Real-time overrun early warning method based on gas concentration characteristic change

Abstract

The invention relates to a real-time overrun early warning method based on gas concentration characteristic change, and belongs to the technical field of coal mine safety monitoring and early warning. The method comprises the following steps: collecting gas concentration time sequence data in real time through gas sensor networks deployed on a working face and a tunneling face; grading concentration intervals, wherein the concentration intervals comprise a threshold interval, an alarm color and an alarm score; the method comprises the following steps of: preprocessing collected gas concentration time sequence data, and then respectively calculating three alarm scores by adopting an amplification change bandwidth judgment method, an exponential smoothing difference method and a sliding window Z score method to correspond to alarm scores in concentration interval grading; voting fusion: at the same sampling moment, synchronously aligning the calculated three alarm scores, and then determining final early warning information according to a voting fusion rule; and outputting early warning information in a visualized manner. According to the method, the concentration interval dynamic grading and voting fusion strategy is combined, refined recognition and grading early warning of the gas risk situation are achieved, and the timeliness and accuracy of alarm are remarkably improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of coal mine safety monitoring and early warning, and relates to a real-time over-limit early warning method based on characteristic changes of gas concentration. Background Art

[0002] Gas (primarily methane) in underground coal mines is one of the most hazardous gases to coal mine safety due to its flammability, explosiveness, and asphyxiation. To ensure the safety of personnel and equipment, real-time and accurate monitoring and early warning of underground gas concentrations are crucial.

[0003] Currently, the gas monitoring systems widely used in coal mines rely primarily on fixed sensors deployed in roadways or working faces. These sensors continuously sample and measure gas concentrations. When the monitored concentration exceeds a pre-set static safety threshold, the system triggers an alarm.

[0004] Although existing monitoring systems achieve real-time data collection, they still have the following three significant limitations in practical applications:

[0005] 1) Limitations of Static Threshold Alarm Mechanisms: Traditional alarm mechanisms rely on fixed concentration thresholds. This approach struggles to effectively distinguish between different types of risks, such as a rapid surge in gas concentration (e.g., a sudden leak) and a slow, sustained increase (e.g., gas accumulation). For the former, the system may miss warnings due to data filtering or sampling delays, missing the optimal warning opportunity. For the latter, false alarms may occur due to environmental factors (e.g., temperature and humidity changes) or sensor drift. Frequent false alarms can reduce staff vigilance.

[0006] 2) Weak anti-interference capabilities of single detection algorithms: Current systems often use simple threshold comparison or first-order difference methods to determine changes in gas concentration. These algorithms are simple models and lack anti-interference capabilities in the complex underground environment. For example, normal airflow fluctuations caused by ventilation equipment or gas pulsation can be misjudged as anomalies, resulting in unstable alarm level determination and reducing the overall reliability of early warning decisions.

[0007] 3) Lack of comprehensive multi-indicator risk assessment capabilities: Current monitoring solutions primarily focus on a single instantaneous concentration value, failing to provide a comprehensive, multi-dimensional assessment of gas risk. These systems overlook key characteristic indicators, such as the steepness of concentration trends, the magnitude of concentration fluctuations, and the duration of concentration excursions. Due to the lack of effective grading and integrated analysis of these indicators, the system is unable to achieve refined and differentiated management of gas risks, making it difficult to provide scientific, multi-dimensional decision support for safe production scheduling based on risk levels.

[0008] In summary, existing coal mine gas monitoring technologies have significant shortcomings in terms of early warning accuracy, reliability, and comprehensive risk assessment. Therefore, there is an urgent need to develop new gas monitoring technologies that can overcome these shortcomings, particularly for critical areas such as underground working faces and tunneling surfaces, to significantly improve early warning accuracy and scientific decision-making. Summary of the Invention

[0009] In view of this, the purpose of the present invention is to provide a real-time over-limit warning method based on changes in gas concentration characteristics, to solve the defects of the existing technology that a single threshold or simple differential algorithm is difficult to take into account both sudden leakage (rapid rise in concentration) and slow accumulation risk, and is susceptible to noise interference such as gas pulsation and ventilation fluctuations, leading to false alarms or missed reports, as well as the lack of a systematic classification and fusion decision-making mechanism for multi-dimensional features such as concentration change trends, fluctuation amplitudes, and over-limit durations. This method combines dynamic classification of concentration intervals with voting fusion strategies to achieve refined identification and graded warning of gas risk situations, significantly improving the timeliness and accuracy of alarms.

[0010] In order to achieve the above object, the present invention provides the following technical solutions:

[0011] Option 1:

[0012] A real-time over-limit warning method based on changes in gas concentration characteristics, comprising:

[0013] (1) Data collection: A gas sensor network deployed at the working face and tunneling face collects real-time gas concentration time series data;

[0014] (2) Concentration interval classification, including threshold interval, alarm color and alarm score;

[0015] (3) Data processing: First, the collected gas concentration time series data are preprocessed, and then the three alarm scores are calculated using the amplification change bandwidth determination method (ACBDI), exponential smoothing difference method (ESDI) and sliding window Z score method (ZDAI), which correspond to the alarm scores in the concentration interval classification;

[0016] (4) Voting fusion: At the same sampling moment, the three calculated alarm scores are aligned synchronously, and then the final warning information is determined according to the voting fusion rules;

[0017] (5) Visual output of warning information.

[0018] Furthermore, for data collection, the gas sensor network includes multiple fixed-mounted gas concentration monitoring devices and portable gas concentration monitoring devices carried by workers. The fixed-mounted gas concentration monitoring devices utilize a portable, retractable support structure, foldable and securely supported by hinges and fixed clips, ensuring stable installation in extremely thin coal seams. The portable monitoring devices are equipped with 5G signal transmission ports for real-time data transmission.

[0019] Furthermore, the concentration interval classification is based on the threshold interval division of the "red, orange, yellow, and blue" alarm levels commonly used in coal mines. The specific threshold intervals and corresponding scores are as follows:

[0020] When the threshold interval is [0%, 0.55%), the alarm color is blue and the alarm score is 1;

[0021] When the threshold interval is [0.55%, 0.7%), the alarm color is yellow and the alarm score is 2;

[0022] When the threshold interval is [0.7%, 1.0%), the alarm color is orange and the alarm score is 3;

[0023] When the threshold interval is 1.0% and above, the alarm color is red and the alarm score is 4.

[0024] Furthermore, during data processing, the Amplitude Change Bandwidth Determination Method (ACBDI) is used to quantify the severity of gas concentration fluctuations and capture rapid concentration increases through bandwidth increase trends. The specific steps are as follows:

[0025] Step 1: Parameter setting:

[0026] Set the sliding window w (default is 30), bandwidth multiplier k (default is 1.5), and continuous growth threshold h1 (default is 5);

[0027] Step 2: Calculate the moving average and standard deviation:

[0028] For each time t, based on the first w sampling points, i.e., the sliding window p t-w+1 ,…,p t Calculate the moving average and standard deviation using the following formula:

[0029]

[0030] Among them, SMA t and STDs t They represent the moving average and standard deviation at time t respectively;

[0031] Step 3: Calculate the upper and lower tracks and bandwidth using the following formula:

[0032] Uppert =SMA t +k*STD t

[0033] Lower t =SMA t -k*STD t

[0034] Band t =Upper t -Lower t

[0035] Among them, Upper t and Lower t Respectively represent the upper and lower rails at time t; Bnad t represents the bandwidth at time t;

[0036] Step 4: Amplification judgment and alarm:

[0037] Compare the moving average, upper track line, and bandwidth of the current moment with those of the previous moment; when the three show an upward trend at the same time, that is, SMA t >SMA t-1 ,Upper t >Upper t-1 ,Band t >Band t-1 If the condition is met continuously within h1 sampling points (time window), the corresponding alarm score score will be triggered according to the concentration interval. a , and output the trigger time point.

[0038] Furthermore, in data processing, the exponential smoothing difference method (ESDI) is used to quantify the trend change and acceleration intensity of gas concentration, and to capture signal reversal or acceleration scenarios through short-term and long-term exponential smoothing differences. The specific steps are as follows:

[0039] Step 1: Parameter setting:

[0040] Set the short-term exponential moving average window s w (Default is 30), long-term exponential moving average window l w (Default is 60); signal line smoothing window sl w (default is 30), continuous growth threshold h2 (default is 2);

[0041] Step 2: Calculate the short-term and long-term exponential moving averages (EMA):

[0042] For each time instant t, use the short-term exponential smoothing factor and the long-term exponential smoothing factor Calculate short-term and long-term exponential moving averages;

[0043]

[0044]

[0045] in, represents the short-term exponential moving average at time t, represents the long-term exponential moving average at time t; P t Indicates the gas concentration value at the current time t.

[0046] Step 3: Find the difference line DIF and the signal line SIG

[0047]

[0048] SIG(t)=(DIF(t-sl w +1)+…+DIF(t)) / sl w

[0049] Histogram(t)=DIF(t)-SIG(t)

[0050] Where DIF(t) represents the difference line at time t, SIG(t) represents the signal line at time t, and Histogram(t) represents the difference value at time t;

[0051] Step 4: Amplification judgment and alarm:

[0052] When any sequence rises continuously at adjacent moments, such as DIF(t)>DIF(t-1), SIG(t)>SIG(t-1) or Histogram(t)>Histogram(t-1); if it is continuously satisfied within the continuous growth of h2 sampling points (time window), the corresponding alarm score score is triggered according to the concentration interval classification. e , and output the trigger time point.

[0053] Furthermore, during data processing, the sliding window Z-score method (ZDAI) is used to quantify abnormal deviations in the statistical distribution of gas concentration. Abnormal concentration peaks are captured by comparing the Z-score within the sliding window with the quantile threshold. The specific steps are as follows:

[0054] Step 1: Parameter setting:

[0055] Set the sliding window w (default is 30), the percentile threshold Q (default is 0.95, corresponding to the 95% percentile), and the continuous exceeding threshold h3 (default is 2);

[0056] Step 2: Calculate the moving average and standard deviation:

[0057] For each time t, based on the first w sampling points, i.e., the sliding window p t-w+1 ,…,p t Calculate the moving average and standard deviation using the following formula:

[0058]

[0059] Among them, SMA t and STDs t They represent the moving average and standard deviation at time t respectively;

[0060] Step 3: Find the Z score at the current moment:

[0061]

[0062] Among them, Z t represents the Z score at time t;

[0063] Step 4: Determine the dynamic threshold:

[0064] In the same sliding window, sort the first w Z scores and take the Qth quantile as the Threshold t ;Threshold t =Quantile({Z t-w+1 ,…,Z t},Q);

[0065] Step 5: Anomaly detection and alarm:

[0066] Determine the current Z t Is it greater than Threshold? t and Z at the previous moment t-1 Compared with the previous period, there is an upward trend, that is, Z t >Z t-1 If the condition is met continuously within h3 sampling points (time window), the corresponding alarm score score will be triggered according to the concentration interval. z , and output the trigger time point.

[0067] Furthermore, the voting fusion rules are:

[0068] If any algorithm is not triggered at time t, its score is considered 0 to avoid missing the voting contribution of the unwarned algorithm;

[0069] Red priority: If any vote is rated as a red warning, a red warning will be output directly and subsequent score comparisons will be terminated;

[0070] All-vote warning: When all three algorithms are triggered (score > 0), and at least one of them is a red warning, the highest score and corresponding level are taken;

[0071] Majority of the same level: When all three algorithms are triggered and there is no red alert, the one with the most votes will be adopted first; if the three scores are different, the highest score will prevail;

[0072] Double vote evaluation: When there are exactly two algorithms triggered, if the two votes have the same score, the higher score will be used; if they are not equal, the higher score will be used;

[0073] Single-vote assessment: When only one vote is triggered, the algorithm score is directly used as the final warning level.

[0074] Furthermore, visual output specifically displays warning information through a graphical interface, including level identification, risk color, and algorithm traceability information, and supports two-dimensional and three-dimensional visualization of data. The system also has a data backup and recovery mechanism to ensure data security and reliability.

[0075] Solution 2: A real-time over-limit warning system based on changes in gas concentration characteristics, including: data acquisition layer, data processing layer, decision analysis layer and visualization output layer.

[0076] The data collection layer comprises a network of gas sensors deployed at the working and tunneling faces of the coal mine. This network consists of multiple fixed-mounted gas concentration monitoring devices and portable gas concentration monitoring devices carried by workers. The fixed-mounted gas concentration monitoring devices utilize a portable, retractable support structure, foldable and securely supported by hinges and fixed clips, ensuring stable installation in extremely thin coal seams. The portable monitoring devices are equipped with 5G signal transmission ports to enable real-time data transmission.

[0077] The data processing layer includes a data preprocessing module and an algorithm calculation module; the data preprocessing module filters the collected gas concentration time series data to eliminate noise interference such as gas pulsation and ventilation fluctuations; the algorithm calculation module executes three real-time detection algorithms in parallel: the amplification change bandwidth determination method (ACBDI), the exponential smoothing difference method (ESDI) and the sliding window Z score method (ZDAI); among them, the amplification change bandwidth determination method (ACBDI) quantifies the severity of gas concentration fluctuations through sliding window and bandwidth multiple calculation; the exponential smoothing difference method (ESDI) uses short-term and long-term exponential moving average calculations to quantify concentration trend changes; the sliding window Z score method (ZDAI) captures abnormal concentration peaks by comparing the Z-score within the sliding window with the quantile threshold.

[0078] The decision analysis layer includes a concentration interval classification module and a voting fusion module; the concentration interval classification module dynamically locates the risk interval according to the gas concentration value of the current sampling point and generates a concentration classification result; the voting fusion module performs fusion analysis based on the calculation results of the algorithm calculation module to generate a dynamic alarm level.

[0079] The visual output layer displays warning information through a graphical interface, including level identification, risk color, and algorithm traceability information. It also supports two-dimensional and three-dimensional visualization of data. The system also has a data backup and recovery mechanism to ensure data security and reliability.

[0080] The beneficial effects of the present invention are:

[0081] (1) By executing three core detection algorithms, namely the amplification change bandwidth determination method (ACBDI), the exponential smoothing difference method (ESDI), and the sliding window Z score method (ZDAI), the dual risk modes of sudden jump and slow accumulation of gas concentration are captured collaboratively, overcoming the shortcomings of a single threshold or simple differential algorithm that is difficult to take into account multiple risks, and significantly improving the accuracy and reliability of early warning.

[0082] (2) The dynamic grading mechanism of concentration intervals, combined with the “red, orange, yellow, and blue” alarm level system, enables refined identification and graded warning of gas risks, enabling the system to more intuitively reflect the risk level of gas concentration changes, effectively reducing the false alarm rate and missed alarm rate.

[0083] (3) Based on the majority voting fusion mechanism, dynamic alarm levels are generated. Through the collaborative analysis of multi-source decision tickets, the system's comprehensive assessment capability of gas risks is improved, the timeliness and accuracy of early warning are enhanced, and a solid line of defense is built for the safety of miners' lives and the stable operation of mines.

[0084] (4) Through the coordinated use of three different algorithms, the system can effectively filter out noise interference such as gas pulsation and ventilation fluctuation, improve the robustness and reliability of the early warning, and effectively reduce the false alarm or missed alarm rate.

[0085] (5) The use of multi-dimensional feature analysis, including concentration trend, fluctuation amplitude, and duration of exceeding the limit, enables a comprehensive understanding of the complex changes in gas concentration and improves the timeliness and accuracy of early warning.

[0086] In summary, the present invention uses a multi-algorithm collaborative monitoring and voting fusion mechanism to accurately capture abnormal gas concentration trends and simultaneously perceive the dual risks of sudden jumps and slow accumulation; the concentration grading system constructs an intuitive risk spectrum and significantly enhances the system's anti-interference robustness; while significantly reducing the false alarm rate and missed alarm rate, it comprehensively improves the timeliness and accuracy of the alarm, deeply meeting the high reliability requirements of underground coal mine safety production, and building a solid line of defense for the safety of miners' lives and the stable operation of mines.

[0087] Other advantages, objects, and features of the present invention will be described in part in the following description and, in part, will be apparent to those skilled in the art upon examination of the following description or may be learned from practice of the present invention. The objects and other advantages of the present invention may be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0088] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below with reference to the accompanying drawings, in which:

[0089] Figure 1 This is a block diagram of a real-time over-limit warning system based on changes in gas concentration characteristics;

[0090] Figure 2 The figure is an overall flow chart of the real-time over-limit warning method based on the characteristic changes of gas concentration. DETAILED DESCRIPTION

[0091] The following describes the embodiments of the present invention by means of specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0092] Among them, the accompanying drawings are only for illustrative purposes and represent only schematic diagrams rather than actual pictures, and should not be understood as limiting the present invention. In order to better illustrate the embodiments of the present invention, some parts of the accompanying drawings may be omitted, enlarged or reduced, and do not represent the dimensions of actual products. For those skilled in the art, it is understandable that some well-known structures and their descriptions may be omitted in the accompanying drawings.

[0093] The same or similar numbers in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if there are terms such as "upper", "lower", "left", "right", "front", "back", etc. indicating directions or positional relationships, they are based on the directions or positional relationships shown in the drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operate in a specific direction. Therefore, the terms describing the positional relationship in the drawings are only used for illustrative purposes and cannot be understood as limiting the present invention. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0094] Example 1:

[0095] See also Figure 1 This embodiment provides a real-time over-limit warning system based on changes in gas concentration characteristics, including: a data acquisition layer, a data processing layer, a decision analysis layer, and a visualization output layer.

[0096] The data collection layer comprises a network of gas sensors deployed at the working and tunneling faces of the coal mine. This network consists of multiple fixed-mounted gas concentration monitoring devices and portable gas concentration monitoring devices carried by workers. The fixed-mounted gas concentration monitoring devices utilize a portable, retractable support structure. Folding hinges and fixed clips allow for folding and secure support, ensuring stable installation in extremely thin coal seams. The portable monitoring devices are equipped with 5G signal transmission ports for real-time data transmission.

[0097] The data processing layer includes a data preprocessing module and an algorithm calculation module; the data preprocessing module filters the collected gas concentration time series data to eliminate noise interference such as gas pulsation and ventilation fluctuations; the algorithm calculation module executes three real-time detection algorithms in parallel: the amplification change bandwidth determination method (ACBDI), the exponential smoothing difference method (ESDI) and the sliding window Z score method (ZDAI); among them, the amplification change bandwidth determination method (ACBDI) quantifies the severity of gas concentration fluctuations through sliding window and bandwidth multiple calculation; the exponential smoothing difference method (ESDI) uses short-term and long-term exponential moving average calculations to quantify concentration trend changes; the sliding window Z score method (ZDAI) captures abnormal concentration peaks by comparing the Z-score within the sliding window with the quantile threshold.

[0098] The decision analysis layer includes a concentration interval classification module and a voting fusion module; the concentration interval classification module dynamically locates the risk interval according to the gas concentration value of the current sampling point and generates a concentration classification result; the voting fusion module performs fusion analysis based on the calculation results of the algorithm calculation module to generate a dynamic alarm level.

[0099] The visual output layer displays warning information through a graphical interface, including level identification, risk color, and algorithm traceability information. It also supports two-dimensional and three-dimensional visualization of data. The system also has a data backup and recovery mechanism to ensure data security and reliability.

[0100] Example 2:

[0101] See also Figure 2 This embodiment provides a real-time over-limit warning method based on changes in gas concentration characteristics. The specific implementation steps are as follows:

[0102] Step 1: Initialize system configuration, including network connection settings, algorithm parameter initialization, database establishment, etc.

[0103] Step 2: The data acquisition layer performs data acquisition tasks, including real-time monitoring of parameters such as gas concentration and ground stress;

[0104] Step 3: The data preprocessing module filters the collected data to remove noise interference and form a valid data set;

[0105] Step 4: The algorithm calculation module executes the Amplitude Change Bandwidth Determination Method (ACBDI), Exponential Smoothing Difference Method (ESDI), and Sliding Window Z-Score Method (ZDAI) in parallel to calculate the fluctuation severity, trend change, and abnormal deviation index of gas concentration respectively;

[0106] Step 5: The concentration interval classification module dynamically generates gas risk levels based on the real-time concentration value and the preset threshold interval;

[0107] Step 6: The voting fusion module generates the final warning level based on the output results of multiple algorithms and the majority voting mechanism;

[0108] Step 8: The visual output layer displays warning information, including level identification, risk color, algorithm traceability information, etc., and supports data visualization;

[0109] Step 9: The system automatically backs up data to ensure data security;

[0110] Step 10: Implement corresponding emergency response measures based on the warning level.

[0111] Example 3:

[0112] The specific implementation of each module of the system in Example 1 is as follows:

[0113] 1) The data collection layer includes:

[0114] (1) A fixed-mounted gas concentration monitoring device consisting of a high-precision capacitive sensor, a data acquisition unit, and a 5G communication module. The sensor uses MEMS technology, with a measurement range of 0-5% and a resolution of 0.001%. The data acquisition unit uses an ARM Cortex-A53 architecture microcontroller with a sampling frequency of 1Hz. The 5G communication module supports full coverage of the coal mine area and a transmission rate of up to 100Mbps.

[0115] (2) A portable gas concentration monitoring device consisting of a high-precision NDIR optical sensor and an intelligent terminal device. The sensor has a measurement range of 0-5% and a resolution of 0.001%. The intelligent terminal device is IP68-rated for water and dust resistance and is equipped with a 2000mAh lithium battery with a battery life of up to 12 hours.

[0116] 2) The data processing layer includes:

[0117] (1) Data preprocessing module uses a filtering algorithm based on median and standard deviation to eliminate outliers and noise interference and retain valid data.

[0118] (2) Algorithm calculation module, which implements:

[0119] ACBDI algorithm: uses a sliding window w of 30, a bandwidth multiplier k of 1.5, and a continuous growth threshold h1 of 5.

[0120] ESDI algorithm: short-term exponential moving average window w is 30, long-term exponential moving average window l w is 60, signal line smoothing window sl w is 30, and the continuous growth threshold h2 is 2.

[0121] ZDAI algorithm: The sliding window size w is 30, the quantile threshold Q is 0.95, and the continuous exceeding threshold h3 is 2.

[0122] 3) The decision analysis layer includes:

[0123] (1) Concentration interval classification module

[0124] During real-time monitoring, the system dynamically locates risk zones based on gas concentrations at the current sampling point and instantly generates concentration grading results. This grading mechanism helps production safety managers clearly identify gas risk levels, supporting graded warning triggering, step-by-step management and control, and differentiated emergency response. Threshold ranges are divided according to the common "red, orange, yellow, and blue" alarm levels in coal mines. The specific threshold ranges and corresponding scores are shown in Table 1 below.

[0125] Table 1 Concentration interval classification mechanism

[0126] Threshold interval Alarm color Score [0％，0.55％) blue 1 [0.55％，0.7％) yellow 2 [0.7％，1.0％) orange color 3 1.0% and above red 4

[0127] (2) Voting Fusion Module

[0128] At the same sampling moment, the three scores (score a 、score e 、score z ) synchronization; if any algorithm is not triggered at that moment, its score is considered to be 0 to avoid missing the voting contribution of the unwarned algorithm.

[0129] Voting Fusion Rules:

[0130] Red priority: If any vote is rated as a red warning, a red warning will be output directly and subsequent score comparisons will be terminated;

[0131] All-vote warning: When all three algorithms are triggered (score > 0), and at least one of them is a red warning, the highest score and corresponding level are taken;

[0132] Majority of the same level: When all three algorithms are triggered and there is no red alert, the one with the most votes will be adopted first; if the three scores are different, the highest score will prevail;

[0133] Double-vote evaluation: When there are exactly two algorithms triggered, if the two votes have the same score, the higher score will be used; if they are not equal, the higher score will be used;

[0134] Single ticket assessment: When only one ticket is triggered, the algorithm score is directly used as the final alarm level.

[0135] 4) Visualization output layer

[0136] A graphical interface developed using HTML5 technology displays warning information, supporting real-time data display in 2D bar charts and 3D scatter plots. The interface supports multi-level menu navigation and real-time refresh, making it easy for operators to quickly understand the cause of the warning and take appropriate measures.

[0137] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solutions, which should all be included in the scope of the claims of the present invention.

Claims

1. A real-time over-limit warning method based on changes in gas concentration characteristics, characterized in that: The method includes: (1) Data collection: A gas sensor network deployed at the working face and tunneling face collects real-time gas concentration time series data; (2) Concentration interval classification, including threshold interval, alarm color and alarm score; (3) Data processing: First, the collected gas concentration time series data is preprocessed, and then the three alarm scores are calculated using the increase change bandwidth determination method, exponential smoothing difference method and sliding window Z score method, which correspond to the alarm scores in the concentration interval classification; (4) Voting fusion: At the same sampling moment, the three calculated alarm scores are aligned synchronously, and then the final warning information is determined according to the voting fusion rules; (5) Visual output of warning information.

2. The real-time over-limit warning method based on gas concentration characteristic changes according to claim 1 is characterized in that: In data collection, the gas sensor network is a plurality of fixedly installed gas concentration monitoring devices underground.

3. The real-time over-limit warning method based on gas concentration characteristic changes according to claim 1 is characterized in that: The concentration interval classification is based on the threshold interval division of the "red, orange, yellow, and blue" alarm level commonly used in coal mines. The specific threshold intervals and corresponding scores are as follows: When the threshold interval is [0%, 0.55%), the alarm color is blue and the alarm score is 1; When the threshold interval is [0.55%, 0.7%), the alarm color is yellow and the alarm score is 2; When the threshold interval is [0.7%, 1.0%), the alarm color is orange and the alarm score is 3; When the threshold interval is 1.0% and above, the alarm color is red and the alarm score is 4.

4. The real-time over-limit warning method based on gas concentration characteristic changes according to claim 1 is characterized in that: During data processing, the bandwidth determination method for increasing gas concentration is used to quantify the severity of gas concentration fluctuations and to capture rapid concentration increases through bandwidth increase trends. The specific steps are as follows: Step 1: Parameter setting: Set the sliding window w, bandwidth multiplier k, and continuous growth threshold h1; Step 2: Calculate the moving average and standard deviation: For each time t, based on the first w sampling points, i.e., the sliding window p t-w+1 ,…,p t Calculate the moving average and standard deviation using the following formula: Among them, SMA t and STDs t They represent the moving average and standard deviation at time t respectively; Step 3: Calculate the upper and lower tracks and bandwidth using the following formula: Upper t =SMA t +k*STD t Lower t =SMA t -k*STD t Band t =Upper t -Lower t Among them, Upper t and Lower t Respectively represent the upper and lower rails at time t; Band t represents the bandwidth at time t; Step 4: Amplification judgment and alarm: Compare the moving average, upper track line, and bandwidth of the current moment with those of the previous moment; when the three show an upward trend at the same time, that is, SMA t >SMA t-1 ,Upper t >Upper t-1 ,Band t >Band t-1 If the condition is met continuously within h1 sampling points, the corresponding alarm score score will be triggered according to the concentration interval. a , and output the trigger time point.

5. The real-time over-limit warning method based on gas concentration characteristic changes according to claim 1 is characterized in that: In data processing, the exponential smoothing difference method is used to quantify the trend change and acceleration intensity of gas concentration, and to capture signal reversal or acceleration scenarios through short-term and long-term exponential smoothing differences. The specific steps are as follows: Step 1: Parameter setting: Set the short-term exponential moving average window s w , long-term exponential moving average window l w ;Signal line smoothing window sl w , continuous growth threshold h2; Step 2: Calculate the short-term and long-term exponential moving averages (EMA): For each time instant t, use the short-term exponential smoothing factor and the long-term exponential smoothing factor Calculate short-term and long-term exponential moving averages; in, represents the short-term exponential moving average at time t, represents the long-term exponential moving average at time t; P t Indicates the gas concentration value at time t; Step 3: Find the difference line DIF and the signal line SIG SIG(t)=(DIF(t-sl w +1)+…+DIF(t)) / sl w Histogram(t)=DIF(t)-SIG(t) Where DIF(t) represents the difference line at time t, SIG(t) represents the signal line at time t, and Histogram(t) represents the difference value at time t; Step 4: Amplification judgment and alarm: When any sequence rises continuously at adjacent moments, such as DIF(t)>DIF(t-1), SIG(t)>SIG(t-1) or Histogram(t)>Histogram(t-1); if it is continuously satisfied within the continuous growth of h2 sampling points, the corresponding alarm score score is triggered according to the concentration interval classification e , and output the trigger time point.

6. The real-time over-limit warning method based on gas concentration characteristic changes according to claim 1 is characterized in that: In data processing, the sliding window Z-score method is used to quantify abnormal deviations in the statistical distribution of gas concentration. Abnormal concentration peaks are captured by comparing the Z-score within the sliding window with the quantile threshold. The specific steps are as follows: Step 1: Parameter setting: Set the sliding window w, the quantile threshold Q, and the continuous exceeding threshold h3; Step 2: Calculate the moving average and standard deviation: For each time t, based on the first w sampling points, i.e., the sliding window p t-w+1 ,…,p t Calculate the moving average and standard deviation using the following formula: Among them, SMA t and STDs t They represent the moving average and standard deviation at time t respectively; Step 3: Find the Z score at the current moment: Among them, Z t represents the Z score at time t; Step 4: Determine the dynamic threshold: In the same sliding window, sort the first w Z scores and take the Qth quantile as the Threshold t ;Threshold t =Quantile({Z t-w+1 ,…,Z t },Q); Step 5: Anomaly detection and alarm: Determine the current Z t Is it greater than Threshold? t and Z at the previous moment t-1 Compared with the previous period, there is an upward trend, that is, Z t >Z t-1 If the condition is met continuously within h3 sampling points, the corresponding alarm score will be triggered according to the concentration interval. z , and output the trigger time point.

7. The real-time over-limit warning method based on gas concentration characteristic changes according to claim 3 is characterized in that: The voting fusion rules are: If any algorithm is not triggered at time t, its score is considered 0; Red priority: If any vote is rated as a red warning, a red warning will be output directly and subsequent score comparisons will be terminated; All-vote warning: When all three algorithms are triggered (score > 0), and at least one of them is a red warning, the highest score and corresponding level are taken; Majority of the same level: When all three algorithms are triggered and there is no red alert, the one with the most votes will be adopted first; if the three scores are different, the highest score will prevail; Double-vote evaluation: When there are exactly two algorithms triggered, if the two votes have the same score, the higher score will be used; if they are not equal, the higher score will be used; Single-vote assessment: When only one vote is triggered, the algorithm score is directly used as the final warning level.

8. The real-time over-limit warning method based on gas concentration characteristic changes according to claim 3 is characterized in that: The visual output specifically displays warning information through a graphical interface, including level identification, risk color and algorithm traceability information, and supports two-dimensional and three-dimensional visual display of data.

9. A system applicable to the real-time over-limit warning method based on gas concentration characteristic changes according to any one of claims 1 to 8, characterized in that: The system includes: data acquisition layer, data processing layer, decision analysis layer and visualization output layer; The data collection layer includes a gas sensor network deployed at the working and tunneling faces of the coal mine. The network consists of multiple fixed-installed gas concentration monitoring devices and portable gas concentration monitoring devices carried by workers. The data processing layer includes a data preprocessing module and an algorithm calculation module; the data preprocessing module filters the collected gas concentration time series data to eliminate gas pulsation and ventilation fluctuations; the algorithm calculation module executes three real-time detection algorithms in parallel: the amplitude change bandwidth determination method, the exponential smoothing difference method, and the sliding window Z score method; among them, the amplitude change bandwidth determination method quantifies the severity of gas concentration fluctuations through sliding window and bandwidth multiplier calculation; the exponential smoothing difference method uses short-term and long-term exponential moving averages to calculate and quantify concentration trend changes; and the sliding window Z score method captures abnormal concentration peaks by comparing the Z-score within the sliding window with the quantile threshold. The decision-making analysis layer includes a concentration interval classification module and a voting fusion module. The concentration interval classification module dynamically locates the risk interval based on the gas concentration value of the current sampling point and generates a concentration classification result. The voting fusion module performs a fusion analysis based on the calculation results of the algorithm calculation module to generate a dynamic alarm level. The visual output layer displays warning information through a graphical interface, including level identification, risk color and algorithm traceability information, and supports two-dimensional and three-dimensional visual display of data.

Images