A method for in-situ identification of coal spontaneous combustion risk in a working face goaf
Patent Information
- Application Number
- CN202611168714.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2026-05-29
- Filing Date
- 2026-08-04
- Publication Date
- 2026-10-27
AI Technical Summary
[0004]针对上述中的相关技术，单凭神经网络模型拟合得到的气体浓度预测值具有较强的经验主义，其预测值无法准确表示气体变化趋势，并且采用固定的安全阈值与预测值进行比较判断，无法适应不同作业区域、不同时间段、不同开采阶段的实际差异，对采空区煤自燃风险做出的判识准确性较低
本申请基于采集的气体浓度历史数据构建相应的数据集，利用样条分位数模型SQM处理数据集，拟合数据的分布结果，得到中位线回归函数，动态生成适配当前具体工况的"标准浓度值"，并通过对应气体实时浓度值与中位线回归函数的比值进行确定气体的趋势范围，得到对应气体的预警程度，实现更加准确精细的判断当前气体浓度偏离正常状态的程度，提高了对采空区气体变化趋势预测的准确性，有利于保障对采空区煤自燃风险的准确判识；
Smart Images
Figure CN122983307A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of coal mine safety, and in particular to a method for on-site identification of the risk of spontaneous combustion of coal in the goaf of a working face. Background Technology
[0002] In coal mine safety operations, assessing the risk of spontaneous combustion of coal in the goaf requires considering both the mechanism of spontaneous combustion and the actual conditions of the goaf. Goafs contain various environmental indicator gases, such as carbon monoxide and ethylene, and the emission characteristics of these gases are closely related to the degree of spontaneous combustion. By monitoring changes in the concentration of these gases within the goaf, the level of risk of spontaneous combustion can be determined.
[0003] In related technologies, such as Figure 1 As shown, currently, the main method relies on various sensor devices (such as gas sensors, temperature and humidity sensors, pressure sensors, etc.) to monitor the environment of the goaf in real time, obtain the corresponding monitoring data, and train the model by inputting the previous monitoring data records into a time-series neural network model (such as a Long Short-Term Memory neural network model) to fit the gas change pattern and obtain the predicted value of gas concentration.
[0004] Regarding the aforementioned technologies, the gas concentration prediction values obtained solely through neural network model fitting are highly empirical. These prediction values cannot accurately represent the gas change trend, and the comparison and judgment between the prediction values and fixed safety thresholds cannot adapt to the actual differences in different operating areas, time periods, and mining stages. Consequently, the accuracy of identifying the risk of spontaneous combustion of coal in goaf areas is relatively low. Summary of the Invention
[0005] To improve the accuracy of predicting gas change trends in goaf areas and ensure accurate identification of coal spontaneous combustion risks in goaf areas, this invention provides a method for on-site identification of coal spontaneous combustion risks in working face goaf areas.
[0006] The present invention adopts the following technical solution: A method for on-site identification of coal spontaneous combustion risk in goaf of working face, comprising: S1, collects gas from the goaf area using gas sampling equipment and uploads the corresponding gas concentration data to the server; S2, the server constructs a corresponding dataset based on the collected historical gas concentration data, processes the dataset using a Spline Quantile Model (SQM), fits the data distribution, and obtains the median regression function. ; This represents the regression function; x is the distance variable, representing the distance x meters from the working face within the goaf area; S3, when the distance variable is x, the real-time gas concentration value collected by monitoring is C, calculate the regression function of C with the median. The ratio P; S4, the server has a built-in gas ratio range setting module. The gas ratio range setting module has the function of dividing the corresponding gas trend range, and can also divide the warning level of coal spontaneous combustion risk in the goaf according to the trend range of the corresponding gas. S5, the gas ratio range setting module determines the trend range of the corresponding gas through the P value obtained in S3, and then obtains the early warning level of the risk of spontaneous combustion of coal in the goaf.
[0007] Optionally, in step S1, the gas sampling device has a built-in communication module for processing gas concentration data and uploading it to a server; the server has a built-in device management module for recording and storing gas concentrations; the communication module is electrically connected to the device management module.
[0008] Optionally, the device management module is designed and written based on the Spring Boot framework, which can build a dataset from the collected historical gas concentration data; and the device management module is associated with the spline quantile model SQM to work in conjunction with the device management module.
[0009] Optionally, in S2, when the server constructs the corresponding dataset based on the collected historical gas concentration data, the historical data shall be at least 40 days of gas concentration data.
[0010] In summary, this application includes at least one of the following beneficial technical effects: This application constructs a corresponding dataset based on collected historical gas concentration data, processes the dataset using the spline quantile model (SQM), fits the data distribution results, and obtains the median regression function. It dynamically generates "standard concentration values" that are adapted to the current specific working conditions, and determines the trend range of the gas by the ratio of the real-time concentration value of the corresponding gas to the median regression function, thereby obtaining the warning level of the corresponding gas. This enables a more accurate and precise judgment of the degree to which the current gas concentration deviates from the normal state, improves the accuracy of predicting the gas change trend in the goaf, and helps to ensure the accurate identification of the risk of spontaneous combustion of coal in the goaf. Traditional monitoring methods only issue alarms when gas concentrations exceed fixed thresholds, representing a typical passive response mode. This application uses a median regression function as a reference system and, based on mine conditions, defines customized concentration range classifications to achieve a refined classification assessment of the safety status of spontaneous combustion of coal in goaf areas. It possesses the ability to proactively identify extreme risks, realizing an early warning function that moves from passive response to proactive prevention. This provides ample preparation for taking preventative measures such as adjusting ventilation and controlling work intensity, thus more effectively ensuring coal mine safety. Attached Figure Description
[0011] Figure 1 This is a schematic diagram of the process of predicting gas concentration using a neural network model in the background technology.
[0012] Figure 2 This is a flowchart illustrating a method for on-site identification of coal spontaneous combustion risk in a working face goaf according to an embodiment of this application.
[0013] Figure 3 This is a schematic diagram of the control system of this application.
[0014] Figure 4 This is a median regression function curve of carbon monoxide in the embodiments of this application.
[0015] Figure 5 This is a median regression function curve of oxygen in an embodiment of this application. Detailed Implementation
[0016] The following combination Figures 2 to 4 This application will be described in further detail.
[0017] This application discloses a method for on-site identification of the risk of spontaneous combustion of coal in the goaf of a working face. Example
[0018] Reference Figure 2 and Figure 3 A method for on-site identification of coal spontaneous combustion risk in goaf areas of working faces, comprising: S1. Gas is collected from the goaf using a gas sampling device, and the corresponding gas concentration data is uploaded to a server. The gas sampling device can be a bundled tube monitoring system; for example, in this embodiment, the third-generation downhole bundled tube gas sampling device produced by Jiangsu Ji'an Mining Technology Co., Ltd. is used. The server in this embodiment uses an NVIDIA 4060ti graphics card and an Intel 14th generation Core i9 CPU. The gas sampling device has a built-in communication module, which can be a wired or wireless communication module. In this embodiment, the communication module is an E103-W04B module, which can be used to process gas concentration data and communicate with the server or other modules wirelessly. The server has a built-in device management module to record and store gas concentrations. The communication module is electrically connected to the device management module.
[0019] The device management module is designed and written based on the Spring Boot framework, which can build a dataset from the collected historical gas concentration data; and the device management module is associated with the spline quantile model SQM to work in conjunction with the device management module.
[0020] S2, the server constructs a corresponding dataset based on the collected historical gas concentration data. To ensure data integrity and validity, the historical gas concentration data must be at least 40 days long. The dataset is then processed using the Spline Quantile Model (SQM) to fit the data distribution and obtain the median regression function. The Spline Quantile Model (SQM) is a statistical model that combines spline functions with the concept of quantile regression. Its core lies in flexibly modeling the nonlinear relationship in quantile regression using spline functions, while utilizing the characteristics of quantile regression to analyze the distribution characteristics of the response variable under different conditional quantiles.
[0021] The median regression function is a nonparametric regression method designed to estimate the linear relationship between the independent variable and the conditional median of the dependent variable. The median regression function represents the distance variable, indicating the distance x meters from the working face within the goaf area. The median regression function can provide a more reliable judgment when there are outliers or non-normal distributions in the data, and can comprehensively describe the influence of independent variables on different quantiles of the dependent variable, capturing the nonlinear characteristics of the data.
[0022] S3, when the distance variable is x, the real-time gas concentration value collected by monitoring is C, calculate the regression function of C with the median. The ratio P; S4, the server has a built-in gas ratio range setting module, which is written in Java. The gas ratio range setting module has the function of dividing the trend range of the corresponding gas, and can also divide the warning level according to the trend range of the corresponding gas. S5, the gas ratio range setting module determines the trend range of the corresponding gas through P obtained in S3, and then obtains the early warning level of the risk of spontaneous combustion of coal in the goaf.
[0023] Reference Figure 4 Taking carbon monoxide (CO) in the goaf of the No. 1 working face of Mengcun Coal Mine in Binchang County, Shaanxi Province as an example, carbon monoxide in the goaf is first collected by a gas sampling device. The communication module organizes the carbon monoxide concentration data and uploads it to the equipment management module on the server. The equipment management module constructs a dataset from the collected historical carbon monoxide concentration data and processes the dataset through the embedded spline quantile module to obtain the median regression function. .
[0024] At a distance of x meters from the working face within the goaf, the real-time concentration of carbon monoxide is monitored as C. The server's gas ratio range setting module divides the carbon monoxide trend range and, based on the carbon monoxide (CO) trend range, classifies the warning level, setting the following three stages: (1) 0≤ <1.0: At this point, the carbon monoxide concentration is below the regression standard value, which indicates a safe state; (2) 1.0≤ <1.5: At this point, the carbon monoxide concentration shows a slight upward trend, which is the initial warning state; (3) 1.5≤ At this point, the carbon monoxide concentration shows a significant upward trend, indicating that this is the alarm stage.
[0025] Depend on Figure 4 As can be seen, the blue dots represent historical data on carbon monoxide gas concentration. When x=30, the measured real-time carbon monoxide gas concentration value is C=0.0195%, and the median regression function value is... =0.015%; Substituting the values, we get a ratio P of 1.3, which falls within the range of 1.0 ≤ P < 1.5. Therefore, we can conclude that the carbon monoxide concentration here shows a slight upward trend, which is the initial alarm stage of spontaneous combustion of coal in the goaf.
[0026] Taking oxygen (O2) as an example, at a distance of x meters from the working face in the goaf, the real-time concentration value of oxygen is C. The server's gas ratio range setting module divides the oxygen trend range and, based on the oxygen (O2) trend range, classifies the warning level, setting the following two stages: (1) 0≤ ≤1.1: At this point, the oxygen concentration is lower than the regression standard value, which is considered a safe state; (2)1.1< At this point, the oxygen concentration shows a clear upward trend, which is the alarm stage. Depend on Figure 5 As can be seen, the blue dots represent historical data on oxygen gas concentration. When x=30, the measured real-time oxygen concentration value C=16.02%, and the median regression function value is... =15.70%; will Substituting the values, we get a ratio P of 1.02, which ≤ 1.1. Therefore, the oxygen concentration here is within a safe range.
[0027] The implementation principle of the on-site identification method for spontaneous combustion risk of coal in the goaf of a working face according to an embodiment of this application is as follows: gas in the goaf is collected by a gas sampling device, and the concentration data of the corresponding gas is uploaded to a server. A corresponding dataset is constructed based on the collected historical gas concentration data. The dataset is processed using the spline quantile model (SQM), and the distribution result of the data is fitted to obtain the median regression function. The trend range of the gas is determined by the ratio of the real-time gas concentration value to the median regression function, thus obtaining the early warning level of the risk of spontaneous combustion of coal in the goaf. This allows for a more accurate and precise assessment of the degree to which the current gas concentration deviates from the normal state, improving the accuracy of predicting the gas change trend in the goaf and ensuring the accurate identification of the risk of spontaneous combustion of coal in the goaf.
[0028] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.
Claims
1. A method for on-site identification of coal spontaneous combustion risk in goaf areas of working faces, characterized in that, include: S1, collects gas from the goaf area using gas sampling equipment and uploads the corresponding gas concentration data to the server; S2, the server constructs a corresponding dataset based on the collected historical gas concentration data, processes the dataset using the spline quantile model (SQM), fits the data distribution results, and obtains the median regression function. ; This represents the regression function; x is the distance variable, indicating the distance x meters from the working face within the goaf area. S3, when the distance variable is x, the real-time gas concentration value collected by monitoring is C. Calculate the regression function of C with the median. The ratio P; S4, the server has a built-in gas ratio range setting module. The gas ratio range setting module has the function of dividing the corresponding gas trend range, and can also divide the warning level of coal spontaneous combustion risk in the goaf according to the trend range of the corresponding gas. S5, the gas ratio range setting module determines the trend range of the corresponding gas through the P value obtained in S3, and then obtains the early warning level of the risk of spontaneous combustion of coal in the goaf.
2. The method for on-site identification of coal spontaneous combustion risk in the goaf of a working face according to claim 1, characterized in that: In step S1, the gas sampling device has a built-in communication module for processing gas concentration data and uploading it to a server; the server has a built-in device management module for recording and storing gas concentrations; the communication module is electrically connected to the device management module.
3. The method for on-site identification of coal spontaneous combustion risk in the goaf of a working face according to claim 2, characterized in that: The device management module is designed and written based on the Spring Boot framework to build a dataset from the collected historical gas concentration data; and the device management module is associated with the spline quantile model SQM to work in conjunction with the device management module.
4. The method for on-site identification of coal spontaneous combustion risk in the goaf of a working face according to claim 1, characterized in that: In step S2, when the server constructs the corresponding dataset based on the collected historical gas concentration data, the historical data must be at least 40 days of gas concentration data.