Self-adaptive gas over-limit long-time prediction method and system

Abstract

The invention relates to the technical field of coal mine gas overrun prediction, and provides a self-adaptive gas overrun long-time prediction method and system. The method comprises the following steps: processing acquired coal mine underground multi-source data, and constructing different time scale characteristics of the data; adaptively screening an optimal prediction index combination according to the Pearson correlation coefficients of the target gas concentration for the different time scale characteristics; and determining an optimization parameter combination of the Informer long-time prediction model, optimizing the Informer long-time prediction model based on a grey wolf optimization algorithm to obtain an optimal prediction model so as to carry out over-limit long-time prediction on the gas concentration of the working face, and carrying out comprehensive evaluation on a prediction result based on a predetermined accuracy evaluation index. Therefore, the problems that in the prior art, gas overrun disaster-causing factors in different mining areas, different coal mines and different working faces are large in difference, the generalization ability is weak, and the long-time prediction ability is insufficient are effectively solved.

Description

Technical Field

[0001] The present application relates to the technical field of coal mine gas over-limit prediction, and in particular to an adaptive gas over-limit long-term prediction method and system. Background Art

[0002] Coal mine gas disasters are one of the major disasters threatening coal mine safety production. Gas disasters directly hinder the normal production of coal mines and hinder the sustained, stable and healthy development of the coal industry. Strengthening gas disaster prevention and control is an important guarantee for ensuring a stable and reliable supply of coal energy and promoting the comprehensive and healthy development of the national economy.

[0003] With the high attention paid to coal mine safety production and the need for the development of coal mining enterprises themselves, accurate prediction of gas concentration, disaster forecasting and prevention of gas outbursts are of great significance. Traditional gas prediction methods are determined based on quantitative indicators of gas-containing coal properties and their occurrence conditions, such as coal seam property indicators, gas indicators, ground stress indicators or comprehensive indicators, to predict whether a single or multiple indicators exceed the critical value, but the prediction accuracy of traditional prediction methods is difficult to meet the requirements of coal mine safety production.

[0004] At present, there is one method to predict the gas concentration in coal mines. By simulating the diffusion law of gas in the coal mining process, a parametric equation is constructed to solve the problem. The prediction results of this method are too idealized and cannot actually reflect the complex environment of coal mines. Another method is to build a prediction model through feature engineering or deep learning technology to predict the gas concentration. However, the features extracted by this method cannot fully express the changing trend of gas concentration under different environments, resulting in low prediction accuracy.

[0005] Therefore, there is an urgent need to provide a technical solution to the above-mentioned deficiencies in the prior art. Summary of the invention

[0006] The purpose of the present application is to provide an adaptive gas over-limit long-term prediction method and system to solve or alleviate the problems existing in the above-mentioned prior art.

[0007] In order to achieve the above objectives, this application provides the following technical solutions:

[0008] The present application provides an adaptive long-term prediction method for gas overlimit, comprising: step S101, processing the acquired multi-source data {T0, T1, T2, T7, FS, W, GT, GL, GW, GP} of a coal mine underground, and constructing different time scale characteristics of the multi-source data of the coal mine underground; wherein, T0 is the gas concentration at the upper corner of the working face, T1 is the gas concentration of the working face, T2 is the gas concentration of the return air lane, T7 is the gas concentration of the refuge chamber, FS is the wind speed of the working face, W is the temperature of the working face, GT is the gas extraction concentration of the goaf, GL is the gas extraction flow rate of the goaf, GW is the gas extraction temperature of the goaf, and GP is the gas extraction pressure of the goaf;

[0009] Step S102: Adaptively screen the best prediction index combination of the working face gas concentration T1 according to the Pearson correlation coefficient between the working face gas concentration T1 and the characteristics of different time scales; Step S103: Based on the Grey Wolf optimization algorithm, optimize the optimization parameter combination of the pre-constructed Informer long-term prediction model to obtain the best prediction model; Step S104: Based on the best prediction model, perform over-limit prediction of the working face gas concentration T1, and comprehensively evaluate the prediction results based on predetermined accuracy evaluation indicators.

[0010] Preferably, in step S101, based on the Holt exponential smoothing algorithm, the acquired multi-source data of the coal mine underground is processed for missing values; the multi-source data of the coal mine underground after the missing value processing is standardized, and the standardized data is identified for outliers based on the 3σ method; based on the autoregressive model, the outliers identified in the multi-source data of the coal mine underground are processed; for the multi-source data of the coal mine underground after the outlier processing, the maximum values ​​Y of 5min, 30min, 1h, and 24h are constructed respectively. t，max , minimum value Y t，min , mean value Y t，ave , difference value Y t，dif .

[0011] Preferably, according to the autoregressive model:

[0012]

[0013] Processing the outliers identified in the multi-source data of the coal mine underground; wherein M is the outlier at time i in the multi-source data of the coal mine underground; c is a constant term; p is an autoregressive order, indicating a time interval; is the autocorrelation coefficient; ε is the random error value; X i-j is the historical observation value at time j before time i in the multi-source data.

[0014] Preferably, according to the formula:

[0015]

[0016] The maximum values ​​Y of 5min, 30min, 1h and 24h are extracted from the multi-source data of the coal mine after outlier processing. t，max , minimum value Y t，min , mean value Y t，ave , difference value Y t，dif ; where t represents different time scales; x t represents the multi-source data of the coal mine after outliers are processed within the time range t; x t-1 It represents the multi-source data of the coal mine underground after the outliers are processed within the last time range of t.

[0017] Preferably, step S102 includes: calculating the Pearson correlation coefficient between the gas concentration T1 of the working face and the characteristics of different time scales, and adaptively extracting the gas over-limit prediction index whose absolute value of the Pearson correlation coefficient is in [0.4, 1] to form the optimal prediction index combination.

[0018] Preferably, in step S103, the number of multi-head probabilistic sparse self-attention mechanism layers, the number of self-attention distillation layers, and the number of fully connected layers of the Informer long-term prediction model are determined as the optimization parameter combination of the Informer long-term prediction model.

[0019] Preferably, in step S103, the Informer long-term prediction model includes: an encoder and a decoder; the encoder includes 3 identical layers stacked together, and each layer includes a multi-head probabilistic sparse self-attention mechanism sublayer and a self-attention distillation sublayer; the decoder includes 2 identical layers stacked together, and each layer includes a masked multi-head probabilistic sparse self-attention mechanism sublayer and a self-attention mechanism sublayer.

[0020] Preferably, in step S104, according to the formula:

[0021]

[0022] Calculate the mean absolute error MAE, mean square error MSE, root mean square error RMSE, and determination coefficient R between the prediction result obtained by the optimal prediction model and the true value of the gas concentration of the target working face. 2 ; Among them, H k is the true value of the kth gas concentration of the working face; h k is the prediction result corresponding to the true value of the kth gas concentration; J k is the average of the K true values ​​of gas concentration in the working face; k∈K, k and K are both positive integers.

[0023] The embodiment of the present application also provides an adaptive gas over-limit long-term prediction system, comprising:

[0024] The source data unit is configured to process the acquired multi-source data {T0, T1, T2, T7, FS, W, GT, GL, GW, GP} of the coal mine underground to construct different time scale characteristics of the multi-source data of the coal mine underground; wherein, T0 is the gas concentration at the upper corner of the working face, T1 is the gas concentration at the working face, T2 is the gas concentration at the return air lane, T7 is the gas concentration at the refuge chamber, FS is the wind speed at the working face, W is the temperature at the working face, GT is the gas extraction concentration in the goaf, GL is the gas extraction flow in the goaf, GW is the gas extraction temperature in the goaf, and GP is the gas extraction pressure in the goaf;

[0025] An indicator unit configured to adaptively select the best prediction indicator combination of the working face gas concentration according to the Pearson correlation coefficient of the working face gas concentration to the different time scale characteristics;

[0026] A parameter optimization unit is configured to optimize the optimization parameter combination of the pre-built Informer long-term prediction model based on the Gray Wolf Optimization Algorithm to obtain the best prediction model;

[0027] The comprehensive evaluation unit is configured to make an over-limit prediction of the gas concentration of the working face based on the optimal prediction model, and to make a comprehensive evaluation of the prediction results based on a predetermined accuracy evaluation index.

[0028] Beneficial effect: The present application provides an adaptive long-term prediction method for gas over-limit, which processes the acquired multi-source sensor data in coal mines, constructs different time scale characteristics of the multi-source data in coal mines, and adaptively screens the best prediction index combination of the gas concentration in the working face according to the Pearson correlation coefficient between the gas concentration T1 in the working face of the coal mine and the characteristics of different time scales; then, based on the Gray Wolf optimization algorithm, optimizes the optimization combination parameters of the pre-constructed Informer long-term prediction model to obtain the best prediction model; finally, based on the best prediction model, predicts the over-limit of the gas concentration T1 in the working face, and comprehensively evaluates the prediction results based on the predetermined accuracy evaluation index. In this way, the problems in the existing technology of large differences in the disaster-causing factors of gas exceeding the limit in different mining areas, different coal mines, and different working faces, weak generalization ability, and insufficient long-term prediction ability are effectively solved. The best prediction indicators are adaptively selected for different mining areas, different coal mines, and different working faces, and the long-term scale characteristics of gas exceeding the limit are constructed, so as to predict the gas concentration of the working face, greatly improve the long-term prediction effect of gas concentration exceeding the limit in the working face, and increase the time for exceeding the limit disposal. It is particularly suitable for the prediction of gas exceeding the limit in high-gas mines. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The drawings constituting part of the present application are used to provide a further understanding of the present application. The exemplary embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. Among them:

[0030] Figure 1 A schematic flow chart of an adaptive long-term prediction method for excessive gas flow provided according to some embodiments of the present application;

[0031] Figure 2 A schematic diagram of a framework of an Informer long-term prediction model provided according to some embodiments of the present application;

[0032] Figure 3 A schematic diagram in the form of a heat map of the gas concentration T1 of a working face and the Pearson correlation coefficient r of different time scale characteristics provided according to some embodiments of the present application;

[0033] Figure 4 A schematic diagram of the structure of an adaptive gas over-limit long-term prediction system provided according to some embodiments of the present application. DETAILED DESCRIPTION

[0034] The present application will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments. Each example is provided by way of explanation of the present application and does not limit the present application. In fact, it will be clear to those skilled in the art that modifications and variations may be made in the present application without departing from the scope or spirit of the present application. For example, a feature shown or described as a part of an embodiment may be used in another embodiment to produce yet another embodiment. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the embodiments of the present invention should belong to the scope of protection of the embodiments of the present invention.

[0035] Gas accidents are the main factor threatening the safe production of coal mines. Real-time and accurate monitoring of gas concentration is an important means to prevent gas accidents. In the safe production of coal mines, accurate and timely prediction of gas concentration can effectively reduce and prevent coal mine gas accidents.

[0036] Gas exceeding the limit is one of the main causes of gas accidents. Coal mining faces are frequent locations of gas exceeding the limit. However, the disaster-causing factors of gas exceeding the limit are complex. They are not only closely related to objective factors such as gas occurrence characteristics, gas geology, coal rock stress and coal body structure, but also affected by human factors such as mine excavation deployment and layout, ventilation methods, coal mining technology, operation methods, and personnel quality. At the same time, the factors that lead to gas exceeding the limit in each mining area, each coal mine, and even each working face are not the same.

[0037] At present, most coal mines only have the function of gas over-limit alarm, but lack the function of gas over-limit prediction. A small number of coal mines with gas over-limit prediction function also have problems such as single prediction index, weak generalization ability, poor prediction effect, and cannot automatically adapt to the characteristics of each working face of each coal mine in each mining area. In addition, the existing methods can only make short-term (5-15 minutes) predictions, and the long-term (greater than 1 hour) predictions have large errors and poor prediction effects. Therefore, on the basis of comprehensively considering the characteristics and differences of each working face of each coal mine in each mining area, adaptively selecting prediction indicators, and deeply exploring the time series characteristics of gas concentration, and making long-term predictions of gas over-limit can move the technical threshold of gas over-limit prevention and control forward and increase the time of emergency decision-making, which is of great significance for preventing and controlling coal mine gas and ensuring coal mine safety production.

[0038] Based on this, the embodiment of the present application proposes an adaptive long-term prediction method and system for gas excess limit, which adaptively selects the optimal prediction index according to the characteristics of each working face of each coal mine in each mining area, and extracts the different time scale characteristics of gas excess limit, so as to make an advance prediction of gas concentration, so as to greatly improve the long-term prediction effect of gas excess limit and increase the excess limit disposal time. It is suitable for the prediction of gas excess limit in mines, especially for the prediction of gas excess limit in high-gas mines.

[0039] like Figures 1 to 3 As shown, the adaptive gas over-limit long-term prediction method includes:

[0040] Step S101: Process the acquired multi-source data {T0, T1, T2, T7, FS, W, GT, GL, GW, GP} of the underground coal mine to construct different time scale characteristics of the multi-source data of the underground coal mine.

[0041] In this application, multi-source data is obtained through multi-source sensors in coal mines. Specifically including: gas concentration T0 in the upper corner of the working face, gas concentration T1 in the working face, gas concentration T2 in the return air lane, gas concentration T7 in the safe-haven chamber, wind speed FS in the working face, temperature W in the working face, gas extraction concentration GT in the goaf, gas extraction flow GL in the goaf, gas extraction temperature GW in the goaf, and gas extraction pressure GP in the goaf. Here, it should be noted that the acquired multi-source data is a data set containing {T0, T1, T2, T7, FS, W, GT, GL, GW, GP} within a certain time period, that is, the multi-source data contains multiple groups of {T0, T1, T2, T7, FS, W, GT, GL, GW, GP} at the same time.

[0042] After obtaining the multi-source data of coal mine underground, the multi-source data of coal mine underground is preprocessed, including: missing value processing, standardization processing and outlier processing. Among them, based on the Holt exponential smoothing algorithm, the missing value processing of the obtained multi-source data of coal mine underground is carried out according to the formula:

[0043]

[0044] α and β are the built-in smoothing parameters of Holt exponential smoothing method, which can be 0.5. i is the missing value at time i, b i , b i-1 are the trend values ​​at time i and time i-1, respectively, S i , S i-1 The smoothing values ​​at time i and time i-1 respectively, L is the prediction step size, which is 5 here.

[0045] Then, the multi-source data of coal mine underground after missing value processing is standardized. Specifically, it is standardized according to the maximum and minimum:

[0046]

[0047] Standardization is performed on the multi-source data of underground coal mines after missing value processing. Among them, y is the standardized value of multi-source data of underground coal mines; x' is the value of multi-source data of underground coal mines after filling with Holt exponential smoothing algorithm; max(x′) is the maximum value of multi-source data of underground coal mines after filling with Holt exponential smoothing algorithm; min(x′) is the minimum value of multi-source data of underground coal mines after filling with Holt exponential smoothing algorithm.

[0048] By standardizing multi-source data, the influence of data dimension differences between different indicators can be effectively eliminated. Then, outliers are identified on the standardized data based on the 3σ method. Specifically, multi-source data obeys a normal distribution with a mean u and a standard deviation σ. If the data is not within the range of (u-3σ, u+3σ), it can be considered an outlier.

[0049] Next, through the autoregressive model:

[0050]

[0051] The outliers identified in the multi-source data of the coal mine are processed. Where M is the outlier at time i in the multi-source data of the coal mine; c is a constant term; p is the autoregressive order, which represents the time interval; is the autocorrelation coefficient; ε is the random error value, which means a white noise signal with a mean of 0 and a constant standard deviation of s; X i-j is the historical observation value at time j before time i in the multi-source data.

[0052] Finally, according to the formula:

[0053]

[0054] The maximum values ​​Y of 5min, 30min, 1h and 24h are extracted from the multi-source data of coal mine underground after outlier processing. t，max , minimum value Y t，min , mean value Y t，ave , difference value Y t，dif . In the formula, t represents different time scales; x t represents the multi-source data of coal mine underground after outliers are processed within the time range t; x t-1 Represents the multi-source data of coal mine underground after outlier processing within the last time range of t.

[0055] Here, it should be noted that the multi-source data include multiple groups of {T0, T1, T2, T7, FS, W, GT, GL, GW, GP} at the same time. When preprocessing the data, each of the multiple groups of {T0, T1, T2, T7, FS, W, GT, GL, GW, GP} at the same time is preprocessed separately; similarly, the different time scale features of the structure include the structural feature {Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of gas concentration T1 at working face {Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of gas concentration T2 in the return air lane {Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of the gas concentration T7 in the refuge chamber {Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of wind speed FS {Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of temperature W{Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of gas extraction concentration GT in goaf {Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of gas drainage flow GL in goaf {Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of gas temperature GW in gas extraction in goaf {Y t，max, Y t，min , Y t，ave , Y t，dif}、Structural characteristics of gas extraction pressure GP in goaf {Y t，max , Y t，min , Y t，ave , Y t，dif}.

[0056] Step S102, adaptively screening the best prediction index combination of the working face gas concentration T1 according to the Pearson correlation coefficient between the working face gas concentration T1 and different time scale characteristics.

[0057] The Pearson correlation coefficient is a statistic that describes the strength of the linear correlation between two variables. According to the formula:

[0058]

[0059] Calculate the Pearson correlation coefficient r between the working face gas concentration T1 and different time scale characteristics. j The main indicator is the gas concentration T1, Y j It is the construction index after multi-source data is constructed at different time scales. Main indicator X j The average value of is the characteristic index Y j is the average value of , and Q is the total number of characteristic variables.

[0060] In this application, the structural characteristics of the gas concentration T0 at the corner of the working surface {Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of gas concentration T1 at working face {Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of gas concentration T2 in the return air lane {Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of the gas concentration T7 in the refuge chamber {Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of wind speed FS {Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of temperature W{Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of gas extraction concentration GT in goaf {Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of gas drainage flow GL in goaf {Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of gas extraction temperature GW in goaf {Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of gas extraction pressure GP in goaf {Y t，max , Y t，min , Y t，ave , Y t，dif}As a structural indicator, the Pearson correlation coefficient r between the working face gas concentration T1 and different time scale characteristics is calculated, that is, the Pearson correlation coefficient between the working face gas concentration T1 and different structural indicators is calculated.

[0061] That is, the Pearson correlation coefficient between the working face gas concentration T1 and each item in the structural index is calculated. For example, the structural characteristics {Y t，max , Y t，min , Y t，ave , Y t，dif}, the structural characteristics of the working face gas concentration T1 and the working face gas concentration T1 {Y t，max , Y t，min , Y t，ave , Y t，dif}, the structural characteristics of the gas concentration T1 at the working face and the gas concentration T2 in the return air lane {Y t，max , Y t，min , Y t，ave , Y t，dif}, the structural characteristics of the gas concentration T1 in the working face and the gas extraction concentration GT in the goaf {Y t，max , Y t，min , Y t，ave , Y t，dif The Pearson correlation coefficient between}.

[0062] The Pearson correlation coefficient r is between [-1, 1], and two variables whose absolute values ​​are between [0.1, 0.3] are called weak correlation, two variables whose absolute values ​​are between [0.4, 0.7] are called moderate correlation, and two variables whose absolute values ​​are between [0.8, 1] are called strong correlation. In this application, according to the Pearson correlation coefficient r, medium and strong correlation indicators are adaptively extracted as the best prediction indicator combination of the long-term prediction model, that is, the gas over-limit prediction indicators whose absolute values ​​of the Pearson correlation coefficient are between [0.4, 1] are extracted to form the best prediction indicator combination.

[0063] Here, it should be noted that the gas over-limit prediction index is the structural index, that is, the structural characteristics of the gas concentration T0 at the corner of the working face {Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of gas concentration T1 at working face {Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of gas concentration T2 in the return air lane {Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of the gas concentration T7 in the refuge chamber {Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of wind speed FS {Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of temperature W{Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of gas extraction concentration GT in goaf {Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of gas drainage flow GL in goaf {Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of gas extraction temperature GW in goaf {Y t，max , Y t，min , Y t，ave , Y t，dif}、Structural characteristics of gas extraction pressure GP in goaf {Y t，max , Y t，min , Y t，ave , Y t，difFrom the calculated Pearson correlation coefficient between the working face gas concentration t1 and each item in the structural index, the gas over-limit prediction index with the absolute value of the Pearson correlation coefficient in [0.4, 1] is extracted to form the best prediction index combination.

[0064] Step S103: Based on the Grey Wolf Optimization Algorithm, the optimization parameter combination of the pre-built Informer long-term prediction model is optimized to obtain the best prediction model.

[0065] In this application, the Informer long-term prediction model is built using the PyTorch open source deep learning framework, which mainly includes: an encoder (Encoder) and a decoder (Decoder). Among them, the encoder is composed of 3 identical layers stacked together, and each layer contains a multi-head probability sparse self-attention mechanism sublayer and a self-attention distillation sublayer; the decoder is composed of 2 identical layers stacked together, and each layer contains a masked multi-head probability sparse self-attention mechanism sublayer and a self-attention mechanism sublayer. The encoder extracts the long-term dependency of the temporal features of multi-source sensor data (multi-source data), and the decoder is mainly used to achieve long-term prediction.

[0066] The best parameter combination is determined by continuously modifying the parameters of the Informer long-term prediction model, and then the best prediction model is obtained. Specifically, the number of multi-head probabilistic sparse self-attention mechanism layers, the number of self-attention distillation layers, and the number of fully connected layers of the Informer long-term prediction model are determined as the optimized parameter combination of the Informer long-term prediction model.

[0067] In a specific example, first, the position and size of the gray wolf group in the gray wolf optimization algorithm are initialized, the number of gray wolf populations is set to 50, and the mean square error MSE of the Informer model is used as the fitness function. Then, the fitness of the gray wolf individuals is calculated 50 times in a loop, the position of the gray wolf individuals is updated according to the fitness, the current fitness of the model is compared with the historical optimal fitness, the optimal gray wolf position and the optimal fitness are output, and then the optimal parameter combination for optimization is obtained, that is, the optimized parameter combination.

[0068] Step S104: Based on the optimal prediction model, the over-limit prediction of the gas concentration T1 of the working face is made, and based on the predetermined accuracy evaluation index, the prediction result is comprehensively evaluated.

[0069] After obtaining the best Informer prediction model, the best prediction model is deployed to the edge box (i.e., edge server) of the underground working face of the coal mine. The edge box is connected to the monitoring substation and monitoring sensor in sequence. Through the established multi-source data real-time acquisition module, the multi-source data {T0, T1, T2, T7, FS, W, GT, GL, GW, GP} is transmitted in real time or periodically, and the incoming multi-source data is processed with missing values, standardized, processed with outliers, constructed with different time scale features, and screened with adaptive indicators. The model receives the processed data and makes a prediction, outputs the gas concentration T1 of the working face in the next hour, compares the gas concentration T1 of the working face in the next hour with the safety threshold of the gas concentration T1 of the working face set by the coal mine, and issues an audible and visual alarm signal when the safety threshold is exceeded.

[0070] In this application, the accuracy of the prediction results of the Informer long-term prediction model is evaluated by predicting the over-limit of the gas concentration T1 of the working face and based on the gas concentration monitoring data in the coal mine. Specifically, according to the formula:

[0071]

[0072] Calculate the mean absolute error (MAE), mean square error (MSE), root mean square error (RMSE), and determination coefficient (R) between the prediction results obtained by the optimal prediction model and the actual value of the gas concentration at the target working face. 2 ; Among them, H k is the true value of the kth gas concentration of the working face, that is, the gas concentration of the working face obtained by monitoring; h k is the prediction result corresponding to the true value of the k-th gas concentration, that is, the gas concentration of the working face predicted by the Informer long-term prediction model; J k is the mean of the K true values ​​of gas concentration in the working face; k∈K, k and K are both positive integers.

[0073] In this application, the multi-source data obtained from underground coal mines are processed through the Holt exponential smoothing algorithm, maximum and minimum standardization, autoregression method, etc., and the data structure features are extracted in a targeted manner; the best prediction index is adaptively selected according to the Pearson correlation coefficient; the Informer time series prediction model constructed based on the encoder (Encoder), decoder (Decoder) and fully connected layer can further capture the long-term dependency of data and realize long-term prediction of gas exceeding the limit. It is flexible and reliable, with strong prediction ability, and is particularly suitable for the prediction of gas exceeding the limit in high-gas mines, especially for the prediction of gas exceeding the limit in intelligent coal mining working faces.

[0074] The present application also provides an adaptive gas over-limit long-term prediction system. Figure 4 As shown, the prediction system includes:

[0075] The source data unit 401 is configured to process the acquired multi-source data {T0, T1, T2, T7, FS, W, GT, GL, GW, GP} of the coal mine underground to construct different time scale characteristics of the multi-source data of the coal mine underground; wherein, T0 is the gas concentration at the upper corner of the working face, T1 is the gas concentration of the working face, T2 is the gas concentration of the return air lane, T7 is the gas concentration of the refuge chamber, FS is the full working wind speed, W is the working face temperature, GT is the gas extraction concentration of the goaf, GL is the gas extraction flow rate of the goaf, GW is the gas extraction temperature of the goaf, and GP is the gas extraction pressure of the goaf;

[0076] The indicator unit 402 is configured to adaptively select the best prediction indicator combination of the gas concentration in the working face according to the Pearson correlation coefficient of the gas concentration in the working face with different time scale characteristics;

[0077] The parameter optimization unit 403 is configured to optimize the optimization parameter combination of the pre-built Informer long-term prediction model based on the Gray Wolf Optimization Algorithm to obtain the best prediction model;

[0078] The comprehensive evaluation unit 404 is configured to perform over-limit prediction of the gas concentration in the working face based on the optimal prediction model, and to perform a comprehensive evaluation of the prediction results based on a predetermined accuracy evaluation index.

[0079] The adaptive gas excess long-term prediction system provided in the embodiments of the present application can implement the steps and processes of the adaptive gas excess long-term prediction method of any of the above embodiments and achieve the same technical effects, which will not be described one by one here.

[0080] In the description of the present invention, the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" etc. mean that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0081] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. An adaptive long-term prediction method for excessive gas limit, characterized in that: include: Step S101, processing the acquired coal mine underground multi-source data {T0, T1, T2, T7, FS, W, GT, GL, GW, GP}, and constructing different time scale characteristics of the coal mine underground multi-source data; wherein, T0 is the gas concentration in the upper corner of the working face, T1 is the gas concentration in the working face, T2 is the gas concentration in the return air lane, T7 is the gas concentration in the refuge chamber, FS is the working face wind speed, W is the working face temperature, GT is the gas extraction concentration in the goaf, GL is the gas extraction flow in the goaf, GW is the gas extraction temperature in the goaf, and GP is the gas extraction pressure in the goaf; Step S102, adaptively screening the best prediction index combination of the working face gas concentration T1 according to the Pearson correlation coefficient between the working face gas concentration T1 and the different time scale characteristics; Step S103: Based on the Gray Wolf Optimization Algorithm, the optimization parameter combination of the pre-built Informer long-term prediction model is optimized to obtain the best prediction model; Step S104: Based on the optimal prediction model, the over-limit prediction of the gas concentration T1 of the working face is performed, and based on the predetermined accuracy evaluation index, the prediction result is comprehensively evaluated.

2. The adaptive long-term prediction method for excessive gas limit according to claim 1, characterized in that: In step S101, Based on the Holt exponential smoothing algorithm, missing value processing is performed on the acquired multi-source data of the coal mine underground; Standardizing the coal mine underground multi-source data after missing value processing, and identifying outliers on the standardized data based on the 3σ method; Based on the autoregressive model, outliers identified in the multi-source data of the coal mine are processed; The maximum values ​​Y of 5min, 30min, 1h and 24h are constructed for the multi-source data of the coal mine after outlier processing. t,max , minimum value Y t,min , mean value Y t,ave , difference value Y t,dif .

3. The adaptive long-term prediction method for excessive gas limit according to claim 2 is characterized in that: According to the autoregressive model: Processing the outliers identified in the multi-source data of the coal mine underground; wherein M is the outlier at time i in the multi-source data of the coal mine underground; c is a constant term; p is an autoregressive order, indicating a time interval; is the autocorrelation coefficient; ε is the random error value; X i-j is the historical observation value at time j before time i in the multi-source data.

4. The adaptive long-term prediction method for excessive gas limit according to claim 2 is characterized in that: According to the formula: The maximum values ​​Y of 5min, 30min, 1h and 24h are extracted from the multi-source data of the coal mine after outlier processing. t,max , minimum value Y t,min , mean value Y t,ave , difference value Y t,dif ; Among them, t represents different time scales; x t represents the coal mine underground multi-source data after outlier processing within the time range t; x t-1 It represents the multi-source data of the coal mine underground after the outliers are processed within the last time range of t.

5. The adaptive long-term prediction method for excessive gas limit according to claim 1 is characterized in that: Step S102 includes: The Pearson correlation coefficient between the gas concentration T1 of the working face and the characteristics of different time scales is calculated, and the gas over-limit prediction index whose absolute value of the Pearson correlation coefficient is in [0.4, 1] is adaptively extracted to form the optimal prediction index combination.

6. The adaptive long-term prediction method for excessive gas limit according to claim 1, characterized in that: In step S103, Determine the number of multi-head probabilistic sparse self-attention mechanism layers, the number of self-attention distillation layers, and the number of fully connected layers of the Informer long-term prediction model as the optimized parameter combination of the Informer long-term prediction model.

7. The adaptive long-term prediction method for excessive gas limit according to claim 1, characterized in that: In step S103, the Informer long-term prediction model includes: an encoder and a decoder; The encoder is composed of three identical layers stacked together, and each layer includes a multi-head probabilistic sparse self-attention mechanism sublayer and a self-attention distillation sublayer; The decoder comprises two identical layers stacked together, and each layer comprises a masked multi-head probabilistic sparse self-attention mechanism sublayer and a self-attention mechanism sublayer.

8. The adaptive long-term prediction method for excessive gas limit according to claim 1 is characterized in that: In step S104, according to the formula: Calculate the mean absolute error MAE, mean square error MSE, root mean square error RMSE, and determination coefficient R between the prediction result obtained by the optimal prediction model and the true value of the gas concentration of the target working face. 2 ; Among them, H k is the true value of the kth gas concentration of the working face; h k is the prediction result corresponding to the true value of the kth gas concentration; J k is the average of the K true values ​​of gas concentration in the working face; k∈K, k and K are both positive integers.

9. An adaptive gas over-limit long-term prediction system, characterized in that: include: The source data unit is configured to process the acquired multi-source data {T0, T1, T2, T7, FS, W, GT, GL, GW, GP} of the coal mine underground to construct different time scale characteristics of the multi-source data of the coal mine underground; wherein, T0 is the gas concentration at the upper corner of the working face, T1 is the gas concentration of the working face, T2 is the gas concentration of the return air lane, T7 is the gas concentration of the refuge chamber, FS is the full working wind speed, W is the working face temperature, GT is the gas extraction concentration of the goaf, GL is the gas extraction flow rate of the goaf, GW is the gas extraction temperature of the goaf, and GP is the gas extraction pressure of the goaf; An indicator unit configured to adaptively select the best prediction indicator combination of the working face gas concentration according to the Pearson correlation coefficient of the working face gas concentration to the different time scale characteristics; A parameter optimization unit is configured to optimize the optimization parameter combination of the pre-built Informer long-term prediction model based on the Gray Wolf Optimization Algorithm to obtain the best prediction model; The comprehensive evaluation unit is configured to make an over-limit prediction of the gas concentration of the working face based on the optimal prediction model, and to make a comprehensive evaluation of the prediction results based on a predetermined accuracy evaluation index.

Images