Integrated miner's suit gas monitoring and early warning device

Abstract

The application discloses an integrated miner's suit gas monitoring and early warning device and relates to the technical field of coal mine safety monitoring.The technical scheme is as follows: when the miner's suit enters a working face, a sensor built in the miner's suit collects gas concentration data; the gas concentration data are subjected to preliminary abnormality judgment, and the gas concentration data are transmitted to an underground gateway; the underground gateway uploads and stores the gas concentration data to a cloud platform; the gas concentration data are input into a gas concentration chaos prediction model, and a gas concentration prediction value is output; if the gas concentration prediction value exceeds a threshold value, early warning information is sent to a gateway of a corresponding area; the gateway sends the early warning information to a target miner's suit, and sound, light and vibration alarms are triggered.The application can predict a gas concentration change trend several minutes in advance.Once a dangerous sign is found, early warning information is directly sent to a corresponding miner's suit, sound, light and vibration alarms are immediately given, and miners can independently take risk-avoiding measures at the first time on site.

Description

Technical Field

[0001] This invention relates to the field of coal mine safety monitoring technology, and more specifically, to an integrated gas monitoring and early warning device for miners' clothing. Background Technology

[0002] Currently, underground gas detection in coal mines mainly relies on two types of hardware structures: fixed gas sensing equipment and handheld portable detectors.

[0003] 1. Fixed gas monitoring device Existing fixed gas monitoring points typically consist of the following components: a gas concentration sensor (catalytic combustion type, electrochemical type, etc.), a mounting bracket or hanging structure fixed to the roadway wall or roof, a power supply unit, which often uses intrinsically safe or explosion-proof power supplies for mining, and a communication module that uploads data to the monitoring system via a wired network or LoRa / industrial bus.

[0004] This type of device relies on sensors to continuously collect gas concentration data in localized areas of the roadway. The data is transmitted to a ground control center, where monitoring software provides visualization and alarm management. It is primarily used at planned fixed monitoring points in main haulage roadways, return air roadways, fixed locations in coal mining faces, and extraction pump stations. The existing operating method involves fixed sensors continuously collecting data → data aggregation at the monitoring center → an alarm being triggered on the ground when limits are exceeded. While its structure is simple and reliable, the monitoring range remains fixed depending on the installation location and cannot be moved with changes in the work area.

[0005] 2. Handheld gas detector Another commonly used device is the portable individual gas detector, whose basic structure includes: a small gas sensor unit, a handheld casing, buttons and a display screen, a battery power module, and an audible and visual alarm. Miners can manually measure the gas concentration in a localized area while working underground. It is mainly used for: gas detection before the work area, safety confirmation of equipment maintenance locations, and temporary work areas. The current working method is characterized by: miners arriving at a certain location and performing a short-term detection → manually reading the concentration → reporting or evacuating upon discovering an anomaly. This device is highly flexible, but relies on manual operation and cannot achieve real-time, continuous, and comprehensive monitoring.

[0006] 3. Typical application scenarios of existing technologies: long-term gas concentration monitoring at fixed monitoring points in roadways, temporary measurement during the expansion of mining faces, safety inspections at airflow intersections and corner areas, safety confirmation before gas extraction, tunneling, and blasting operations, and manual inspection tasks for underground inspection personnel.

[0007] Despite the maturity of existing devices, the following problems still exist: 1. Unavoidable monitoring blind spots: Fixed sensors have difficulty covering new areas after the work face advances, roadway corners, behind equipment, temporary mining points, and other locations.

[0008] 2. Long alarm transmission chain: Alarms from fixed monitoring points need to be transmitted to the ground dispatch center first, and then personnel need to notify the miners, which can result in a delay of tens of seconds or even longer.

[0009] 3. Inability to achieve individualized monitoring and early warning: Existing equipment mainly relies on "regional alarms" and cannot provide accurate alerts for "local risks at the location of a specific miner".

[0010] 4. Inability to continuously monitor personnel movement: Handheld instruments rely on manual detection and cannot achieve dynamic and continuous monitoring; data real-time performance cannot be guaranteed when miners are moving.

[0011] In view of this, the present invention proposes an integrated gas monitoring and early warning device for miners' clothing. Summary of the Invention

[0012] The above-mentioned technical objective of the present invention is achieved through the following technical solution: The first aspect of this invention provides an integrated method for monitoring and early warning of methane gas in miners' clothing, comprising the following steps: When the miner's suit enters the working face, the sensors built into the suit collect data on methane concentration. A preliminary anomaly assessment is performed on the gas concentration data, and the gas concentration data is transmitted to the downhole gateway. The downhole gateway then uploads and stores the data to the cloud platform, inputs the gas concentration chaotic prediction model, and outputs the gas concentration prediction value. If the predicted gas concentration exceeds the threshold, an early warning message will be sent to the gateway of the corresponding area. The gateway sends the warning information to the target miner's suit, triggering an audible, visual, and vibration alarm.

[0013] In conjunction with the first aspect, the present invention is further configured such that: the miner's suit has a built-in gas sensor, a main control module, a built-in Wi-Fi and Bluetooth communication module, and an audible and visual vibration alarm.

[0014] In conjunction with the first aspect, the present invention is further configured such that: the preliminary anomaly judgment is a comparison between the gas concentration data and the gas concentration safety threshold; If the gas concentration exceeds the safe threshold, an audible, visual, and vibration alarm will be triggered, and the gas concentration data will be transmitted to the underground gateway. If the gas concentration does not exceed the safe threshold, the gas concentration data will be transmitted to the underground gateway.

[0015] In conjunction with the first aspect, the present invention is further configured such that: the gas concentration prediction is a gas concentration prediction value for the next 3-5 minutes.

[0016] A second aspect of the present invention also provides an integrated gas monitoring and early warning device / equipment / system for miners' clothing, comprising a memory, a processor, and a computer program stored in the memory, characterized in that the processor executes the computer program to implement the steps of the above method.

[0017] A third aspect of the present invention also provides a computer-readable storage medium having a computer program / instructions stored thereon, characterized in that the computer program / instructions, when executed by a processor, implement the steps of the above-described method.

[0018] A fourth aspect of the present invention also provides a computer program product, including a computer program / instructions, characterized in that, when the computer program / instructions are executed by a processor, they implement the steps of the above-described method.

[0019] In summary, the present invention has the following beneficial effects: Monitoring coverage: Achieve 100% dynamic coverage of the downhole operation area, completely eliminating monitoring blind spots; Early warning timeliness: The early warning response time has been shortened from the traditional 2-3 minutes to less than 20 seconds, and the early warning lead time has reached 3-5 minutes; Prediction accuracy: The chaotic prediction model achieved an accuracy of 85% in predicting sudden changes in gas concentration; System reliability: Employing a three-tiered redundancy mechanism across cloud, edge, and endpoint, the system availability reaches 99.9%. Deployment and maintenance costs: By retrofitting existing miner suits, the cost of single-point retrofitting is reduced by 40% compared to fixed sensor deployment, and the maintenance cycle is extended from 3 months to 1 year. Attached Figure Description

[0020] Figure 1 This is a diagram illustrating the architecture of the chaotic prediction model for gas concentration in Embodiments 1 and 2 of this invention. Figure 2 This is a graph showing the combined analysis results of the accuracy and false alarm rate of the chaotic time series analysis and prediction model in Embodiments 1 and 2 of the present invention; Figure 3 This is a comparative analysis result of the gas concentration prediction models in Embodiments 1 and 2 of the present invention. Detailed Implementation

[0021] 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.

[0022] Example 1: An integrated method for monitoring and early warning of methane gas in miners' clothing includes the following steps: S100. Collect methane concentration data; S200. Perform preliminary anomaly judgment on gas concentration data, transmit gas concentration data to the underground gateway, upload and store it to the cloud platform through the underground gateway, input the gas concentration chaotic prediction model, and output the gas concentration prediction value; If the predicted gas concentration exceeds the threshold, an early warning message will be sent to the gateway of the corresponding area. The S300 gateway sends the warning information to the target miner's suit via the LoRa network, triggering an audible, visual, and vibration alarm.

[0023] In step S100 of this embodiment, the miner's suit has a built-in gas sensor, a main control module, a built-in Wi-Fi and Bluetooth communication module, and an audible and visual vibration alarm.

[0024] In this embodiment, miners wear smart mining suits to enter the working face. Each suit is equipped with a high-precision gas sensor (detection range 0-5% VOL, resolution 0.01%), an ESP-32 main control module, built-in Wi-Fi and Bluetooth communication modules, and an audible, visual, and vibration alarm. The outer shell is made of flame-retardant ABS engineering plastic, and the sensor has an IP67 protection rating. Gas concentration data is collected through the gas sensor inside the mining suit. In step S200 of this embodiment, the preliminary anomaly judgment is to compare the gas concentration data with the gas concentration safety threshold. If the gas concentration exceeds the safe threshold, an audible, visual, and vibration alarm will be triggered, and the gas concentration data will be transmitted to the underground gateway. If the gas concentration does not exceed the safe threshold, the gas concentration data will be transmitted to the underground gateway.

[0025] In step S200 of this embodiment, the architecture diagram of the gas concentration chaotic prediction model is as follows: Figure 1 As shown. The predicted gas concentration value is the predicted gas concentration value for the next 3-5 minutes.

[0026] Example 2: Chaotic Prediction Model for Gas Concentration Due to the nonlinear and nonstationary nature of gas concentration data, this invention employs chaotic dynamics theory as the core of the prediction model, replacing traditional linear models (such as ARIMA). Based on Takens' embedding theorem, chaotic time series analysis uses phase space reconstruction technology to analyze and predict this nonlinear time series of gas concentration.

[0027] Phase space reconstruction: For a d-dimensional deterministic dynamical system, if a scalar time series is observed... The phase space, which is topologically equivalent to the original system, can be reconstructed using the delayed coordinate vector: in, To delay time, Let be the embedding dimension. When When ≥ 2d+1, the reconstructed phase space Y(t) is equivalent to the dynamics of the original system in the sense of embedding.

[0028] The main steps include: A100: Data collection and preprocessing Storing methane concentration data on a cloud service platform allows for automatic identification of the data and handles missing values ​​through direct deletion, making it suitable for industrial scenarios requiring high-quality data. This design ensures the code's versatility and maintainability. The dataset is then cleaned, and a methane concentration time series is generated from the cleaned data.

[0029] A200: By analyzing the gas concentration time series, chaotic parameters are identified to obtain the delay time. and embedding dimension ; Delay time The delay time is determined using the mutual information function method. The classic method is based on information theory concepts to measure the degree of correlation between two random variables.

[0030] In its implementation, the algorithm constructs different delays We explore this relationship using sequence pairs. For each candidate... Value, the original sequence Its delayed sequence Data pairs are formed, and their joint probability distribution is estimated using a two-dimensional histogram statistical method. This distribution reflects the likelihood that the two variables will simultaneously take specific values.

[0031] The core of mutual information calculation lies in comparing the product of the joint probability distribution and the individual marginal probability distributions. If two variables are completely independent, then the joint probability should equal the product of the marginal probabilities, and the mutual information is zero. If a dependency exists, the joint probability will deviate from this product, and the mutual information value will increase.

[0032] In actual gas concentration data, when When the value is very small, adjacent concentration readings are highly correlated, with a large mutual information value. However, this strong correlation causes the reconstructed phase space points to be overly concentrated near the diagonal, failing to fully represent the system's dynamic characteristics. With... As the value increases, the correlation gradually weakens, and the mutual information value decreases. When the number of sequences continues to increase to a certain extent, they become almost completely independent, and the mutual information value will stabilize at a low level. However, at this point, too much evolutionary information of the system has been lost.

[0033] Therefore, this invention seeks the first local minimum point on the mutual information function curve. This point represents the optimal balance between correlation and independence: it ensures sufficient independence between coordinate components, avoiding information redundancy, while maintaining appropriate correlation to ensure no loss of the system's dynamic evolution information. In the code implementation, this invention achieves this by traversing... Calculate the possible values ​​for each. The corresponding mutual information values ​​are then scanned to identify the first point that satisfies the local minimum condition.

[0034] The specific implementation is as follows: Mutual information function Used for measurement and Statistical dependencies between them: In the formula, For marginal probability distribution, The joint probability distribution is used. The optimal delay time is... Pick The first local minimum value, at which point the sequence components maintain both correlation and independence.

[0035] Embedding dimension Determination of: False Nearest Neighbor Method The algorithm's execution process involves a step-by-step exploration from low to high dimensions. For each candidate embedding dimension... This invention first reconstructs the corresponding phase space, transforming the one-dimensional time series into... A set of points in 3D space. In 3D space, this invention finds the nearest neighbor for each point and records the distance R_m between them.

[0036] A key step is to examine how these points are embedded in the present invention. In a 3D space, do these nearest neighbor relationships still remain? Specifically, this invention calculates the distance R_( between the original nearest neighbor pairs after adding a dimension.) +1). If the distance ratio R_( ) / R_ Exceeding the set threshold (usually 10-20) means that two points that appear close in low-dimensional space are actually far apart in high-dimensional space, and their proximity is an illusion caused by projection.

[0037] In practical applications of gas concentration prediction, this invention observes that as the embedding dimension increases... As the dimension increases, the proportion of false nearest neighbors typically shows a significant decreasing trend. In the initial stage, due to a severe lack of dimensionality, the proportion of false nearest neighbors is very high. With... As it increases, the proportion decreases rapidly. When When a certain value is reached, the proportion of pseudo-nearest neighbors will drop to a very low level (such as below 10%). At this point, further increasing the dimensionality will only bring minor improvements, but will significantly increase computational complexity and sensitivity to noise.

[0038] Therefore, this invention selects the dimension in which the proportion of pseudo-nearest neighbors first drops below the threshold as the optimal embedding dimension. This avoids unnecessary redundancy and makes the selection of the optimal embedding dimension more accurate.

[0039] Specifically as follows: For a given The criteria for identifying false nearest neighbors are: in express The first phase space A vector, Its nearest neighbor, The tolerance threshold (usually taken as...) ).when Increase the proportion of pseudo-nearest neighbors First time below the threshold (e.g., 10%): corresponding That is, the optimal embedding dimension. .

[0040] A300: Use to obtain delay time and embedding dimension The gas concentration time series is converted into a trajectory in a high-dimensional phase space to obtain real-time gas concentration data. This data is then transformed into a current state vector, and a neighbor search is performed. Weights are assigned based on distance to obtain a weighted reference point set. The linear regression coefficients are solved by weighted least squares method to construct a mapping relationship from the current phase space state to the concentration at the next time step. The current state vector is substituted into the fitted mapping relationship to output the predicted gas concentration value.

[0041] Specifically, a weighted first-order local prediction algorithm is used to achieve time-series prediction and inference of gas concentration. This algorithm is based on phase space reconstruction technology in chaos theory, which maps a one-dimensional gas concentration sequence to a high-dimensional phase space to achieve short-term prediction. In the algorithm implementation, this invention first converts historical gas concentration data into trajectories in a high-dimensional phase space through phase space reconstruction; For each current state point to be predicted, this invention searches for its nearest set of reference points in the historical phase space. The selection of reference points is based on the Euclidean distance metric; points that are closer together are considered to have more similar evolutionary patterns.

[0042] After obtaining a weighted reference point set, this invention establishes a local linear model to describe the evolution in phase space. The linear regression coefficients are solved using the weighted least squares method to construct a mapping relationship from the current phase space state to the concentration at the next time step. The final predicted value is obtained by substituting the current state point into the obtained linear model.

[0043] The specific steps are as follows: (a) Neighborhood search For the current state point Find its k nearest neighbors in historical data. satisfy: ; (b) Distance-weighted Assign an exponentially decaying weight to each neighboring point: ; (c) Local linear modeling Assume that within the neighborhood of Y(t), the system evolution is approximately linear: ; The parameters are estimated using the weighted least squares method: ; Its analytical solution is: ; ; in, ; (d) Predicted output Substituting the current state into the fitted model, we obtain the predicted value for the next time step: ; Final predicted gas concentration That is The first component.

[0044] The sensor continuously uploads gas data, and the model continuously adjusts and optimizes based on the new data. Specifically, the model dynamically calculates chaotic parameters based on the new data. sum dimension To achieve model training and iteration.

[0045] The loss function used is the weighted sum of squared residuals, and its mathematical expression is: ; This loss function uses different parameters for each prediction point and places greater emphasis on historical data. It is suitable for predicting non-stationary and nonlinear gas concentration sequences and can quickly adapt to dynamic changes in the system.

[0046] Combination Figure 2 and Figure 3 The comprehensive performance evaluation results of the chaotic prediction model for gas concentration are as follows: 1. Chaotic time series prediction methods have advantages in processing nonlinear and non-stationary gas concentration data; 2. The weighted first-order local prediction algorithm based on the chaotic prediction algorithm can effectively capture the chaotic characteristics of the data. The chaotic prediction model has a prediction accuracy of 85% for sudden changes in gas concentration.

[0047] Example 3: The present invention also provides an integrated gas monitoring and early warning device / equipment / system for miners, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the above method.

[0048] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. 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. An integrated method for monitoring and early warning of methane gas in miners' clothing, characterized by: Includes the following steps: Collect gas concentration data; A preliminary anomaly assessment is performed on the gas concentration data, and the gas concentration data is transmitted to the downhole gateway. The downhole gateway then uploads and stores the data to the cloud platform, inputs the gas concentration chaotic prediction model, and outputs the gas concentration prediction value. If the predicted gas concentration exceeds the threshold, an early warning message will be sent to the gateway of the corresponding area. The gateway sends the warning information to the target miner's suit, triggering an audible, visual, and vibration alarm.

2. The early warning method according to claim 1, characterized in that: The gas concentration data is collected by a gas sensor built into the miner's suit.

3. The early warning method according to claim 1, characterized in that: The initial anomaly assessment involves comparing the gas concentration data with the safe gas concentration threshold. If the gas concentration exceeds the safe threshold, an audible, visual, and vibration alarm will be triggered, and the gas concentration data will be transmitted to the underground gateway. If the gas concentration does not exceed the safe threshold, the gas concentration data will be transmitted to the underground gateway.

4. The early warning method according to claim 1, characterized in that: The gas concentration prediction is the predicted gas concentration value for the next 3-5 minutes.

5. The early warning method according to claim 1, characterized in that: The chaotic prediction model for gas concentration includes: preprocessing gas concentration data to obtain a gas concentration time series; Chaotic parameters were identified and the delay time was obtained by using the gas concentration time series. and embedding dimension ; Use delay time and embedding dimension The gas concentration time series is converted into a trajectory in a high-dimensional phase space to obtain real-time gas concentration data and convert it into a current state vector. Neighbor point search is performed, and weights are assigned based on distance to obtain a weighted reference point set. The linear regression coefficients are solved by weighted least squares method to construct a mapping relationship from the current phase space state to the concentration at the next time moment. The current state vector is substituted into the fitted mapping relationship to output the gas concentration prediction value.

6. The early warning method according to claim 5, characterized in that: The delay time The embedding dimension is determined using the mutual information function method. Determined by the pseudo-nearest neighbor method.

7. The early warning method according to claim 6, characterized in that: The nearest neighbor search involves calculating the Euclidean distance between the current state vector and the trajectory in the high-dimensional phase space.

8. A device / equipment / system for integrated gas monitoring and early warning in miners' clothing, comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method according to any one of claims 1-7.

9. A computer-readable storage medium having a computer program / instructions stored thereon, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method described in any one of claims 1-7.

10. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method described in any one of claims 1-7.

Images