Coal mine tunnel ventilation dynamic adjusting system based on Internet of Things

Abstract

The invention discloses a coal mine tunnel ventilation dynamic adjusting system based on the Internet of Things, and particularly relates to the technical field of coal mine tunnel ventilation control. Various gas concentration data, dust distribution data, environmental parameter data, emergency parameter data and electromechanical equipment state data in a coal mine tunnel are detected on line through deployed sensors; the construction of a more comprehensive risk perception system is facilitated; feature data is extracted from coal mine tunnel online detection data passing data quality verification and input into the trained logistic regression model for reasoning a single risk probability and a comprehensive risk probability, so that whether risk regulation is needed or not is judged, and early risk pre-judgment of multi-parameter association can be realized; a regulation scheme is generated based on the single risk probability, the comprehensive risk probability, the single risk event regulation demand and the comprehensive risk event regulation demand, the comprehensive risk probability and the single risk probability are minimized while the optimal parameter adjustment amount is solved, and blind regulation is avoided.

Description

Technical Field

[0001] This invention relates to the field of coal mine roadway ventilation control technology, and more specifically, to a dynamic adjustment system for coal mine roadway ventilation based on the Internet of Things. Background Technology

[0002] In coal mine production, the stable operation of the roadway ventilation system is crucial for ensuring the safety of underground workers and the safe production of the mine. Traditional methods of regulating coal mine roadway ventilation mainly rely on manual experience, controlling the ventilation volume by setting up fixed ventilation facilities such as air doors and windows. However, with the increase in coal mining depth, changes in mining conditions, and the increasing complexity of the underground working environment, this traditional method has many shortcomings and needs to be upgraded and optimized.

[0003] The existing patent document, CN115030763B, entitled "A Method for Intelligent Pressure Equalization Ventilation and Fire Prevention in Mines," collects and analyzes ventilation environment parameters to establish a coal mine ventilation model, plans the location of hardware equipment, and builds a three-dimensional ventilation tunnel model. It achieves pressure control of airflow by controlling the frequency of fans and the opening of ventilation windows. Furthermore, it builds a cloud computing intelligent analysis platform to optimize the coal mine ventilation model, establish a coal mine ventilation data model, set early warning thresholds, and establish an early warning data model. This patent realizes the management of mine ventilation tunnels through the constructed mine ventilation model. Figure 3 The integrated display allows for dynamic monitoring of mine ventilation and air pressure changes based on adjusted parameters. Furthermore, the pre-set early warning model can provide early warnings of abnormal underground ventilation, enabling timely detection of related problems and elimination of safety hazards.

[0004] However, existing methods still have some shortcomings: the data collected by existing methods is relatively simple, mainly relying on basic ventilation parameters (wind pressure, wind speed, etc.) and static roadway characteristics, lacking real-time monitoring of dynamic safety parameters such as gas concentration, dust, and the status of electromechanical equipment, making it difficult to comprehensively perceive fire hazards; the risk warning of existing methods is lagging, only triggering alarms through thresholds, without establishing a risk probability model with multi-parameter correlation, and unable to predict complex risks in advance; the control flexibility of existing methods is insufficient, relying on fixed ventilation models and pressure equalization control, without dynamically combining real-time environmental data to optimize the adjustment amount, which may lead to insufficient or excessive control. Therefore, existing methods still need to be optimized. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art, the present invention provides a dynamic adjustment system for ventilation in coal mine roadways based on the Internet of Things to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a dynamic adjustment system for ventilation in coal mine roadways based on the Internet of Things, comprising: Sensor control module: Based on the sensor deployment scheme, multiple types of sensors are deployed in the coal mine roadway, and different maintenance strategies are implemented for the sensors based on their health levels; Online monitoring module for coal mine roadways: It uses deployed sensors to monitor various gas concentrations, dust distribution, environmental parameters, emergency parameters, and the status of electromechanical equipment in coal mine roadways online. Data quality verification module: Performs physical range verification, timeliness verification, spatial consistency verification, equipment status correlation verification, multi-source data mutual verification, and missing value filling on the data detected in the coal mine roadway in sequence, and outputs the verification results. Coal mine roadway risk diagnosis module: Extract feature data from online detection data of coal mine roadways that have passed data quality verification and input it into the trained logistic regression model to infer the probability of single risk and the probability of comprehensive risk, and determine whether risk control is needed based on the inference results; Risk control scheme generation module: Analyzes the risk diagnosis results and finally generates a risk control scheme that includes target control parameters, the optimal adjustment amount of the target control parameters, the expected effect of single risk probability, and the expected effect of comprehensive risk probability. Risk control plan execution module: Performs dynamic control according to the generated risk control plan.

[0007] The technical effects and advantages of this invention are as follows: 1. This invention sets up an online detection module for coal mine roadways, which uses deployed sensors to detect various gas concentration data, dust distribution data, environmental parameter data, emergency parameter data, and electromechanical equipment status data in coal mine roadways online. By deploying multiple types of sensors (gas, dust, equipment status, etc.), it supplements the dynamic safety parameters missing in existing methods and builds a more comprehensive risk perception system.

[0008] 2. The present invention sets up a coal mine roadway risk diagnosis module to extract feature data from online detection data of coal mine roadways that has passed data quality verification and input it into a trained logistic regression model to infer single risk probability and comprehensive risk probability. Based on the inference results, it determines whether risk control is needed, which replaces the simple threshold alarm of the existing method and can realize early risk prediction with multi-parameter correlation.

[0009] 3. The present invention sets up a risk control scheme generation module that constructs a risk-parameter correlation matrix based on single risk probability, comprehensive risk probability, single risk event control requirements, and comprehensive risk event control requirements. It generates a set of control requirements and prioritizes them, predicts the change in the probability of risk events after each control parameter is adjusted, minimizes the comprehensive risk probability and single risk probability while solving for the optimal parameter adjustment amount, and combines the expected effect evaluation to avoid blind control. Attached Figure Description

[0010] Figure 1 This is a system structure block diagram of the present invention.

[0011] Figure 2 This is a flowchart illustrating the risk control method of the present invention.

[0012] Figure 3 This is a flowchart illustrating the device association verification method of the present invention.

[0013] Figure 4 This is a flowchart illustrating the steps of the risk-parameter correlation matrix construction method of the present invention. Detailed Implementation

[0014] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0015] like Figure 1 The embodiment shown provides an IoT-based dynamic adjustment system for coal mine roadway ventilation, including a sensor control module, a coal mine roadway online detection module, a data quality verification module, a coal mine roadway risk diagnosis module, a risk control scheme generation module, a control scheme execution module, and a database. The sensor control module, coal mine roadway online detection module, data quality verification module, coal mine roadway risk diagnosis module, risk control scheme generation module, and control scheme execution module are connected sequentially. The data quality verification module is connected to the sensor control module, and the sensor control module is connected to the risk control scheme generation module. All modules in the system are connected to the database.

[0016] The sensor control module deploys multiple types of sensors in the coal mine roadway based on the sensor deployment scheme, and performs different maintenance strategies on the sensors based on their health levels. Furthermore, the sensor control module includes a sensor deployment plan input unit, a sensor deployment plan implementation control unit, a sensor health assessment result receiving unit, and a sensor maintenance strategy execution unit. The sensor deployment plan input unit is used to input the sensor deployment plan, including sensor ID, model, technical parameters, deployment location, and installation requirements. The sensor deployment plan implementation control unit is used to ensure that the actual sensor deployment is consistent with the sensor deployment plan. The sensor health assessment result receiving unit is used to receive the sensor health level and maintenance strategy. The sensor maintenance strategy execution unit is used to execute the sensor maintenance strategy.

[0017] Specifically, in this embodiment, the deployed sensors are used to detect the concentrations of various gases, dust distribution, environmental parameters, emergency parameters, and the status of electromechanical equipment within the coal mine roadway. Infrared absorption sensors are installed 5-10m from the face of the working face to detect CH4 concentration, and electrochemical sensors are installed to detect CO concentration. In the return airway, infrared absorption sensors are deployed at 50m and 100m intervals to detect CH4 and CO2 concentrations, electrochemical sensors to detect CO concentration, and paramagnetic oxygen analyzers to detect O2 concentration. Infrared absorption sensors are installed near air leakage points on the sealed wall of the goaf to detect CO2 concentration. A paramagnetic oxygen analyzer is installed on the top of the electromechanical chamber to detect O2 concentration. Laser scattering sensors are deployed 10m below the coal mining machine in the coal mine roadway to detect PM2.5 and PM10 concentrations. Beta-ray absorption sensors are deployed at the belt conveyor transfer points in the coal mine roadway to detect total dust concentration. Electrostatic induction sensors are deployed in the personnel operating area of ​​the coal mine roadway to detect respirable dust concentration. An ultrasonic anemometer is installed at the center of the coal mine roadway cross-section to detect wind speed; platinum resistance temperature sensors and capacitive humidity sensors are installed in pairs at the entrance of the mining area (10m from the roadway opening) and 5m before the intersection of the air intake and exhaust roadways at the mining area exit to detect the temperature and humidity at the mining area entrance and exit; micro differential pressure transmitters are installed before and after the ventilation doors to detect ventilation resistance. Infrared thermal imagers are installed in areas with flammable material accumulation to detect flue gas temperature; fiber optic sensors are installed in geological structural zones to detect roadway deformation; acoustic emission sensors are installed in high-stress areas to detect coal and rock fracture information. Vibration sensors are installed in the main fan bearing housing to detect fan bearing vibration; current transformers are installed in the power distribution cabinet to detect motor power consumption; and absolute encoders are installed on the regulating damper shaft to detect damper opening.

[0018] The online detection module for coal mine roadways uses deployed sensors to detect various gas concentration data, dust distribution data, environmental parameter data, emergency parameter data, and electromechanical equipment status data in the coal mine roadways. Furthermore, the online detection module for coal mine roadways uses deployed sensors to detect various gas concentration data within the roadways, including CH4, CO, CO2, and O2 concentrations; dust distribution data, including PM2.5, PM10, and total dust concentrations; environmental parameter data, including wind speed, temperature and humidity at the mine entrance, temperature and humidity at the mine exit, and ventilation resistance; emergency parameter data, including flue gas temperature, roadway deformation, and coal and rock fracture information, with coal and rock fracture information including acoustic emission event counts, acoustic emission energy, ringing counts, amplitude, rise time, and frequency characteristics; and electromechanical equipment status data, including fan bearing vibration information, motor power consumption, and damper opening, with fan bearing vibration information including XYZ triaxial vibration acceleration, vibration velocity, vibration displacement, high-frequency impact pulses, and bearing characteristic frequencies.

[0019] The data quality verification module performs physical range verification, timeliness verification, spatial consistency verification, equipment status correlation verification, multi-source data mutual verification, and missing value filling on the data detected in the coal mine roadway in sequence before outputting the verification results. Furthermore, the data quality verification module includes a detection data receiving unit, a data verification unit, a missing value imputation unit, a confidence level calculation unit, and a verification result output unit. The detection data receiving unit is used to receive detection data from the coal mine roadway; the data verification unit sequentially performs physical range verification, timeliness verification, spatial consistency verification, equipment status association verification, and multi-source data mutual verification on the data detected in the coal mine roadway; the missing value imputation unit is used to impute missing data points; the confidence level calculation unit is used to calculate the comprehensive confidence level of each detection data in the coal mine roadway; and the verification result output unit is used to output the comprehensive confidence level, quality level, and processing method of each detection data.

[0020] Specifically, in this embodiment, when the overall confidence level of the detection data ∈ [0, 80%), the data quality level is 1, and the data is processed by disabling and triggering a maintenance work order; when the overall confidence level of the detection data ∈ [80%, 95%], the data quality level is 2, and the data is processed by labeling uncertainty before use; when the overall confidence level of the detection data ∈ (95%, 100%), the data quality level is 3, and the data is processed for risk diagnosis. The overall confidence level C of the i-th type of detection data... i The calculation formula is: N pi N ti N si N di N mi The numbers, in order, represent the number of physical anomalies, time-related anomalies, spatial anomalies, equipment-related anomalies, and multi-source contradictions within the detection period of the i-th type of detection data, a. p a t a s a d a m The weighting coefficients for the following categories are, in order: physical anomalies, time-related anomalies, spatial anomalies, equipment-related anomalies, and multi-source conflicts. N zi The total number of data within the detection period for the i-th type of detection data.

[0021] In this embodiment, the specific steps for data verification are as follows: A1. Physical range verification: Check whether the data is within the physical possible range. If the data exceeds the limit, it is marked as invalid data and a sensor fault alarm is triggered. For example, the physical possible range of CH4 concentration is [0%, 100%]. If the CH4 concentration is greater than 100% or less than 0%, the data is marked as exceeding the limit. A2. Timeliness Verification: Check the timestamp to confirm the actual reception time delay and the reception time delay threshold under real-time control requirements. If the actual reception time delay is greater than the reception time delay threshold, start the backup communication channel to verify the data update frequency. Compare the calculated data update frequency with the set data update frequency. If the absolute value of the difference between the two is less than the preset frequency deviation, it is determined that the data update frequency is abnormal and switch to the historical data prediction mode. A3. Spatial Consistency Verification: Check the rationality of gas concentration gradient and temperature gradient. If the gas concentration gradient exceeds the diffusion limit or the temperature gradient exceeds the heat conduction limit, spatial interpolation correction is initiated to verify the airflow direction. That is, if the CH4 concentration in the intake airway is less than the CH4 concentration in the return airway, and the airflow direction is contradictory, ventilation fault diagnosis is triggered. A4. Equipment Status Correlation Verification: Evaluate the sensor health index and match the sensor health level, sensor maintenance strategy, and detection data usage strategy based on the evaluation results; A5. Multi-source data mutual verification: The detection data is sequentially cross-compared with heterogeneous sensors, verified by physical conservation laws, and fused with dynamic weights. Finally, the verified data value, data confidence level, and verification status flag are output.

[0022] like Figure 3 The embodiment shown provides a method for device status association verification, including the following steps: B1. Record the sensor's voltage-related parameters, temperature-related parameters, mechanical vibration-related parameters, communication signal-related parameters, and historical status-related parameters. Voltage-related parameters include the measured voltage value, nominal voltage value, permissible voltage deviation, maximum voltage failure threshold, and minimum voltage failure threshold. Temperature-related parameters include the measured temperature value, nominal temperature value, permissible temperature deviation, maximum temperature failure threshold, and minimum temperature failure threshold. Mechanical vibration-related parameters include the root mean square value of vibration in the 500-1000Hz frequency band and the reference vibration value when the equipment is newly installed. Communication signal-related parameters include the number of successfully received packets, the number of packets to be received, the received signal strength indication, and the critical strength of wireless communication in coal mine roadways. Historical status-related parameters are the sensor's early warning accuracy rate over the past 30 days. B2. Calculate the sensor's voltage deviation, temperature deviation, vibration degradation, communication reliability, and historical reliability. The formula for calculating the voltage deviation ΔV is: where V c V0, δ V V lim,max V lim,min The voltage measurements, nominal voltage values, permissible voltage deviation, maximum voltage failure threshold, and minimum voltage failure threshold are listed in order. The formula for calculating the temperature deviation ΔT is: where T... c 、T0、δ T T lim,maxT lim,min The parameters are, in order: measured temperature value, nominal temperature value, permissible temperature deviation, maximum temperature failure threshold, and minimum temperature failure threshold; the formula for calculating the vibration degradation degree ΔVib is: RMS(f 500-1000Hz ), RMS ref The values ​​are, in order, the root mean square value of vibration in the 500-1000Hz frequency band and the reference vibration value when the equipment is newly installed; the formula for calculating communication reliability ΔCom is: ,m ac m ay RSSI, -85dBm, and I (RSSI < -85dBm) represent, respectively, the number of successfully received packets, the number of packets to be received, the received signal strength indication, the critical strength for wireless communication in coal mine roadways, and the weak signal indication function (=1 if condition is met, =0 otherwise); the formula for calculating historical reliability ΔHis is: , α 30 0.9 represents the accuracy rate of the sensor's early warning over the past 30 days, 0.9 represents the minimum allowable accuracy rate of the coal mine safety system under the AQ1029 standard, and 5 represents the slope adjustment factor for controlling the steepness of the Sigmoid function. B3. Retrieve the failure sensitivity coefficient λ based on voltage deviation, temperature deviation, vibration degradation, communication reliability, and historical reliability. i The values ​​are 8.0, 6.0, 12.0, 5.0, and 3.0, respectively. B4. Determine the basic weights for voltage deviation, temperature deviation, vibration degradation, communication reliability, and historical reliability. 0,i The values ​​are 0.25, 0.2, 0.2, 0.15, and 0.2, respectively. The dynamic weight w is obtained by dynamically adjusting the basic weights based on real-time conditions and then normalizing them. t,i w i,new The new weights for the i-th parameter after dynamic adjustment are as follows: the base weight for voltage deviation increases dynamically by 0.1 when powered by battery, the base weight for temperature deviation increases dynamically by 0.15 when in a high humidity environment (RH>85%), the base weight for vibration degradation increases dynamically by 0.10 when using mobile devices (such as inspection instruments), the base weight for communication reliability increases dynamically by 0.05 when transmitting wirelessly, and the base weight for historical reliability increases dynamically by 0.10 when the equipment has been in service for more than 3 years. B5. Calculate the sensor health index HI, the specific formula is: Δp i Let be the i-th parameter index of the sensor. When i=1, Δp1=ΔV, when i=2, Δp2=ΔT, when i=3, Δp3=ΔVib, when i=4, Δp4=ΔCom, and when i=5, Δp5=ΔHis. B6. When the sensor health index HI∈[0,0.29], the equipment health level is critical, the maintenance strategy is immediate power-off and replacement, and the detection data usage strategy is to trigger an emergency shutdown; when the sensor health index HI∈[0.30,0.49], the equipment health level is poor, the maintenance strategy is to replace within 48 hours, the detection data usage strategy is to disable and switch to a backup sensor for re-detection; when the sensor health index HI∈[0.50,0.69], the equipment health level is medium, the maintenance strategy is on-site diagnosis within 3 days, and the detection data usage strategy is to use it only for trend analysis; when the sensor health index HI∈[0.70,0.84], the equipment health level is good, the maintenance strategy is to calibrate within 7 days, and the detection data usage strategy is to use it with reduced weight (weight = HI); when the sensor health index HI∈[0.85,1], the equipment health level is excellent, the maintenance strategy is routine inspection, and the detection data usage strategy is to use it directly for risk analysis and control decisions.

[0023] The coal mine roadway risk diagnosis module extracts feature data from the online detection data of coal mine roadways that has passed data quality verification and inputs it into the trained logistic regression model to infer the single risk probability and the comprehensive risk probability, and determines whether risk control is needed based on the inference results; Furthermore, the coal mine roadway risk diagnosis module includes a risk event definition unit, a feature processing unit, a single-risk probability inference unit, a comprehensive risk probability calculation unit, a risk diagnosis unit, and a diagnosis result output unit. The risk event definition unit defines six key risk events based on coal mine safety regulations: gas explosion risk, coal dust explosion risk, fire risk, ventilation system risk, structural instability risk, and equipment failure risk. The feature processing unit extracts relevant feature parameters from the online detection data of the coal mine roadway that has passed data quality verification for each risk event. After normalizing all feature parameters to the range [0,1], feature fusion is performed, outputting a feature vector group for each risk event. The single-risk probability inference unit inputs the feature vector group of each risk event into... The trained logistic regression model is used to infer the probability of occurrence of each risk event. The comprehensive risk probability calculation unit aggregates the individual risk probabilities to obtain the comprehensive risk probability. The risk diagnosis unit compares the inferred individual risk probability with the corresponding set control value. If the inferred value is greater than the set control value, it is determined that the probability of occurrence of the risk event needs to be reduced. The comprehensive risk probability calculation value is compared with the set control value. If the calculated value is greater than the set control value, it is determined that the comprehensive risk probability needs to be reduced. The diagnosis result output unit transmits the risk diagnosis results to the risk control plan generation module. The risk diagnosis results include the individual risk probability, the comprehensive risk probability, the control requirements for individual risk events, and the control requirements for comprehensive risk events.

[0024] In this embodiment, it should be specifically noted that e1, e2, e3, e4, e5, and e6 represent the risks of gas explosion, coal dust explosion, fire, ventilation system, structural instability, and equipment failure, respectively. The risk event set E = {e1, e2, e3, e4, e5, e6}. Gas explosion risk is mainly associated with CH4 concentration, O2 concentration, temperature, and wind speed; coal dust explosion risk is mainly associated with dust concentration, O2 concentration, temperature, and wind speed; fire risk is mainly associated with CO concentration, flue gas temperature, and O2 concentration; ventilation system risk is mainly associated with wind speed, ventilation resistance, and inlet / outlet temperature and humidity difference; structural instability risk is mainly associated with roadway deformation and coal / rock fracture information such as acoustic emission event counts and energy; and equipment failure risk is mainly associated with fan vibration information, motor power consumption, and damper opening. The feature processing steps are as follows: For e1 (gas explosion risk): extract CH4 concentration, O2 concentration, average temperature (average of inlet and outlet temperatures), and wind speed; for e2 (coal dust explosion risk): extract PM2.5 concentration, PM10 concentration, O2 concentration, and flue gas temperature; for e5 (structural instability risk): extract acoustic emission event count, acoustic emission energy, and tunnel deformation values ​​(such as displacement); for e6 (equipment failure risk): extract fan X-axis vibration acceleration, high-frequency impact pulse, and motor power consumption. All features are normalized to the range [0,1] to eliminate dimensional influence. The normalization formula used is the minimum-maximum normalization formula, specifically: where F is the original feature value, F... min and F max These are the historical minimum and maximum values ​​of this feature (based on training data). The normalized features are then fused to obtain the risk event e. i Feature vector group F i .

[0025] In this embodiment, it should be specifically noted that a logistic regression model is used because its output probability value range is [0,1], and it is easy to interpret. The model parameters are trained based on historical accident data. For each risk event e... i The probability of occurrence is P(e) i The formula for calculating β0 is: i For risk event e i The intercept term, representing the baseline risk level, reflects the default log probability when there are no feature inputs, β. i For risk event e i The weight vector (e.g., β) 1 =[β1 1 ,β2 1, β k 1 ]), dimension k equals risk event e i The number of features, · represents the dot product, and the model parameters β0 and β... iThe data was trained using historical data, with the training objective being binary classification (event occurrence / non-occurrence). For e1 (gas explosion risk): the feature vector F1 = [CH4 concentration, O2 concentration, average temperature, wind speed], and the formula for calculating the risk event probability Pe1 is:

[0026] In this embodiment, the specific calculation process for the comprehensive risk probability is as follows: C1. Define the base weights w for gas explosion risk, coal dust explosion risk, fire risk, ventilation system risk, structural instability risk, and equipment failure risk based on the probability of death, economic loss coefficient, and production interruption coefficient. ai And calculate the normalized weight w bi The specific formula is: C2. Based on coal mine accident chain analysis, establish a 6×6 correlation matrix R: , element r ij ∈[0,1], representing risk event e i For e j The intensity of the impact; C3. For each risk event e i Calculate the correlation enhancement factor α xi The specific formula is: r ji For risk event e j For e i Influence intensity, P(e j ) is a risk event e j The probability of occurrence is calculated, and the correlation-adjusted probability P(e) is calculated. i ) * The specific formula is: C4. Calculate the weighted overall probability P w The specific formula is: , to calculate the combined probability P of independent events. d The specific formula is: C5. Calculate the final overall risk probability P z The specific formula is: , where a0 is an empirical coefficient, usually taken as a0=0.3; C6. Dynamic Environment Calibration: Obtaining the Production Intensity Coefficient α p , Job type coefficient α w Production intensity coefficient α p The ratio of actual output to designed output is given by the operation type coefficient, which is α when the operation type is tunneling. w =1.2, during the recovery, α w =1.0, during maintenance α w =0.7, calibrating the final probability yields P z,final The specific formula is:

[0027] Risk control scheme generation module: Analyzes the risk diagnosis results and finally generates a risk control scheme that includes target control parameters, the optimal adjustment amount of the target control parameters, the expected effect of single risk probability, and the expected effect of comprehensive risk probability. Furthermore, the risk control scheme generation module includes a data receiving unit, a risk-parameter correlation matrix construction unit, a control demand analysis unit, a parameter control effect prediction unit, a multi-objective optimization solution unit, a control value calibration and verification unit, and a control scheme output unit. The data receiving unit receives risk diagnosis results and sensor data. The risk-parameter correlation matrix construction unit defines the set of controllable parameters, initially correlates risk events with parameters, quantifies the control effectiveness, and performs dynamic correlation verification to obtain the risk-parameter correlation matrix. The control demand analysis unit generates a set of control demands and prioritizes them. The parameter control effect prediction unit predicts the change in the probability of risk events after adjusting each control parameter. The multi-objective optimization solution unit minimizes the comprehensive risk probability and the single risk probability while solving for the optimal parameter adjustment amount. The control value calibration and verification unit calibrates and verifies the optimal parameter adjustment amount of the target control parameter. The control scheme output unit outputs the target control parameter, the optimal adjustment amount of the target control parameter, the expected effect of the single risk probability, and the expected effect of the comprehensive risk probability to the control scheme execution module.

[0028] like Figure 4 This embodiment provides a method for constructing a risk-parameter correlation matrix, including the following steps: D1. Define the adjustable parameter set: List the parameters that can be adjusted in real time in the coal mine roadway, verify the adjustability of the parameters, and output the adjustable parameter set u={u1,u2,...,u...} after unifying the units of the verified parameters into percentages or standard physical units. n The parameters that can be adjusted in real time in coal mine roadways include, but are not limited to, the fan speed u1 and damper opening u2 of the main ventilation system, the spray water volume u3 and nozzle angle u4 of the dust suppression system, the local fan frequency u5 and duct length u6 of the local ventilation equipment, and the motor load distribution u7 and voltage u8 of the power system. The adjustability of the parameters is verified to confirm that the parameters can be adjusted in real time through the automation system or manual commands, and parameters with a response delay of >5 minutes are excluded. D2. Preliminary Correlation between Risk Events and Parameters: The risk event set E={e1,e2,e3,e4,e5,e6} is correlated with the adjustable parameter set u={u1,u2,...,u...} n The initial matrix M is obtained through preliminary correlation. 6×n a For each risk event e i The root cause parameter, u, was determined through fault tree analysis. j For e i If there is no physical influence, then Mij =0; D3. Calculate the initial effectiveness value: Retrieve historical control records and sensor time-series data, and extract all u j Adjustment events and corresponding P(e) i Record changes and exclude interference periods; apply linear regression to continuous parameters to calculate the regulation effectiveness coefficient k. ij The specific formula is: , and the linear regression fitting equation is: , ΔP(e i ), Δu j b0 and b0 are parameters u respectively. j Adjusted risk event e i The probability change and the control parameter u j Adjustment amount, Δu j =0 when ΔP(e i The baseline term is used to calculate the control effectiveness coefficient k by applying the mean effect calculation method to discrete parameters. ij The specific formula is: The calculated regulatory efficiency coefficient is normalized to obtain the initial efficiency value M. ij a The specific formula is: D4. Expert Knowledge Correction: The organization's safety engineers evaluate the matrix values ​​using the Delphi method, correcting contradictory terms to obtain the correction effectiveness value M. ij b ; D5. Dynamic Correlation Verification: Periodically perform small-amplitude perturbation tests on high-risk parameters and record the P(e) after the perturbation. i The stable value of M is then used to recalculate the effectiveness value. ij c The specific formula is: P(e i ) post 、P(e i ) pre u j post u j pre Risk events e after parameter adjustment i Stability probability, risk event e before parameter adjustment i The stability probability and parameter u j Adjusted target value, parameter u j The adjusted target value, when |M ij b -M ij c If a matrix update is triggered when |>0.15, the final effectiveness value M will be... ij d The specific calculation formula is: otherwise, retain the original value, i.e., M. ij d =M ijb .

[0029] In this embodiment, it is specifically necessary to explain that P(e) is used. i ) y This indicates the single-risk probability setting control value, P. z y If the overall risk probability is used to set the control value, then the set of control demand D = {e i |P(e i )>P(e i ) y}∪{“Comprehensive Risk”|P z >P z y}, Risk event control priority coefficient α yi The specific calculation formula is: c0 is the associated risk enhancement coefficient, which is usually taken as an empirical value of 0.3.

[0030] In this embodiment, the specific steps for solving the multi-objective optimization problem are as follows: E1. Optimization Problem Modeling: Determine the decision variables as Δu = {Δu1, Δu2, ..., Δu} n The objective function is , where b1 and g are... i The weighting coefficients are, in order, the overall risk weighting coefficient and the control weighting coefficient for individual risk events. b1 typically takes the default value of 0.6. i The calculation formula is: |D| represents the number of risk events to be regulated, and P... z new 、P(e i ) new The parameters are, in order, the overall risk probability after regulation and the probability of occurrence of a single risk event after regulation. The constraints are defined as parameter range, equipment safety, and risk threshold. The parameter range constraint is: u jmin ≦u jc +Δu j ≦u jmax The equipment safety constraint is: motor power consumption ≤ 110% of rated value, and the risk threshold constraint is: P(e i ) new ≦0.9×P(e i ) y ; E2. Solution: If the optimization problem model is linear, the simplex method is used to solve it. For nonlinear problems, an approximate solution is obtained through iterative solutions. E3. Output the optimal solution: Calculate the optimal adjustment amount Δu * ={Δu1 * ,Δu2 * ,...,Δu n *}, single risk probability expected value and comprehensive risk expected value.

[0031] Risk control plan execution module: Performs dynamic control according to the generated risk control plan.

[0032] Database: Used to store data information for each module in the system.

[0033] In this embodiment, it should be noted that the preset values ​​and set values ​​used are all set based on actual needs or experience, and no specific value limit is imposed here.

[0034] like Figure 2 The embodiment shown provides a method for dynamic adjustment of ventilation in coal mine roadways based on the Internet of Things, including the following steps: S1: Based on the sensor deployment scheme, deploy multiple types of sensors in the coal mine roadway and implement different maintenance strategies for the sensors based on their health levels; S2: Online detection of various gas concentration data, dust distribution data, environmental parameter data, emergency parameter data, and electromechanical equipment status data in coal mine roadways through deployed sensors; S3: Perform physical range verification, timeliness verification, spatial consistency verification, equipment status correlation verification, multi-source data mutual verification, and missing value imputation on the data detected in the coal mine roadway in sequence, and then output the verification results. S4: Extract feature data from the online detection data of coal mine roadways that have passed data quality verification and input it into the trained logistic regression model to infer the single risk probability and the comprehensive risk probability. Based on the inference results, determine whether risk control is needed. S5: Analyze the risk diagnosis results and finally generate a risk control plan that includes target control parameters, the optimal adjustment amount of the target control parameters, the expected effect of single risk probability, and the expected effect of comprehensive risk probability. S6: Dynamically adjust according to the generated risk control plan.

[0035] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A dynamic adjustment system for ventilation in coal mine roadways based on the Internet of Things, characterized in that: include: Sensor control module: Based on the sensor deployment scheme, multiple types of sensors are deployed in the coal mine roadway, and different maintenance strategies are implemented for the sensors based on their health levels; Online monitoring module for coal mine roadways: It uses deployed sensors to monitor various gas concentrations, dust distribution, environmental parameters, emergency parameters, and the status of electromechanical equipment in coal mine roadways online. Data quality verification module: Performs physical range verification, timeliness verification, spatial consistency verification, equipment status correlation verification, multi-source data mutual verification, and missing value filling on the data detected in the coal mine roadway in sequence, and outputs the verification results. Coal mine roadway risk diagnosis module: Extract feature data from online detection data of coal mine roadways that have passed data quality verification and input it into the trained logistic regression model to infer the probability of single risk and the probability of comprehensive risk, and determine whether risk control is needed based on the inference results; Risk control scheme generation module: Analyzes the risk diagnosis results and finally generates a risk control scheme that includes target control parameters, the optimal adjustment amount of the target control parameters, the expected effect of single risk probability, and the expected effect of comprehensive risk probability. Risk control plan execution module: Performs dynamic control according to the generated risk control plan.

2. The dynamic adjustment system for coal mine roadway ventilation based on the Internet of Things as described in claim 1, characterized in that: The sensor control module includes a sensor deployment scheme input unit, a sensor deployment scheme implementation control unit, a sensor health assessment result receiving unit, and a sensor maintenance strategy execution unit. The sensor deployment scheme input unit is used to input the sensor deployment scheme, including sensor ID, model, technical parameters, deployment location, and installation requirements. The sensor deployment scheme implementation control unit is used to ensure that the actual sensor deployment is consistent with the sensor deployment scheme; the sensor health assessment result receiving unit is used to receive the sensor health level and maintenance strategy; and the sensor maintenance strategy execution unit is used to execute the sensor maintenance strategy.

3. The dynamic adjustment system for coal mine roadway ventilation based on the Internet of Things as described in claim 1, characterized in that: The online detection module for coal mine roadways uses deployed sensors to detect various gas concentrations within the roadways, including CH4, CO, CO2, and O2 concentrations; dust distribution data, including PM2.5, PM10, and total dust concentrations; environmental parameter data, including wind speed, temperature and humidity at the mine entrance, temperature and humidity at the mine exit, and ventilation resistance; emergency parameter data, including flue gas temperature, roadway deformation, and coal and rock fracture information, where coal and rock fracture information includes acoustic emission event counts, acoustic emission energy, ringing counts, amplitude, rise time, and frequency characteristics; and electromechanical equipment status data, including fan bearing vibration information, motor power consumption, and damper opening, where fan bearing vibration information includes XYZ triaxial vibration acceleration, vibration velocity, vibration displacement, high-frequency impact pulses, and bearing characteristic frequencies.

4. The dynamic adjustment system for coal mine roadway ventilation based on the Internet of Things as described in claim 1, characterized in that: The data quality verification module includes a detection data receiving unit, a data verification unit, a missing value filling unit, a confidence calculation unit, and a verification result output unit. The detection data receiving unit is used to receive detection data in the coal mine roadway. The data verification unit sequentially performs physical range verification, timeliness verification, spatial consistency verification, equipment status correlation verification, and multi-source data mutual verification on the data detected in the coal mine roadway; the missing value imputation unit is used to impute missing data points; and the confidence calculation unit is used to calculate the comprehensive confidence of each detection data in the coal mine roadway. The verification result output unit is used to output the overall confidence level, quality level, and processing method of each test data.

5. The dynamic adjustment system for coal mine roadway ventilation based on the Internet of Things according to claim 1, characterized in that: The coal mine roadway risk diagnosis module includes a risk event definition unit, a feature processing unit, a single risk probability reasoning unit, a comprehensive risk probability calculation unit, a risk diagnosis unit, and a diagnosis result output unit. The risk event definition unit defines six key risk events in the coal mine safety regulations, namely gas explosion risk, coal dust explosion risk, fire risk, ventilation system risk, structural instability risk, and equipment failure risk. The feature processing unit extracts relevant feature parameters from the online detection data of coal mine roadways that has passed data quality verification for each risk event. After normalizing all feature parameters to the range of [0,1], feature fusion is performed to output the feature vector group of each risk event. The single risk probability inference unit inputs the feature vector group of each risk event into the trained logistic regression model to infer the probability of occurrence of each risk event. The comprehensive risk probability calculation unit is used to aggregate the probabilities of individual risks to obtain the comprehensive risk probability; The risk diagnosis unit compares the inferred value of a single risk probability with the corresponding set control value of the single risk probability. If the inferred value is greater than the set control value, it is determined that the probability of the occurrence of the risk event needs to be reduced. The unit also compares the calculated value of the comprehensive risk probability with the set control value of the comprehensive risk probability. If the calculated value is greater than the set control value, it is determined that the comprehensive risk probability needs to be reduced. The diagnostic result output unit is used to transmit the risk diagnostic results to the risk control plan generation module. The risk diagnostic results include single risk probability, comprehensive risk probability, single risk event control requirements, and comprehensive risk event control requirements.

6. The dynamic adjustment system for coal mine roadway ventilation based on the Internet of Things according to claim 1, characterized in that: The risk control scheme generation module includes a data receiving unit, a risk-parameter correlation matrix construction unit, a control demand analysis unit, a parameter control effect prediction unit, a multi-objective optimization solution unit, a control value calibration and verification unit, and a control scheme output unit. The data receiving unit is used to receive risk diagnosis results and sensor data. After defining the adjustable parameter set, the risk-parameter correlation matrix construction unit initially correlates risk events with parameters, then quantifies the control effectiveness and performs dynamic correlation verification to obtain the risk-parameter correlation matrix; The regulation demand analysis unit is used to generate a set of regulation demands and prioritize them; the parameter regulation effect prediction unit is used to predict the change in the probability of risk events after each regulation parameter is adjusted; the multi-objective optimization solution unit minimizes the comprehensive risk probability and the single risk probability while solving for the optimal parameter adjustment amount. The control value calibration and verification unit is used to calibrate and verify the optimal parameter adjustment amount of the target control parameter; The control scheme output unit is used to output the target control parameters, the optimal adjustment amount of the target control parameters, the expected effect of single risk probability, and the expected effect of comprehensive risk probability to the control scheme execution module.

7. The dynamic adjustment system for coal mine roadway ventilation based on the Internet of Things according to claim 6, characterized in that: The specific steps for constructing the risk-parameter correlation matrix using the risk-parameter correlation matrix construction unit are as follows: D1. Define the adjustable parameter set: List the parameters that can be adjusted in real time in the coal mine roadway, verify the adjustability of the parameters, and output the adjustable parameter set u={u1,u2,...,u...} after unifying the units of the verified parameters into percentages or standard physical units. n }; D2. Preliminary Correlation between Risk Events and Parameters: The risk event set E={e1,e2,e3,e4,e5,e6} is correlated with the adjustable parameter set u={u1,u2,...,u...} n The initial matrix M is obtained through preliminary correlation. 6×n a For each risk event e i The root cause parameter, u, was determined through fault tree analysis. j For e i If there is no physical influence, then M ij =0; D3. Calculate the initial effectiveness value: Retrieve historical control records and sensor time-series data, and extract all u j Adjustment events and corresponding P(e) i Changes are recorded and interference is eliminated. Linear regression is applied to continuous parameters to calculate the regulation efficiency coefficient. The mean effect method is applied to discrete parameters to calculate the regulation efficiency coefficient. The calculated regulation efficiency coefficient is normalized to obtain the initial efficiency value. D4. Expert Knowledge Correction: Safety engineers evaluate the matrix values ​​using the Delphi method, and obtain the corrected effectiveness value after correcting contradictory terms. D5. Dynamic Correlation Verification: Periodically perform small-amplitude perturbation tests on high-risk parameters and record the P(e) after the perturbation. i After stabilizing the value of the element, the power value is recalculated. When the absolute value of the difference between the corrected power value and the recalculated power value of the same element is greater than the preset value, the matrix is ​​updated. The final power value of the element is the average of the sum of the corrected power value and the recalculated power value. Otherwise, the original value is retained, that is, the final power value of the element is the corrected power value.

Images