mining section," thereby achieving an early warning 3-4 minutes earlier than portable instruments.

[0044] The droplet current data of the electrostatic dry fog generator is collected by the electrodes and collecting electrodes of the electrostatic dry fog generator; the electrostatic dry fog generator is arranged on both sides of the head of the end-face coal mining machine, the nozzle particle size is 6μm, it is equipped with a 30kV negative high-voltage electrode, and the water consumption is ≤0.3L・min. -1 Furthermore, the 30kV high-voltage power supply and the metal structure maintain an air gap of ≥150mm.

[0045] The electrostatic dry fog generator is positioned slightly above the cutoff point, creating an "interception zone" between the rising and returning dust paths. A row of spray booms is symmetrically arranged on each side, with 2-3 nozzles per row spaced 1.2–1.5m apart. The spray pattern forms a narrow cone angle facing the main dust channel. The spray booms and electrodes are aligned in the same direction, with the electrodes downstream of the droplets (relative to the airflow) to enhance charging efficiency and collection rate. A ≥150mm air gap is maintained between the 30kV high-voltage power supply and the metal structure. Shielding and grounding are added for humid environments (without altering the "low humidity" objective). In conjunction with ventilation, the dry fog jet direction is slightly counter-current, facilitating initial collection before the "directional air curtain" and reducing the dust load entering the air curtain.

[0046] A2. Dust concentration is retrieved from the droplet current data of the electrostatic dry fog generator to obtain a dust concentration time series. In this embodiment, the current change of the electrode / collector electrode of the electrostatic dry fog generator is used to indirectly estimate the captured dust load. Combined with the droplet flux and wind field, the equivalent mass concentration estimate of PM2.5 / PM10 is obtained. In practical applications, 1-2 intrinsically safe dust sensors can be added for verification.

[0047] A3. Input methane concentration, temperature and humidity data, strain data, and dust concentration time series into a pre-trained Transformer-LSTM model to predict the peak dust concentration and peak methane concentration within the next 5 minutes and issue an early warning. In addition to methane concentration, temperature and humidity data, strain data, and dust concentration time series, auxiliary parameters also include equipment pose / power, etc., which are input as control features into the Transformer-LSTM model.

[0048] Specifically, warnings will be issued according to the following three-level warning process: When the predicted peak methane concentration is ≥1% or the peak dust concentration is >50 mg·m³ within the next 5 minutes... -3 When this occurs, a yellow alert is triggered, and an audio-visual warning is issued.

[0049] When the predicted peak methane concentration is ≥1.5% or the peak dust concentration is >80 mg·m³ within the next 5 minutes. -3 When this occurs, an orange alert is triggered, and the adjustment range of fog droplet flow and wind speed is increased.

[0050] When the predicted peak methane concentration is ≥2% or the peak dust concentration is >120 mg·m³ within the next 5 minutes. -3 At that time, a red alert was triggered, forcing a shutdown and cutting off the power supply to the mining tunnel.

[0051] A4. The predicted peak dust concentration and peak methane concentration for the next 5 minutes are input into the model predictive control algorithm to obtain adjustment commands for droplet flow rate, directional air curtain spray angle, and wind speed. Specifically, the objective function of the model predictive control algorithm is: .

[0052] in, J The objective function value, w d , w g These are the weighting factors for dust and gas, respectively. and These are the predicted peak dust concentration and peak methane concentration for the next 5 minutes, respectively. The permissible concentration limit for dust, such as 2 mg / m³. -3 Daily average, or 120 mg / m -3 Peak warning value; For example, a safety limit for methane, such as 2%; r Q , r θ , r V These are the penalty coefficients for controlling actions. ΔQ , Dth , ΔV These represent the changes in droplet flow rate, directional air curtain spray angle, and wind speed within adjacent control cycles. Q , i , V These are droplet flow rate, directional air curtain spray angle, and wind speed, respectively.

[0053] The constraints of the model predictive control algorithm include: 0 ≤ Q ≤ Q max , Q max This represents the maximum droplet flow rate of the electrostatic dry fog generator; 0°≤ i ≤45°; 0≤ V ≤ V max , V max The maximum wind speed of the ventilation system; and the humidity increase inside the end-side mining tunnel is ≤3%.

[0054] The control cycle of the model predictive control algorithm is 10-20 seconds. It solves the nonlinear programming once per cycle and outputs adjustment commands.

[0055] The main objectives of model predictive control include dust mass concentration C. dust (PM available) 2.5 The main indicator, supplemented by PM 10 The spatial mean or quantile (p90 / p95) of methane volume fraction in the driver's field of view ROI. CH4 (%CH4) is weighted in the top plate region and downstream of the wind curtain, areas prone to enrichment. Distributed Raman output naturally supports spatial weighting.

[0056] Key points for actual parameter tuning: Droplet flow rate Q, main influencing factor C dust (Capture + Coagulation); Secondary Impact C CH4 (By altering the microscopic flow field / humidity, dilution efficiency is affected, albeit to a small extent). The directional air curtain spray angle θ affects both: a large angle baffle suppresses backflow → beneficial for dust control; a small angle horizontal push → beneficial for gas dilution. Wind speed V primarily affects C. CH4 (Dilution / Displacement), Secondary Effect C dust (High wind speeds can also push uncollected dust away from the driver's field of vision). It needs to be explained here that in end-side mining, since the driver is not in the actual driver's seat of the end-side mining machine, the end-side mining machine is operated from the field of vision of the camera on the cutting drum of the continuous mining machine.

[0057] Trade-off strategy: Used in Model Predictive Control (MPC) algorithm w d , w gIt embodies "safety priority"; in the rule-based approach, different combined actions are driven by thresholds / levels (yellow / orange / red) (see the graded early warning methods below).

[0058] The control strategy is as follows: like If the dominant factor is V, then prioritize increasing V (ventilation volume) and decreasing θ (more gradual dilution), with minor modifications to Q; like If the dominant factor is Q (droplet flow rate), then prioritize increasing Q (droplet flow rate) and adjusting θ (to further isolate backflow), followed by V. If both are high, then V and Q are increased simultaneously, but θ is compromised (20–30°), and then continuously optimized and fine-tuned.

[0059] A5. Based on the adjustment commands for droplet flow rate, directional air curtain spray angle, and wind speed, control the electrostatic dry fog generator and the directional steerable air curtain to perform dust-gas suppression actions.

[0060] The directional steerable air curtain includes a "gate-type" linear spray bar (with spray holes evenly distributed along the width of the roadway) below the tail conveyor, left and right side compensation nozzles, a 0.6MPa compressed air source, and a servo spray angle adjustment unit; through adaptive adjustment of the spray angle (0°-45°), it realizes the switching between "air curtain isolation of dust" and "push flow dilution of gas", blocking dust backflow and introducing fresh air.

[0061] As an optional implementation, before retrieving the dust concentration from the droplet current data of the electrostatic dry fog generator to obtain the dust concentration time series, the method further includes the following steps: B1. Filter the droplet current data to remove electromagnetic interference noise.

[0062] B2. Based on the droplet flux of the electrostatic dry fog generator and the wind field data in the end-side mining tunnel, establish a mapping relationship between droplet current data and dust concentration; dust concentration includes PM2.5 and PM10 mass concentrations.

[0063] B3. The established mapping relationship is verified using the measured data collected by the intrinsically safe dust sensor to correct the inversion error.

[0064] In another exemplary embodiment of this application, prior to step A3, the method further includes the following steps: C1. Collect historical monitoring data; historical monitoring data includes methane concentration, temperature and humidity data, strain data, droplet current data, and historical control parameters collected by the distributed fiber optic sensing module; historical control parameters include droplet flow rate, directional air curtain spray angle, and wind speed.

[0065] C2. Using historical monitoring data from the past 300 seconds as the input time window, and the measured peak concentration of dust, peak concentration of methane, and peak occurrence time of the next 300 seconds as labels, a training sample set is constructed.

[0066] C3. The AdamW optimizer and layer normalization technique are adopted, and the mean squared error is used as the loss function. The Transformer-LSTM model is trained based on the training sample set to obtain the pre-trained Transformer-LSTM model. During model training, monotonicity constraints are introduced and the model is optimized through regularization terms. The sliding window incremental training method is adopted, and the model is updated online based on edge GPU nodes to adapt to changes in coalbed methane content and humidity.

[0067] In this embodiment, the trained Transformer-LSTM model structure is as follows: Input layer: methane concentration sequence (one point per 1m, possibly 300 dimensions), droplet current → dust concentration time series, auxiliary features such as temperature, humidity, and wind speed, and historical control variables ( Q , i , V Concatenate them into a multidimensional time-series input matrix X∈R T×d .

[0068] The Transformer encoder uses a multi-head attention mechanism to capture the dependencies between gas / dust points at different locations (such as the impact of upstream gas accumulation on downstream areas). It uses a feed-forward network for non-linear mapping to improve expressive power. The output is a "global feature embedding".

[0069] The LSTM decoder receives features from the Transformer, learns the local dynamic changes of the time series, and outputs the dust concentration curve and peak value for the next 5 minutes and the methane concentration curve and peak value for the next 5 minutes through two branches of the output layer.

[0070] The prediction loss function is as follows: .

[0071] In the formula, The model predicts the peak dust concentration (mg·m³) for the next 5 minutes. -3 ), C dust,peak This represents the actual observed peak dust concentration. For the predicted peak methane volume fraction (%) in the next 5 minutes, C CH4,peak The peak value of methane is the actual observed value, and MSE() is the mean squared error function.

[0072] Through the above-mentioned solution, this application addresses the issue of reducing particulate matter and methane concentrations in end-face mining tunnels without increasing humidity or electrical conductivity risks within the tunnels. The daily average PM2.5 concentration is reduced by 65%, far below 2 mg / m³. -3 Limits are no longer obstructing the driver's view. Methane warnings are delivered 3 minutes in advance (based on a distributed fiber optic sensing module), enabling dynamic and coordinated control of source containment and end-of-pipe dilution, reducing outages due to exceeding limits by 40%; droplet water consumption is only 15% of ordinary spraying, and humidity increase is less than 3%.

[0073] Based on the same inventive concept, this application also provides a system for implementing the aforementioned method for dual-channel suppression and early warning of dust and gas in end-face mining tunnels. The solution provided by this system is similar to the implementation described in the above method. In an exemplary embodiment, such as... Figure 2 As shown, a dust-gas dual-channel suppression and early warning system for end-face mining tunnels is provided, including the following functional modules: The raw monitoring data acquisition unit is used to acquire raw monitoring data within the end-side sampling chamber. The raw monitoring data includes methane concentration, temperature and humidity data, strain data, and droplet current data from the electrostatic dry fog generator, all collected by the distributed fiber optic sensing module.

[0074] The dust concentration sequence inversion unit is used to invert dust concentration based on the droplet current data of the electrostatic dry fog generator to obtain the dust concentration time series.

[0075] The future concentration prediction and early warning unit is used to input methane concentration, temperature and humidity data, strain data and dust concentration time series into a pre-trained Transformer-LSTM model to predict the peak dust concentration and peak methane concentration within the next 5 minutes and issue an early warning.

[0076] The system control parameter determination unit is used to input the predicted peak dust concentration and peak methane concentration within the next 5 minutes into the model predictive control algorithm to obtain adjustment commands for droplet flow rate, directional air curtain spray angle and wind speed.

[0077] The dust-gas combined suppression unit is used to control the electrostatic dry fog generator and the directional steerable air curtain to perform dust-gas suppression actions according to the adjustment commands of droplet flow rate, directional air curtain spray angle and wind speed.

[0078] certainly, Figure 2 The architecture shown is merely exemplary; it can be omitted as needed when implementing different functionalities. Figure 2 One or at least two components of the system shown.

[0079] The following details the implementation process of this application, taking into account the actual working conditions of an end-face mining tunnel (using an end-face mining tunnel in an open-pit coal mine as an example, with a cross-sectional dimension of 4m × 3m, a length of 200m, and a coal mining machine power of 315kW): From a hardware structure design perspective, the following hardware is first deployed around the end-face coal mining machine: Electrostatic dry fog generating module: Install one row of spray bars on each side of the coal mining machine head, with 3 nozzles in each row (spaced 1.5m apart). The nozzle height is slightly higher than the cutting point (0.8m from the coal wall), the spray cone angle is 30°, and it faces the main dust channel; the 30kV high voltage power supply maintains a 200mm air gap with the metal shell of the coal mining machine, and the grounding resistance is ≤4Ω.

[0080] Directional steerable air curtain module: Install a "gate-type" spray bar (4m in length, 2mm in diameter of nozzle, and 50mm in spacing) below the tail conveyor of the coal mining machine, and install one set of compensating nozzles on each of the left and right sides (1.5m from the bottom plate); adjust the compressed air source pressure to 0.6MPa, and initially set the spray angle to 25° (compromise state).

[0081] Distributed fiber optic sensing module: A main fiber optic cable (200m) is laid along the centerline of the tunnel roof. Two dense monitoring points (0.5m apart) are added at the head (0m) and tail (200m). A branch fiber optic cable (200m) is laid on the right side and a branch fiber optic cable (100m) is laid on the left side. A temperature and humidity MEMS chip is installed at 50m, 100m and 150m. DSS strain fiber optic cables are arranged at the head (0-20m) and downstream of the air curtain (200-220m).

[0082] AI Adaptive Control Module: It adopts an intrinsically safe control cabinet for mining, with a built-in edge GPU (NVIDIA Jetson AGX), pre-installed Transformer-LSTM model (pre-trained data from 3 months of historical monitoring data of the same coal seam) and MPC solver; the model input window is set to 300s, the prediction window is set to 300s, and the MPC control cycle is 20s.

[0083] Early warning module: Install one audible and visual alarm at 10m, 100m, and 200m inside the tunnel. Deploy early warning terminals at the ground monitoring center. The threshold is set as follows: Yellow Alert (CH4 ≥ 1% / PM2.5 > 50mg·m³). -3 Orange alert (CH4 ≥ 1.5% / PM2.5 > 80mg·m³) -3 Red alert (CH4 ≥ 2% / PM2.5 > 120 mg·m³) -3 ).

[0084] When applying this system, follow the procedure below to perform dual-channel dust and gas suppression and early warning: (1) Data acquisition stage: The distributed optical fiber sensing module acquires data once per second, and the droplet current is acquired once every 0.5 seconds. The data is synchronously transmitted to the AI ​​control module via the downhole Ethernet and then timestamped after filtering and noise reduction.

[0085] (2) Dust inversion stage: based on droplet current (measured range 0.1-5mA) and preset mapping model (e.g., I=k×C) 0.8 (k was calibrated to 0.02 in the preliminary experiment) to retrieve PM2.5 concentration (e.g., a current of 2 mA corresponds to a PM2.5 concentration of 40 mg·m³). -3 And it was calibrated by an intrinsically safe dust sensor (placed at 50m), and the error was controlled within ±5%.

[0086] (3) Peak prediction stage: The model input includes the methane concentration sequence (200 points), PM2.5 concentration sequence (600 data points), temperature and humidity (3 MEMS data points), and historical control parameters within 300 seconds. Q =0.2L / min, i =25°, V =1.5m / s), output the prediction for the next 5 minutes: C dust,peak =75mg・m -3 , C CH4,peak =1.3%.

[0087] (4) Control and Regulation Stage: The MPC algorithm constructs the objective function ( w d =0.4, w g =0.6, r Q =0.1, r θ =0.1, r V =0.1), the adjustment command is obtained by solving: Q Increase to 0.25 L / min. i Increase to 30° (enhanced dust isolation). V Increase to 1.8 m / s (enhanced gas dilution), adjustment range ΔQ =0.05L / min, Dth =5°, ΔV =0.3m / s, which meets the smoothness constraint.

[0088] (5) Early warning triggering stage: due to C CH4,peak =1.3% (≥1% and <1.5%), triggering a yellow alert. The alarm inside the tunnel emits intermittent audible and visual alerts, and the ground terminal displays the warning information. There is no forced shutdown action.

[0089] Incremental model training is performed once every 7 days via edge GPU nodes, using measured data from the past 8-10 minutes (including robust samples of wind volume perturbations and humidity fluctuations) to update model parameters, ensuring that the prediction error is always <10%. When the coal seam gas content increases (e.g., the basic methane concentration increases from 0.5% to 0.8%), the model can automatically adapt without retraining.

[0090] All electrical components of the device in this application comply with the intrinsically safe design requirements of the "Coal Mine Safety Regulations". The high-voltage system of the electrostatic dry fog generator is equipped with overcurrent protection, and the distributed optical fiber has no risk of electric sparks. The acquisition, storage and transmission of monitoring data are all authorized by the mine and comply with user information security regulations.

[0091] Furthermore, the technical solution of this application needs to be adapted to the actual dimensions of the end-face mining tunnel, the coal seam gas content, and ventilation conditions: the fiber optic cable layout length should cover the entire length of the mining tunnel + a 50m extension section to ensure full-range monitoring of gas diffusion; the nozzle spacing of the electrostatic dry fog generator needs to be adjusted according to the cutting width of the coal mining machine (e.g., when the cutting width is 5m, the nozzle spacing can be reduced to 1.2m); the weighting factor of the MPC algorithm... w d , w g Adjustments need to be made according to the mine's safety regulations (e.g., for high-gas mines, adjustments can be made). w g Increased to 0.8).

[0092] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 3 As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. 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, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it can implement the dust-gas dual-channel suppression and early warning method for end-face mining tunnels provided in the previous embodiment.

[0093] Those skilled in the art will understand that Figure 3The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0094] In one exemplary embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0095] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0096] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0097] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0098] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0099] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0100] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for dust-gas dual-channel suppression and early warning in end-face mining tunnels, characterized in that, include: Obtain raw monitoring data from the end-slope mining tunnel; The raw monitoring data includes methane concentration, temperature and humidity data, strain data collected by the distributed fiber optic sensing module, and droplet current data from the electrostatic dry fog generator. Dust concentration is retrieved based on the droplet current data from the electrostatic dry fog generator to obtain a dust concentration time series. The methane concentration, temperature and humidity data, strain data, and dust concentration time series are input into a pre-trained Transformer-LSTM model to predict the peak dust concentration and peak methane concentration within the next 5 minutes and issue an early warning. The predicted peak dust concentration and peak methane concentration for the next 5 minutes are input into the model predictive control algorithm to obtain adjustment commands for droplet flow rate, directional air curtain spray angle and wind speed. Based on the adjustment commands for droplet flow rate, directional air curtain spray angle, and wind speed, the electrostatic dry fog generator and the directional steerable air curtain are controlled to perform dust-gas suppression actions.

2. The method for dust-gas dual-channel suppression and early warning in end-face mining tunnels according to claim 1, characterized in that, The distributed optical fiber sensing module adopts a "Λ" shaped main line arrangement of "top plate - right side - loop"; the main optical fiber is arranged close to the center line of the top plate of the end-side mining tunnel, and dense monitoring points are set near the head and tail of the end-side coal mining machine; the distributed optical fiber sensing module also includes branch optical fibers that branch to the left and right sides, temperature and humidity MEMS chips arranged every 10-15m, and 10-20m sections of DSS strain optical fiber are arranged near the head of the end-side coal mining machine and in easily deformable areas downstream of the air curtain. The distributed optical fiber sensing module is based on Raman spectroscopy distributed methane sensing technology, with a methane concentration monitoring resolution of 0.05%CH4, a spatial monitoring spacing of 1m, and a maximum monitoring length of 300m. The droplet current data of the electrostatic dry fog generator is collected by the electrodes and collecting electrodes of the electrostatic dry fog generator; the electrostatic dry fog generator is arranged on both sides of the head of the end-face coal mining machine, the nozzle particle size is 6μm, it is equipped with a 30kV negative high-voltage electrode, and the water consumption is ≤0.3L・min. -1 Furthermore, the 30kV high-voltage power supply and the metal structure maintain an air gap of ≥150mm.

3. The method for dust-gas dual-channel suppression and early warning in end-face mining tunnels according to claim 1, characterized in that, Before obtaining the dust concentration time series by inverting the dust concentration based on the droplet current data of the electrostatic dry fog generator, the following steps are also included: The droplet current data is filtered to remove electromagnetic interference noise; By combining the droplet flux of the electrostatic dry fog generator and the wind field data in the end-side mining tunnel, a mapping relationship between droplet current data and dust concentration is established; the dust concentration includes the mass concentrations of PM2.5 and PM10. The established mapping relationship is verified and the inversion error is corrected by using measured data collected by an intrinsically safe dust sensor.

4. The method for dust-gas dual-channel suppression and early warning in end-face mining tunnels according to claim 1, characterized in that, Before inputting the methane concentration, temperature and humidity data, strain data, and dust concentration time series into a pre-trained Transformer-LSTM model to predict the peak dust concentration and peak methane concentration within the next 5 minutes and issue an early warning, the process also includes: Historical monitoring data is collected; the historical monitoring data includes methane concentration, temperature and humidity data, strain data, droplet current data, and historical control parameters collected by the distributed fiber optic sensing module; the historical control parameters include droplet flow rate, directional air curtain spray angle, and wind speed. Using historical monitoring data from the past 300 seconds as the input time window, and the measured peak concentrations of dust, methane, and the time of their occurrence in the next 300 seconds as labels, a training sample set is constructed. The AdamW optimizer and layer normalization technique are used, with mean squared error as the loss function. The Transformer-LSTM model is trained based on the training sample set to obtain a pre-trained Transformer-LSTM model. During model training, monotonicity constraints are introduced, and the model is optimized through regularization terms. A sliding window incremental training method is adopted, and the model is updated online based on edge GPU nodes to adapt to changes in coalbed methane content and humidity.

5. The method for dust-gas dual-channel suppression and early warning in end-face mining tunnels according to claim 1, characterized in that, The objective function of the model predictive control algorithm is: ； in, J The objective function value, w d , w g These are the weighting factors for dust and gas, respectively. and These are the predicted peak dust concentration and peak methane concentration for the next 5 minutes, respectively. The dust concentration limit is... For methane safety limits, ρ Q , ρ θ , ρ V These are the penalty coefficients for controlling actions. Δ Q , Δθ , ΔV These represent the changes in droplet flow rate, directional air curtain spray angle, and wind speed within adjacent control cycles. Q , θ , V These are droplet flow rate, directional air curtain spray angle, and wind speed, respectively. The constraints of the model predictive control algorithm include: 0 ≤ Q ≤ Q max , Q max This represents the maximum droplet flow rate of the electrostatic dry fog generator; 0°≤ θ ≤45°; 0≤ V ≤ V max , V max The maximum wind speed of the ventilation system; and the humidity increase inside the end-side mining tunnel ≤3%; The control cycle of the model predictive control algorithm is 10-20 seconds. It solves the nonlinear programming once per cycle and outputs adjustment commands.

6. The method for dust-gas dual-channel suppression and early warning in end-face mining tunnels according to claim 1, characterized in that, The following procedure will be used to issue warnings: When the predicted peak methane concentration is ≥1% or the peak dust concentration is >50 mg·m³ within the next 5 minutes... -3 When this is triggered, a yellow alert will be issued, along with an audio-visual warning. When the predicted peak methane concentration is ≥1.5% or the peak dust concentration is >80 mg·m³ within the next 5 minutes. -3 When this occurs, an orange alert is triggered, and the adjustment range of fog droplet flow and wind speed is increased; When the predicted peak methane concentration is ≥2% or the peak dust concentration is >120 mg·m³ within the next 5 minutes. -3 At that time, a red alert was triggered, forcing a shutdown and cutting off the power supply to the mining tunnel.

7. A dust-gas dual-channel suppression and early warning system for end-face mining tunnels, characterized in that, include: The raw monitoring data acquisition unit is used to acquire raw monitoring data within the end-side mining tunnel. The raw monitoring data includes methane concentration, temperature and humidity data, strain data collected by the distributed fiber optic sensing module, and droplet current data from the electrostatic dry fog generator. The dust concentration sequence inversion unit is used to invert dust concentration based on the droplet current data of the electrostatic dry fog generator to obtain a dust concentration time series; The future concentration prediction and early warning unit is used to input the methane concentration, temperature and humidity data, strain data and dust concentration time series into the pre-trained Transformer-LSTM model to predict the peak dust concentration and peak methane concentration within the next 5 minutes and issue an early warning. The system control parameter determination unit is used to input the predicted peak dust concentration and peak methane concentration within the next 5 minutes into the model predictive control algorithm to obtain adjustment commands for droplet flow rate, directional air curtain spray angle and wind speed. The dust-gas combined suppression unit is used to control the electrostatic dry fog generator and the directional steerable air curtain to perform dust-gas suppression actions according to the adjustment commands of the droplet flow rate, the directional air curtain spray angle and the wind speed.

8. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the dust-gas dual-channel suppression and early warning method for end-face mining tunnels according to any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the dust-gas dual-channel suppression and early warning method for end-face mining tunnels as described in any one of claims 1-6.

10. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the dust-gas dual-channel suppression and early warning method for end-face mining tunnels as described in any one of claims 1-6.

Images