A coal mining face fire early warning monitoring system and fire prevention method

Abstract

The present invention discloses a coal mining face fire early warning monitoring system and a fire prevention method, which specifically relate to the field of coal mine safety technology. The system divides the tunnel into monitoring areas and obtains detailed data. The system evaluates data accuracy through a time series prediction model, extracts dust and vibration characteristics to evaluate interference when the data is inaccurate, dynamically adjusts the safety threshold and issues a real-time early warning based on the evaluation results and fuzzy logic, and divides the risk level according to the early warning signal and takes corresponding fire prevention measures. This not only improves the accuracy of the monitoring data and the responsiveness of the system, but also effectively prevents safety hazards and ensures the safe and efficient operation of the coal mining face.

Description

Technical Field

[0001] The present invention relates to the technical field of coal mine safety, and in particular to a coal mining face fire prevention early warning monitoring system and a fire prevention method. Background Art

[0002] After more than 20 years of mining, the coal resources of the 3 coal seams of Yunhe Coal Mine have been gradually exhausted, and only the corner blocks of the 7th and 9th mining areas are left. In order to ensure the sustainable development of the mine and extend the service life of the mine, according to the amount of coal recovered by the comprehensive mining and top coal caving technology used in the 3 coal seams of Yunhe Coal Mine in the early production period and the coal pillars left between the original comprehensive mining working faces, after evaluation and demonstration, it was found that there were still 30%-40% of coal resources left in the goaf of the first mining area. It was decided to re-mine the coal resources left in the goaf, which has become an important way for the sustainable survival of the enterprise. In order to ensure the safe and reliable recovery of the coal resources left in the first mining area and prevent the hidden dangers of spontaneous combustion, the comprehensive fire prevention and extinguishing technology of the re-mining working face was studied.

[0003] In the coal mining face, fire prevention warning is to monitor the temperature and gas concentration of the coal rock mass in real time through temperature sensors and gas sensors arranged in the tunnel. When it is detected that the temperature or gas concentration exceeds the preset safety threshold, the system's alarm device will immediately issue an audible and visual alarm to notify the on-site staff. At the same time, the communication module will send the warning signal to the remote monitoring center so that appropriate fire prevention measures can be taken in time to ensure the safety of the working face. However, during the coal mining process, due to the complex environment of the re-mining working face, the temperature and gas sensors may be interfered with, affecting the accuracy of the monitoring data. And if the monitoring data is inaccurate, it may cover up the actual high-risk areas, causing the staff to mistakenly believe that the environment is safe, thereby ignoring the necessary protection and preventive measures, increasing the safety hazards in the operation process. Summary of the invention

[0004] The purpose of the present invention is to provide a coal mining face fire early warning monitoring system and a fire prevention method to solve the deficiencies in the background technology.

[0005] In order to achieve the above-mentioned object, the present invention provides the following technical solutions: a coal mining face fire early warning monitoring system, comprising a data acquisition module, a data processing module, a feature extraction and analysis module, a safety threshold adjustment module and an early warning module;

[0006] Data acquisition module: used to divide the tunnel into several monitoring areas and obtain the temperature and gas concentration data of the coal and rock mass in each monitoring area and the environmental data of the re-mining working face;

[0007] Data processing module: used to divide the acquired coal rock temperature and gas concentration data into different time periods, and evaluate the accuracy of the data by building a time series prediction model;

[0008] Feature extraction and analysis module: When the data is inaccurate, feature extraction is performed on the environmental data of the re-mining working face, and the extracted dust settling rate change characteristics on the sensor surface and the vibration frequency interference characteristics of the coal mining equipment operation on the sensor are analyzed to evaluate the interference degree of the re-mining working face on the sensor;

[0009] Safety threshold adjustment module: Based on the accuracy of the data and the degree of interference to the sensors faced by the re-collection work, the safety threshold pre-set by the early warning system is dynamically adjusted by using fuzzy logic;

[0010] Early warning module: Based on the adjusted safety threshold, early warning is issued for the temperature and gas concentration of the coal rock mass collected in real time.

[0011] Preferably, in the data processing module, the accuracy of the data is evaluated by constructing a time series prediction model, wherein the time series prediction model is an ARIMA model;

[0012] Preprocess the data in each sliding window, use the time period data to train the ARIMA model, apply the trained model to the new time period data, generate predicted values, compare the predicted values ​​with the actual collected data, calculate the prediction error for each time period, statistically calculate the mean and standard deviation of the error, calculate the ratio of the standard deviation of the error to the mean of the error, record it as the data anomaly index, compare the calculated data anomaly index with the data anomaly index reference threshold, if the data anomaly index is greater than or equal to the data anomaly index reference threshold, the data accuracy is low, and a data inaccurate signal is generated; if the data anomaly index is less than the data anomaly index reference threshold, the data accuracy is high, and a data accurate signal is generated.

[0013] Preferably, in the feature extraction and analysis module, when the data is inaccurate, feature extraction is performed on the environmental data of the re-mining working face, and the dust concentration in the tunnel is monitored in real time using a dust sensor. The dust settling rate anomaly index is generated according to the rate change characteristics of dust settling on the sensor surface. The dust settling rate anomaly index is obtained by:

[0014] Dust sensors are evenly arranged in the tunnel to obtain dust concentration data in real time within a time period of s and calculate the change in dust concentration between two consecutive time points. The expression is: ΔC t =C t -C t-1 Among them, C t is the dust concentration at the current moment, C t-1 is the dust concentration at the previous moment, and the dust settling rate is calculated using the expression: Where Δt is the time interval, R t is the dust settling rate, and the average settling rate in the time period s is calculated. The expression is: In the formula, Rˉ is the average sedimentation rate, N is the number of sampling times in the time period, and R i For each sampling sedimentation rate, calculate the sedimentation rate change rate ΔR t , the expression is: ΔR t =R t -Rˉ; After standardizing the sedimentation rate change rate, the dust sedimentation rate anomaly index is calculated, and the expression is: Where, μΔR t is the mean value of the sedimentation rate change rate, σΔR t is the standard deviation of the settling rate change rate, and HK is the dust settling rate anomaly index.

[0015] Preferably, in the feature extraction and analysis module, a vibration sensor is used to monitor the vibration frequency and amplitude of the coal mining face equipment during operation, and a vibration frequency fluctuation index is generated according to the vibration frequency interference characteristics of the sensor caused by the operation of the coal mining equipment. The method for obtaining the vibration frequency fluctuation index is:

[0016] The vibration sensor is arranged near the coal mining face equipment. The vibration sensor collects vibration signals at fixed time intervals. After removing the DC from the vibration signal, the time domain signal of the vibration signal is converted into a frequency domain signal. The frequency component and amplitude of the vibration are extracted. The expression is: Where X(f) is the value of the frequency signal at the fth frequency point, x(n) is the value of the time domain signal at the nth sample point, N is the signal length, and e -j2πfn / N is the kernel function of Fourier transform, j is the imaginary unit; extract the main frequency component and its corresponding amplitude value from the spectrum, select the k frequency components with the highest amplitude in the spectrum, and calculate the rate of change of the main frequency component in a continuous time period. The expression is: Δf t =f t -ef; where Δf t is the frequency change rate, f t is the main frequency at the current moment, ef is the average frequency in the T time period, and the frequency change rate range under the preset standard state is obtained: Δf tmin ~Δf tmax ; Compare the frequency change rate calculated in real time with it and calculate the vibration frequency fluctuation index. If Δf t >Δf tmax , the expression is: DR = |Δf t -Δf tmax |; If Δf t <Δf tmin , the expression is: DR = Δf tmin -Δf t |; Where DR is the vibration frequency fluctuation index.

[0017] Preferably, the dust settling rate anomaly index and the vibration frequency fluctuation index are converted into a first eigenvector, and the first eigenvector is used as the input of the machine learning model. The machine learning model uses each group of first eigenvectors to predict the interference degree value label of the re-mining working face to the sensor as the prediction target, and takes minimizing the sum of prediction errors of all re-mining working face interference degree value labels to the sensor as the training target. The machine learning model is trained until the sum of prediction errors converges and the model training is stopped. The interference degree value of the re-mining working face to the sensor is determined according to the model output results, wherein the machine learning model is a polynomial regression model.

[0018] Preferably, the instantaneous power offset value and the signal frequency fluctuation index are converted into a first eigenvector, and the first eigenvector is used as the input of the machine learning model. The machine learning model uses each group of first eigenvectors to predict the risk value label of invisible faults inside the inverter as the prediction target, and takes minimizing the sum of prediction errors of energy-saving effect value labels under all operating modes as the training target. The machine learning model is trained until the sum of prediction errors converges and the model training is stopped. The risk value of invisible faults inside the inverter is determined according to the model output results, wherein the machine learning model is a polynomial regression model.

[0019] Preferably, in the safety threshold adjustment module, the safety threshold preset by the early warning system is dynamically adjusted using fuzzy logic, specifically:

[0020] The interference degree value and data anomaly index of the sensor faced by the re-mining work are used as the input items of fuzzy logic, and the safety threshold pre-set by the early warning system is used as the output item;

[0021] The interference degree value and data anomaly index of the sensor faced by the re-mining work are fuzzy processed and divided into several fuzzy sets. According to the different combinations of the interference degree value and data anomaly index of the sensor faced by the re-mining work, fuzzy rules are formulated to determine the safety threshold preset by the output item early warning system;

[0022] According to the fuzzy rule base and the interference degree value and data anomaly index of the sensor in the current re-collection work, fuzzy reasoning is performed to obtain the fuzzy output result;

[0023] The fuzzy output result obtained by fuzzy reasoning is defuzzified to obtain a specific safety threshold.

[0024] Preferably, in the early warning module, according to the adjusted safety threshold, an early warning is issued for the temperature and gas concentration of the coal rock mass collected in real time, specifically:

[0025] Start the data acquisition module, ensure that the temperature and gas concentration sensors are working properly, and import the dynamically adjusted safety thresholds. The temperature threshold is T_threshold and the gas concentration threshold is G_threshold.

[0026] The temperature and gas concentration data of the coal rock mass are collected once a minute through sensors; the collected data are transmitted to the early warning module;

[0027] Compare the current temperature data T_current with T_threshold: if T_current>T_threshold, the temperature exceeds the safety threshold;

[0028] Compare the current gas concentration data G_current with G_threshold: if G_current>G_threshold, the gas concentration safety threshold is exceeded;

[0029] If the temperature or gas concentration data exceeds the threshold, the warning level is determined according to the degree of excess, and a warning signal of the corresponding level is generated. If the temperature or gas concentration data exceeds the threshold by less than 10%, a low-level warning signal is triggered; if the temperature or gas concentration data exceeds the threshold by 10%-30%, a medium-level warning signal is triggered; if the temperature or gas concentration data exceeds the threshold by more than 30%, a high-level warning signal is triggered.

[0030] The present invention also provides a coal mining face fire prevention method, comprising the following steps:

[0031] S1: It is used to divide the tunnel into several monitoring areas and obtain the temperature and gas concentration data of the coal and rock mass in each monitoring area and the environmental data of the re-mining working face;

[0032] S2: used to divide the acquired coal rock temperature and gas concentration data into different time periods, and evaluate the accuracy of the data by constructing a time series prediction model;

[0033] S3: When the data is inaccurate, extract the characteristics of the environmental data of the re-mining working face, analyze the characteristics of the rate change of dust settling on the sensor surface and the vibration frequency interference characteristics of the coal mining equipment operation on the sensor, and evaluate the interference degree of the re-mining working face on the sensor;

[0034] S4: Based on the accuracy of the data and the degree of interference to the sensors in the re-collection work, the safety thresholds pre-set by the early warning system are dynamically adjusted using fuzzy logic;

[0035] S5: Based on the adjusted safety threshold, early warning is issued for the temperature and gas concentration of the coal and rock mass collected in real time;

[0036] S6: Divide each monitoring area into risk areas of different levels according to the early warning signals generated in a fixed time period, and take corresponding fire prevention measures according to the risk level.

[0037] Preferably, in S6, each monitoring area is divided into risk areas of different levels according to the early warning signals generated in a fixed time period, and corresponding fire prevention measures are taken according to the risk level, specifically:

[0038] In a fixed time period [t1, t2], the total number of warning signals Si in each monitoring area is calculated, and the expression is: Compare the total number of warning signals obtained within a fixed time period with the gradient warning signal risk ratio threshold, the gradient warning signal risk ratio threshold includes a first risk ratio threshold and a second risk ratio threshold, and the first risk ratio threshold is less than the second risk ratio threshold, and compare the total number of warning signals within the fixed time period with the first risk ratio threshold and the second risk ratio threshold respectively;

[0039] If the total number of warning signals within a fixed time period is greater than the second risk ratio threshold, the possibility of a risk event occurring in the area is high, and the area is classified as a high-risk area;

[0040] If the total number of warning signals within a fixed time period is greater than or equal to the first risk ratio threshold and less than or equal to the second risk ratio threshold, the possibility of a risk event occurring in the area is average, and the area is classified as a medium-risk area;

[0041] If the total number of warning signals within a fixed time period is less than the first risk ratio threshold, the possibility of a risk event occurring in the area is low, and it is classified as a low-risk area.

[0042] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0043] 1. The present invention ensures the efficiency and accuracy of the fire warning system for coal mining working faces through the collaborative work of multiple modules. First, by dividing the tunnel into several monitoring areas and obtaining detailed temperature and gas concentration data, the data accuracy is evaluated in combination with the time series prediction model. When the data is inaccurate, environmental characteristics such as dust settling rate and vibration frequency are extracted for analysis to evaluate the degree of interference. Then, based on the evaluation results, fuzzy logic is used to dynamically adjust the safety threshold of the early warning system, thereby ensuring the reliability and accuracy of the early warning signal during real-time monitoring. The early warning module issues an early warning based on the adjusted threshold and takes corresponding fire prevention measures based on risk signals of different levels.

[0044] 2. The present invention significantly improves the data accuracy and response capability of the fire warning system by refining the monitoring area and accurately processing data, effectively preventing high-risk areas that may be covered up due to inaccurate sensor data, and ensuring timely discovery and response to fire hazards. In addition, by dynamically adjusting the warning threshold and intelligent risk level classification, it can flexibly respond to actual conditions, reduce false alarms and missed alarms, improve the accuracy and timeliness of fire prevention measures, and comprehensively ensure the safe operation of the coal mining face. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0046] Figure 1 This is a module diagram of a coal mining face fire warning monitoring system.

[0047] Figure 2 The present invention is a flow chart of a fire prevention method for a coal mining working face. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0049] Example 1, please refer to Figure 1 As shown, a coal mining face fire early warning monitoring system described in this embodiment includes a data acquisition module, a data processing module, a feature extraction and analysis module, a safety threshold adjustment module and an early warning module;

[0050] Data acquisition module: used to divide the tunnel into several monitoring areas and obtain the temperature and gas concentration data of the coal and rock mass in each monitoring area and the environmental data of the re-mining working face;

[0051] Data processing module: used to divide the acquired coal rock temperature and gas concentration data into different time periods, and evaluate the accuracy of the data by building a time series prediction model;

[0052] Feature extraction and analysis module: When the data is inaccurate, feature extraction is performed on the environmental data of the re-mining working face, and the extracted dust settling rate change characteristics on the sensor surface and the vibration frequency interference characteristics of the coal mining equipment operation on the sensor are analyzed to evaluate the interference degree of the re-mining working face on the sensor;

[0053] Safety threshold adjustment module: Based on the accuracy of the data and the degree of interference to the sensors faced by the re-collection work, the safety threshold pre-set by the early warning system is dynamically adjusted by using fuzzy logic;

[0054] Early warning module: Based on the adjusted safety threshold, early warning is issued for the temperature and gas concentration of the coal rock mass collected in real time.

[0055] Among them, in the data acquisition module, the tunnel is divided into several monitoring areas, and the temperature and gas concentration data of the coal rock mass in each monitoring area and the environmental data of the re-mining working face are obtained, specifically:

[0056] Conduct preliminary survey of the tunnel to determine its length, width, height and tortuosity. Identify key monitoring points, such as turns, ventilation openings, equipment concentration areas, etc.

[0057] The tunnel is divided into several monitoring areas according to its length and structure. The length of each monitoring area can be determined according to the actual situation, generally 20-50 meters. Each area is numbered (such as A1, A2, A3...) to ensure that the area is clearly marked.

[0058] Evenly distribute temperature sensors and gas sensors in each monitoring area to ensure comprehensive monitoring coverage of each area. The density of sensors can be appropriately increased in key areas such as turns, vents, and equipment concentration areas.

[0059] At least one temperature sensor is arranged in each monitoring area. The sensor is installed on the wall or top of the tunnel to avoid being affected by mechanical equipment and personnel activities.

[0060] At least one gas sensor is arranged in each monitoring area to detect the concentration of key gases such as carbon monoxide (CO), methane (CH4) and carbon dioxide (CO2). The sensor should be installed in areas prone to gas accumulation, such as the top of the tunnel or low-lying areas.

[0061] In addition to temperature and gas sensors, dust sensors, humidity sensors and vibration sensors can also be arranged to obtain environmental data of the re-mining working face. Dust sensors are installed in areas prone to dust accumulation, and humidity sensors and vibration sensors are distributed in different locations in the tunnel to ensure the representativeness of the data.

[0062] All sensors collect data in real time and transmit the data to the central data processing unit through wired or wireless networks. The Internet of Things (IoT) technology is used to ensure the reliability and real-time performance of data collection and transmission. The data processing unit stores and preliminarily processes the collected data, including data cleaning and denoising. The data collected by each sensor is timestamped to ensure the time sequence and accuracy of the data.

[0063] Data processing module: used to divide the acquired coal rock temperature and gas concentration data into different time periods, and evaluate the accuracy of the data by building a time series prediction model.

[0064] Use sliding window technology to divide continuous data into multiple time periods. The size and step size of the sliding window can be set according to specific needs. For example, if the sliding window size is 10 minutes and the step size is 1 minute, each time period contains 10 minutes of data and slides once every minute. For each sensor data, it is divided according to the timestamp to ensure that the data in each time period is complete.

[0065] The accuracy of the data is evaluated by building a time series prediction model, where the time series prediction model is an ARIMA model, specifically:

[0066] Preprocess the data in each sliding window, including:

[0067] Missing value processing: To handle missing values ​​in data, interpolation or other data completion methods can be used.

[0068] Data standardization: Standardize or normalize the data to improve model training results.

[0069] Use the ADF (Augmented Dickey-Fuller) test to determine whether the time series data is stationary. If not, perform difference processing to make it stationary. Draw ACF and PACF graphs to determine the parameters p, d, and q of the ARIMA model.

[0070] Using time period data to train the ARIMA model, the expression is:

[0071] PL t =W+T t -(s1T t-1 +s2T t-2 +...s p T t-p +m1PL t-1 +m2PL t-2 +...+m q PL t-q );

[0072] Where PLt is the predicted data of the temperature and gas concentration of the coal rock mass at time point t, W is a constant term, s i and m i are the autoregressive coefficients and lagged error coefficients of the ARIMA model, and p and q are the number of autoregressive terms and lagged error terms, respectively.

[0073] Use common time series evaluation metrics such as mean squared error (MSE), root mean square error (RMSE), or mean absolute error (MAE) to evaluate model performance.

[0074] Apply the trained model to the data of the new time period to generate the predicted value. Compare the predicted value with the actual collected data and calculate the error.

[0075] Calculate the prediction error for each time period, the mean and standard deviation of the statistical error, calculate the ratio of the standard deviation of the error to the mean of the error, and record it as the data anomaly index. Compare the calculated data anomaly index with the data anomaly index reference threshold. If the data anomaly index is greater than or equal to the data anomaly index reference threshold, it means that the data error in this time period is large and the data accuracy is low. At this time, an inaccurate data signal is generated; if the data anomaly index is less than the data anomaly index reference threshold, it means that the data error in this time period is small and the data accuracy is high. At this time, an accurate data signal is generated.

[0076] According to the error analysis results, evaluate the accuracy of the data. If the error is large, further check the sensor and environmental factors. Score the data accuracy for each time period. You can use a 0-1 scoring standard, where 1 means the data is accurate and 0 means the data is inaccurate.

[0077] Feature extraction and analysis module: When the data is inaccurate, feature extraction is performed on the environmental data of the re-mining working face, and the extracted dust settling rate change characteristics on the sensor surface and the vibration frequency interference characteristics of the coal mining equipment operation on the sensor are analyzed to evaluate the degree of interference of the re-mining working face on the sensor.

[0078] When the data is inaccurate, feature extraction is performed on the environmental data of the re-mining working face, and the dust concentration in the tunnel is monitored in real time using a dust sensor. The dust settling rate anomaly index is generated based on the change characteristics of the dust settling rate on the sensor surface. The dust settling rate anomaly index is obtained as follows:

[0079] Dust sensors are evenly arranged in the tunnel to monitor the dust concentration in real time, ensuring that the position of the sensor can represent the overall dust environment in the tunnel. The dust sensor obtains the dust concentration data within the s time period in real time. Each collected data should be timestamped to ensure the time sequence of the data.

[0080] Calculate the dust concentration change between two consecutive time points, the expression is: ΔC t =C t -C t-1 Among them, C t is the dust concentration at the current moment, C t-1 is the dust concentration at the previous moment, and the dust settling rate is calculated using the expression: Where Δt is the time interval, R t is the dust settling rate, the unit can be mg / m3 / min, calculate the average settling rate in the time period s, the expression is: In the formula, Rˉ is the average sedimentation rate, N is the number of sampling times in the time period, and R i For each sampling sedimentation rate, calculate the sedimentation rate change rate ΔR t , reflecting the dynamic changes of the dust environment, the expression is: ΔR t =R t -Rˉ; Standardize the sedimentation rate change rate to eliminate the difference between different time periods and calculate the dust sedimentation rate anomaly index, the expression is: Where, μΔR t is the mean value of the sedimentation rate change rate, σΔR t is the standard deviation of the settling rate change rate, and HK is the dust settling rate anomaly index.

[0081] The larger the dust settling rate anomaly index is, the more serious the interference to the sensor will be. A high anomaly index indicates that the dust concentration has changed dramatically in a short period of time. This change may cause dust to quickly settle on the sensor surface, affecting the normal function of the sensor. For example, dust covering the sensor surface will cause the sensor to be unable to accurately detect temperature and gas concentration, thus affecting the reliability of the fire warning system.

[0082] As the dust settling rate abnormality index increases, the sensor may have unstable readings or large reading deviations. This will not only reduce the accuracy of the sensor data, but may also lead to false or missed fire risks. If the sensor frequently has erroneous data, the system may not be able to issue early warning signals in time, thereby increasing the risk of fire and endangering mine safety.

[0083] Therefore, the dust settling rate abnormality index is an important indicator for evaluating the degree of sensor interference faced by re-mining work. Regular monitoring and analysis of this index can help timely discover and solve the problem of dust interference on sensors and ensure the effective operation of the fire warning system. If necessary, the impact of dust on sensors can be reduced by cleaning sensors, optimizing ventilation systems, and other measures to improve the overall reliability and safety of the system.

[0084] The vibration frequency and amplitude of the coal mining working face equipment during operation are monitored by vibration sensors. The vibration frequency fluctuation index is generated according to the vibration frequency interference characteristics of the coal mining equipment operation on the sensor. The method for obtaining the vibration frequency fluctuation index is as follows:

[0085] Vibration sensors are placed near the coal mining face equipment to ensure accurate monitoring of the vibration frequency and amplitude of the equipment during operation. The vibration sensors collect vibration signals at fixed time intervals (such as every second). Each collected data should be timestamped to ensure the time sequence of the data.

[0086] The vibration signal is processed to remove DC to ensure that the center of the signal is at zero. The vibration signal is filtered to remove high-frequency and low-frequency noise and retain the effective signal.

[0087] Convert the time domain signal into a frequency domain signal and extract the frequency component and amplitude of the vibration. The expression is: Where X(f) is the value of the frequency signal at the fth frequency point, x(n) is the value of the time domain signal at the nth sample point, N is the signal length, and e -j2πfn / N is the kernel function of Fourier transform, j is the imaginary unit; extract the main frequency component and its corresponding amplitude value from the spectrum, select the k frequency components with the highest amplitude in the spectrum, and calculate the rate of change of the main frequency component in a continuous time period. The expression is: Δf t =f t -ef; where Δf t is the frequency change rate, f t is the main frequency at the current moment, ef is the average frequency in the T time period, and the frequency change rate range under the preset standard state is obtained: Δf tmin ~Δf tmax ; Compare the frequency change rate calculated in real time with it and calculate the vibration frequency fluctuation index. If Δf t >Δf tmax , the expression is: DR = |Δf t -Δf tmax |; If Δf t <Δf tmin , the expression is: DR = |Δf tmin -Δf t |; Where DR is the vibration frequency fluctuation index.

[0088] The greater the vibration frequency fluctuation index, the more serious the interference to the sensor during the re-mining work. This means that the vibration frequency of the equipment during operation has changed significantly compared to normal conditions, which may cause unstable or erroneous data readings from the sensor, thus affecting the accuracy and reliability of the sensor. For example, in the case of large frequency fluctuations, the sensor may not be able to accurately detect temperature and gas concentration, resulting in false alarms or missed fire risks, endangering mine safety.

[0089] As the vibration frequency fluctuation index increases, the possibility and degree of sensor being affected by vibration also increases. This interference will not only cause data fluctuations, but may also cause sensor hardware failures or accelerate its aging, further affecting the effectiveness of the fire warning system. If the sensor is frequently subjected to high-intensity vibration interference, the system's early warning capability will be greatly reduced, increasing the risk of fire hazards.

[0090] Therefore, real-time monitoring and analysis of the vibration frequency fluctuation index is crucial to assess the degree of interference to sensors during re-mining. By timely identifying and responding to these interferences, appropriate measures can be taken, such as optimizing equipment operation or enhancing sensor protection, to ensure the normal operation of the fire warning system and safe production in the mine.

[0091] The dust settling rate anomaly index and the vibration frequency fluctuation index are converted into the first eigenvector, and the first eigenvector is used as the input of the machine learning model. The machine learning model uses each group of first eigenvectors to predict the interference degree value label of the re-mining working face to the sensor as the prediction target, and takes minimizing the sum of prediction errors of all re-mining working face interference degree value labels to the sensor as the training target. The machine learning model is trained until the sum of prediction errors converges and the model training is stopped. The interference degree value of the re-mining working face to the sensor is determined according to the model output results, wherein the machine learning model is a polynomial regression model.

[0092] The method for obtaining the interference degree value of the re-mining work on the sensor is: from the first eigenvector training data of the trained machine learning model, obtain the corresponding function expression: CR=f1(SQ, Eθ); where f1 is the output function of the model, HK is the dust settling rate anomaly index, DR is the vibration frequency fluctuation index, and CR is the interference degree value of the re-mining work on the sensor.

[0093] Safety threshold adjustment module: Based on the accuracy of the data and the degree of interference to the sensors faced by the re-collection work, the safety threshold pre-set by the early warning system is dynamically adjusted using fuzzy logic.

[0094] The interference degree value CR and data anomaly index DT of the sensor faced by the re-mining work are used as the input items of fuzzy logic, and the safety threshold value AH pre-set by the early warning system is used as the output item;

[0095] CR and DT are fuzzified and divided into several fuzzy sets. For example, CR can be divided into three fuzzy sets: "low", "medium", and "high", and DT can be divided into three fuzzy sets: "normal", "abnormal", and "severely abnormal".

[0096] Establish a fuzzy rule base: According to different combinations of CR and DT, formulate fuzzy rules to determine the output item AH. For example: if CR is low and DT is normal, AH remains the same; if CR is low and DT is abnormal, AH decreases a little; if CR is medium and DT is normal, AH remains the same; if CR is medium and DT is abnormal, AH decreases a lot; if CR is high and DT is normal, AH increases a little; if CR is high and DT is abnormal, AH decreases significantly; if CR is high and DT is seriously abnormal, AH decreases to the minimum.

[0097] According to the fuzzy rule base and the current CR and DT values, fuzzy reasoning is performed to obtain the fuzzy output result.

[0098] The fuzzy output result obtained by fuzzy reasoning is defuzzified to obtain a specific AH value. For example, a defuzzification method such as the centroid method and the maximum membership method can be used.

[0099] According to the AH value obtained by defuzzification, the pre-set safety threshold of the early warning system is dynamically adjusted to ensure the safety and accuracy of the system under different interference levels and data anomalies.

[0100] Early warning module: Based on the adjusted safety threshold, early warning is issued for the temperature and gas concentration of the coal rock mass collected in real time.

[0101] Start the sensor data acquisition module and ensure that the temperature and gas concentration sensors are working properly.

[0102] Import the dynamically adjusted safety thresholds, where the temperature threshold is T_threshold and the gas concentration threshold is G_threshold.

[0103] The temperature and gas concentration data of the coal rock mass are collected once a minute through sensors, and the collected data are transmitted to the early warning module.

[0104] The collected temperature and gas concentration data are denoised to eliminate sensor noise. The sliding average algorithm is used to smooth the data and reduce data fluctuations.

[0105] Compare the current temperature data T_current with T_threshold: if T_current>T_threshold, the temperature exceeds the safety threshold.

[0106] Compare the current gas concentration data G_current with G_threshold: if G_current>G_threshold, the gas concentration safety threshold is exceeded.

[0107] If the temperature or gas concentration data exceeds the threshold, the warning level is determined according to the degree of excess and a warning signal of the corresponding level is generated, specifically:

[0108] If the threshold is exceeded by less than 10%, a low-level warning signal is triggered; if the threshold is exceeded by 10%-30%, a medium-level warning signal is triggered; if the threshold is exceeded by more than 30%, a high-level warning signal is triggered.

[0109] Warning response:

[0110] Low-level warning: record the event, send an alert to the monitoring personnel, and recommend observation.

[0111] Intermediate warning: Record the event, send an alarm and notify the person in charge to conduct an inspection.

[0112] Advanced early warning: record the event, send an alarm, immediately notify the responsible person and activate the emergency plan.

[0113] Record warning event information and generate detailed reports, including time, location, temperature, gas concentration, warning level, etc. Regularly summarize warning data, analyze system performance, and make improvement suggestions. Regularly check sensors and warning systems to ensure their normal operation. Adjust safety thresholds and warning rules based on the latest data to optimize warning system performance.

[0114] In this embodiment, the tunnel is divided into several monitoring areas, and the temperature and gas concentration data of the coal and rock mass in each monitoring area and the environmental data of the re-mining working face are obtained; the obtained temperature and gas concentration data of the coal and rock mass are divided into different time periods, and the accuracy of the data is evaluated by constructing a time series prediction model; when the data is inaccurate, the environmental data of the re-mining working face is feature extracted, and the extracted dust settling rate change characteristics on the sensor surface and the vibration frequency interference characteristics of the coal mining equipment operation on the sensor are analyzed to evaluate the degree of interference of the re-mining working face on the sensor; according to the accuracy of the data and the degree of interference of the re-mining working face on the sensor, the pre-set safety threshold of the early warning system is dynamically adjusted after using fuzzy logic; according to the adjusted safety threshold, the temperature and gas concentration of the coal and rock mass collected in real time are warned. This not only significantly improves the accuracy and responsiveness of the early warning system, but also effectively ensures the safety and stability of the coal mine working environment.

[0115] Example 2, please refer to Figure 2 As shown, a coal mining face fire prevention method described in this embodiment includes the following steps:

[0116] S1: It is used to divide the tunnel into several monitoring areas and obtain the temperature and gas concentration data of the coal and rock mass in each monitoring area and the environmental data of the re-mining working face;

[0117] S2: used to divide the acquired coal rock temperature and gas concentration data into different time periods, and evaluate the accuracy of the data by constructing a time series prediction model;

[0118] S3: When the data is inaccurate, extract the characteristics of the environmental data of the re-mining working face, analyze the characteristics of the rate change of dust settling on the sensor surface and the vibration frequency interference characteristics of the coal mining equipment operation on the sensor, and evaluate the interference degree of the re-mining working face on the sensor;

[0119] S4: Based on the accuracy of the data and the degree of interference to the sensors in the re-collection work, the safety thresholds pre-set by the early warning system are dynamically adjusted using fuzzy logic;

[0120] S5: Based on the adjusted safety threshold, an early warning is issued for the temperature and gas concentration of the coal rock mass collected in real time.

[0121] S6: Divide each monitoring area into risk areas of different levels according to the early warning signals generated in a fixed time period, and take corresponding fire prevention measures according to the risk level.

[0122] Among them, in S6, each monitoring area is divided into risk areas of different levels according to the early warning signals generated in a fixed time period, and corresponding fire prevention measures are taken according to the risk level, specifically:

[0123] In a fixed time period [t1, t2], the total number of warning signals Si in each monitoring area is calculated, and the expression is: Compare the total number of warning signals obtained within a fixed time period with the gradient warning signal risk ratio threshold, the gradient warning signal risk ratio threshold includes a first risk ratio threshold and a second risk ratio threshold, and the first risk ratio threshold is less than the second risk ratio threshold, and compare the total number of warning signals within the fixed time period with the first risk ratio threshold and the second risk ratio threshold respectively;

[0124] If the total number of warning signals within a fixed time period is greater than the second risk ratio threshold, it means that the possibility of risk events occurring in the area is high, and it is classified as a high-risk area;

[0125] If the total number of warning signals within a fixed time period is greater than or equal to the first risk ratio threshold and less than or equal to the second risk ratio threshold, it means that the possibility of risk events occurring in the area is average, and it is classified as a medium-risk area;

[0126] If the total number of warning signals within a fixed time period is less than the first risk ratio threshold, it means that the possibility of risk events occurring in the area is low, and it is classified as a low-risk area.

[0127] According to the risk level of each monitoring area, take corresponding fire prevention measures:

[0128] High-risk areas: Immediately initiate emergency response measures, such as evacuating personnel, shutting down equipment, and activating fire extinguishing systems. Increase the frequency of inspections to ensure that problems are discovered and addressed in a timely manner.

[0129] Medium-risk areas: Strengthen monitoring, increase the frequency of use of ventilation equipment, and reduce temperature and gas concentration. Reduce or suspend high-risk operations in the area and assign dedicated personnel to monitor.

[0130] Low-risk areas: Maintain normal operations, conduct regular routine inspections, continue monitoring but no special response is required. Record early warning signs and reassess regional risks when necessary.

[0131] In this embodiment, by calculating the total number of warning signals in each monitoring area within a fixed time period and comparing it with the gradient warning signal risk ratio threshold, it is divided into high, medium and low risk areas. Emergency response measures are immediately initiated in high-risk areas; medium-risk areas strengthen monitoring and ventilation to reduce high-risk operations; low-risk areas maintain normal operations and conduct regular inspections. This solution effectively improves regional safety management and emergency response capabilities by accurately dividing risk levels and taking targeted fire prevention measures.

[0132] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0133] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0134] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0135] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0136] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0137] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A coal mining face fire early warning monitoring system, characterized in that: It includes data acquisition module, data processing module, feature extraction and analysis module, safety threshold adjustment module and early warning module; Data acquisition module: used to divide the tunnel into several monitoring areas and obtain the temperature and gas concentration data of the coal and rock mass in each monitoring area and the environmental data of the re-mining working face; Data processing module: used to divide the acquired coal rock temperature and gas concentration data into different time periods, and evaluate the accuracy of the data by building a time series prediction model; Feature extraction and analysis module: When the data is inaccurate, feature extraction is performed on the environmental data of the re-mining working face, and the extracted dust settling rate change characteristics on the sensor surface and the vibration frequency interference characteristics of the coal mining equipment operation on the sensor are analyzed to evaluate the interference degree of the re-mining working face on the sensor; Safety threshold adjustment module: Based on the accuracy of the data and the degree of interference to the sensors faced by the re-collection work, the safety threshold pre-set by the early warning system is dynamically adjusted by using fuzzy logic; Early warning module: Based on the adjusted safety threshold, early warning is issued for the temperature and gas concentration of the coal rock mass collected in real time.

2. A coal mining face fire early warning monitoring system according to claim 1, characterized in that: In the data processing module, the accuracy of the data is evaluated by building a time series prediction model, where the time series prediction model is an ARIMA model; Preprocess the data in each sliding window, use the time period data to train the ARIMA model, apply the trained model to the new time period data, generate predicted values, compare the predicted values ​​with the actual collected data, calculate the prediction error for each time period, statistically calculate the mean and standard deviation of the error, calculate the ratio of the standard deviation of the error to the mean of the error, record it as the data anomaly index, compare the calculated data anomaly index with the data anomaly index reference threshold, if the data anomaly index is greater than or equal to the data anomaly index reference threshold, the data accuracy is low, and a data inaccurate signal is generated; if the data anomaly index is less than the data anomaly index reference threshold, the data accuracy is high, and a data accurate signal is generated.

3. A coal mining face fire early warning monitoring system according to claim 1, characterized in that: In the feature extraction and analysis module, when the data is inaccurate, feature extraction is performed on the environmental data of the re-mining working face, and the dust concentration in the tunnel is monitored in real time using a dust sensor. The dust settling rate anomaly index is generated based on the rate change characteristics of dust settling on the sensor surface. The dust settling rate anomaly index is obtained as follows: Dust sensors are evenly arranged in the tunnel to obtain dust concentration data in real time within a time period of s and calculate the change in dust concentration between two consecutive time points. The expression is: ΔC t =C t -C t-1 ; Among them, C t is the dust concentration at the current moment, C t-1 is the dust concentration at the previous moment, and the dust settling rate is calculated using the expression: Where Δt is the time interval, R t is the dust settling rate, and the average settling rate in the time period s is calculated. The expression is: In the formula, R - is the average sedimentation rate, N is the number of sampling times in the time period, R i For each sampling sedimentation rate, calculate the sedimentation rate change rate ΔR t , the expression is: ΔR t =R t -R - ; After standardizing the sedimentation rate change rate, the dust sedimentation rate anomaly index is calculated, and the expression is: Where, μΔR t is the mean value of the sedimentation rate change rate, σΔR t is the standard deviation of the settling rate change rate, and HK is the dust settling rate anomaly index.

4. A coal mining face fire early warning monitoring system according to claim 3, characterized in that: In the feature extraction and analysis module, the vibration sensor is used to monitor the vibration frequency and amplitude of the coal mining working face equipment during operation. The vibration frequency fluctuation index is generated according to the vibration frequency interference characteristics of the sensor caused by the operation of the coal mining equipment. The method for obtaining the vibration frequency fluctuation index is as follows: The vibration sensor is arranged near the coal mining face equipment. The vibration sensor collects vibration signals at fixed time intervals. After removing the DC from the vibration signal, the time domain signal of the vibration signal is converted into a frequency domain signal. The frequency component and amplitude of the vibration are extracted. The expression is: Where X(f) is the value of the frequency signal at the fth frequency point, x(n) is the value of the time domain signal at the nth sample point, N is the signal length, and e -j2πfn / N is the kernel function of Fourier transform, j is the imaginary unit; extract the main frequency component and its corresponding amplitude value from the spectrum, select the k frequency components with the highest amplitude in the spectrum, and calculate the rate of change of the main frequency component in a continuous time period. The expression is: Δf t =f t -ef; where Δf t is the frequency change rate, f t is the main frequency at the current moment, ef is the average frequency in the T time period, and the frequency change rate range under the preset standard state is obtained: Δft min ~Δf tmax ; Compare the frequency change rate calculated in real time with it and calculate the vibration frequency fluctuation index. If Δf t >Δf tmax , the expression is: DR = |Δf t -Δf tmax |; If Δf t <Δf tmin , the expression is: DR = |Δf tmin -Δf t |; Where DR is the vibration frequency fluctuation index.

5. A coal mining face fire early warning monitoring system according to claim 4, characterized in that: The dust settling rate anomaly index and the vibration frequency fluctuation index are converted into the first eigenvector, and the first eigenvector is used as the input of the machine learning model. The machine learning model uses each group of first eigenvectors to predict the interference degree value label of the re-mining working face to the sensor as the prediction target, and takes minimizing the sum of prediction errors of all re-mining working face interference degree value labels to the sensor as the training target. The machine learning model is trained until the sum of prediction errors converges and the model training is stopped. The interference degree value of the re-mining working face to the sensor is determined according to the model output results, wherein the machine learning model is a polynomial regression model.

6. A coal mining face fire early warning monitoring system according to claim 5, characterized in that: The instantaneous power offset value and the signal frequency fluctuation index are converted into the first eigenvector, and the first eigenvector is used as the input of the machine learning model. The machine learning model predicts the risk value label of the invisible fault inside the inverter for each group of first eigenvectors as the prediction target, and takes minimizing the sum of the prediction errors of the energy-saving effect value labels under all operating modes as the training target. The machine learning model is trained until the sum of the prediction errors converges and the model training is stopped. The risk value of the invisible fault inside the inverter is determined according to the model output results, wherein the machine learning model is a polynomial regression model.

7. A coal mining face fire early warning monitoring system according to claim 1, characterized in that: In the safety threshold adjustment module, fuzzy logic is used to dynamically adjust the safety threshold pre-set by the early warning system, specifically: The interference degree value and data anomaly index of the sensor faced by the re-mining work are used as the input items of fuzzy logic, and the safety threshold pre-set by the early warning system is used as the output item; The interference degree value and data anomaly index of the sensor faced by the re-mining work are fuzzy processed and divided into several fuzzy sets. According to the different combinations of the interference degree value and data anomaly index of the sensor faced by the re-mining work, fuzzy rules are formulated to determine the safety threshold preset by the output item early warning system; According to the fuzzy rule base and the interference degree value and data anomaly index of the sensor in the current re-collection work, fuzzy reasoning is performed to obtain the fuzzy output result; The fuzzy output result obtained by fuzzy reasoning is defuzzified to obtain a specific safety threshold.

8. A coal mining face fire early warning monitoring system according to claim 7, characterized in that: In the early warning module, according to the adjusted safety threshold, the temperature and gas concentration of the coal rock mass collected in real time are warned, specifically: Start the data acquisition module, ensure that the temperature and gas concentration sensors are working properly, and import the dynamically adjusted safety thresholds. The temperature threshold is T_threshold and the gas concentration threshold is G_threshold. The temperature and gas concentration data of the coal rock mass are collected once a minute through sensors; the collected data are transmitted to the early warning module; Compare the current temperature data T_current with T_threshold: if T_current>T_threshold, the temperature exceeds the safety threshold; Compare the current gas concentration data G_current with G_threshold: if G_current>G_threshold, the gas concentration safety threshold is exceeded; If the temperature or gas concentration data exceeds the threshold, the warning level is determined according to the degree of excess, and a warning signal of the corresponding level is generated. If the temperature or gas concentration data exceeds the threshold by less than 10%, a low-level warning signal is triggered; if the temperature or gas concentration data exceeds the threshold by 10%-30%, a medium-level warning signal is triggered; if the temperature or gas concentration data exceeds the threshold by more than 30%, a high-level warning signal is triggered.

9. A method for preventing fire in a coal mining face, implemented based on the early warning monitoring system according to any one of claims 1 to 8, characterized in that: The fire prevention method comprises the following steps: S1: It is used to divide the tunnel into several monitoring areas and obtain the temperature and gas concentration data of the coal and rock mass in each monitoring area and the environmental data of the re-mining working face; S2: used to divide the acquired coal rock temperature and gas concentration data into different time periods, and evaluate the accuracy of the data by constructing a time series prediction model; S3: When the data is inaccurate, extract the characteristics of the environmental data of the re-mining working face, analyze the characteristics of the rate change of dust settling on the sensor surface and the vibration frequency interference characteristics of the coal mining equipment operation on the sensor, and evaluate the interference degree of the re-mining working face on the sensor; S4: Based on the accuracy of the data and the degree of interference to the sensors in the re-collection work, the safety thresholds pre-set by the early warning system are dynamically adjusted using fuzzy logic; S5: Based on the adjusted safety threshold, early warning is issued for the temperature and gas concentration of the coal and rock mass collected in real time; S6: Divide each monitoring area into risk areas of different levels according to the early warning signals generated in a fixed time period, and take corresponding fire prevention measures according to the risk level.

10. A method for preventing fire in a coal mining face according to claim 9, characterized in that: In S6, each monitoring area is divided into risk areas of different levels according to the early warning signals generated in a fixed time period, and corresponding fire prevention measures are taken according to the risk level, specifically: In a fixed time period [t1, t2], the total number of warning signals Si in each monitoring area is calculated, and the expression is: Compare the total number of warning signals obtained within a fixed time period with the gradient warning signal risk ratio threshold, the gradient warning signal risk ratio threshold includes a first risk ratio threshold and a second risk ratio threshold, and the first risk ratio threshold is less than the second risk ratio threshold, and compare the total number of warning signals within the fixed time period with the first risk ratio threshold and the second risk ratio threshold respectively; If the total number of warning signals within a fixed time period is greater than the second risk ratio threshold, the possibility of a risk event occurring in the area is high, and the area is classified as a high-risk area; If the total number of warning signals within a fixed time period is greater than or equal to the first risk ratio threshold and less than or equal to the second risk ratio threshold, the possibility of a risk event occurring in the area is average, and the area is classified as a medium-risk area; If the total number of warning signals within a fixed time period is less than the first risk ratio threshold, the possibility of a risk event occurring in the area is low, and it is classified as a low-risk area.

Images