Gas concentration over-limit early warning method and intelligent analysis equipment based on big data technology

Abstract

The present invention proposes a gas concentration over-limit early warning method and intelligent analysis equipment based on big data technology, the method comprising: obtaining historical time series data of gas concentration in a mine; performing model fitting according to the historical time series data of gas concentration and an improved Prophet model, and obtaining a fitted Prophet model, wherein the improved Prophet model comprises a gas concentration growth rate trend model, a seasonal trend model and a holiday effect model, the gas concentration growth rate trend model is obtained based on a tanh function, the seasonal trend model is a wavelet expansion of the seasonal trend of gas concentration, and the holiday effect model comprises the influence of statutory holidays and preset specific holidays on gas concentration; according to the fitted Prophet model, predicting the gas concentration in a future preset time period, and issuing an alarm for abnormal values. The present invention utilizes the fitted Prophet model to more accurately predict the gas concentration in a future preset time period and accurately alarm.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a gas concentration over-limit early warning method and intelligent analysis equipment based on big data technology. Background Art

[0002] Gas is an associated gas of coal resources and an important clean energy source. Its flammable and explosive nature makes it an important factor affecting coal mine safety production. With the mining of coal resources, the mining depth of mines increases year by year, and the gas content in coal seams gradually increases, making gas control work more complicated.

[0003] The mining face is the main production area of ​​the mine. During the mining process of gas mines, a large amount of gas flows into the tunnel from the coal wall, coal falling, goaf, adjacent layers and other areas, forming a mixed gas with the air and moving with the wind. Accurately judging the gas concentration is the basis for gas outburst prediction, gas explosion prevention, ventilation design and other work.

[0004] There are a large number of sensors in the underground monitoring system, such as gas detection sensors, carbon dioxide detection sensors, temperature sensors, humidity sensors, wind flow sensors, etc. In-depth data analysis can have significant benefits for mine mining. Scholars have done a lot of work on gas concentration prediction using sensor data and proposed many methods, including chaotic time series, autoregressive moving average model (ARMA), neural network, support vector machine, and various combination algorithms. The multi-source characteristics of working face gas and the migration characteristics of gas mixture make the gas concentration have both certain regularity and certain complexity, which is a typical nonlinear prediction problem.

[0005] The above algorithms are often trained by manually extracting features, which have shortcomings such as experience dependence, poor timeliness, and limitation to shallow data features. In addition, there are many factors that affect gas outburst, such as geological conditions, coal seam gas content, coal seam burial depth, adjacent layers, goaf and mining process, etc. Some factors are not time series data and cannot be quantified in real time and cannot be well integrated into the prediction model, which brings challenges to the accurate prediction of coal mine gas. Summary of the invention

[0006] The present invention provides a gas concentration over-limit early warning method and intelligent analysis equipment based on big data technology, the main purpose of which is to better fit the gas concentration data and effectively improve the accuracy of gas concentration prediction.

[0007] In a first aspect, an embodiment of the present invention provides a gas concentration over-limit early warning method based on big data technology, comprising:

[0008] Obtain historical time series data of gas concentration in mines;

[0009] According to the gas concentration historical time series data and the improved Prophet model, model fitting is performed to obtain the fitted Prophet model, wherein the improved Prophet model includes a gas concentration growth rate trend model, a seasonal trend model and a holiday effect model, the gas concentration growth rate trend model is obtained based on the tanh function, the seasonal trend model is a wavelet expansion of the seasonal trend of gas concentration, and the holiday effect model includes the influence of statutory holidays and preset specific holidays on gas concentration;

[0010] According to the fitted Prophet model, the gas concentration within a preset time period in the future is predicted, and an alarm is issued for abnormal values.

[0011] In a second aspect, an embodiment of the present invention provides a gas concentration over-limit early warning system based on big data technology, comprising:

[0012] A data acquisition module is used to obtain historical time series data of gas concentration in the mine;

[0013] A model fitting module, used for performing model fitting according to the gas concentration historical time series data and the improved Prophet model, and obtaining a fitted Prophet model, wherein the improved Prophet model includes a gas concentration growth rate trend model, a seasonal trend model and a holiday effect model, the gas concentration growth rate trend model is obtained based on a tanh function, the seasonal trend model is a wavelet expansion of the seasonal trend of gas concentration, and the holiday effect model includes the influence of statutory holidays and preset specific holidays on gas concentration;

[0014] The prediction and alarm module is used to predict the gas concentration within a preset time period in the future according to the fitted Prophet model and issue an alarm for abnormal values.

[0015] In the third aspect, an embodiment of the present invention provides an intelligent analysis device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned gas concentration excessive limit warning method based on big data technology when executing the computer program.

[0016] In a fourth aspect, an embodiment of the present invention provides a computer storage medium, wherein the computer storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned gas concentration excessive limit warning method based on big data technology are implemented.

[0017] The present invention proposes a gas concentration over-limit warning method and intelligent analysis equipment based on big data technology. According to the analysis of the gas concentration historical time series data, combined with the improved Prophet model, data fitting is performed to obtain the fitted Prophet model. Since the fitted Prophet model includes a gas concentration growth rate trend model, a seasonal trend model and a holiday effect model, the gas concentration growth rate trend model in the present invention is obtained based on the tanh function, and the tanh function can better retain the difference in the distribution amplitude of the gas concentration data, so that the gas concentration growth rate trend model can better simulate the trend distribution of the gas concentration time series data; the seasonal trend model in the present invention is the wavelet expansion of the seasonal trend of the gas concentration, and the wavelet transform can be constructed according to the signal characteristics of the gas concentration historical time series data, so as to more accurately describe the local characteristics of the gas concentration historical time series data; the holiday effect model in the present invention can better reflect the characteristics of the gas concentration time series data on the preset specific holidays. Therefore, using the fitted Prophet model, the gas concentration in the preset time period in the future can be more accurately predicted, and an alarm can be accurately issued. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A flowchart of a gas concentration over-limit early warning method based on big data technology provided in an embodiment of the present invention;

[0019] Figure 2 A schematic diagram of historical time series data of gas concentration provided by an embodiment of the present invention;

[0020] Figure 3 A schematic diagram of the structure of a gas concentration over-limit early warning system based on big data technology provided by an embodiment of the present invention;

[0021] Figure 4 A schematic diagram of the structure of an intelligent analysis device provided by an embodiment of the present invention.

[0022] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0023] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be understood as limiting the present application.

[0024] In order to enable those skilled in the art to better understand the solutions of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of the present application.

[0025] References to "one embodiment" or "some embodiments" etc. described in this specification mean that one or more embodiments of the present application include a particular feature, structure or characteristic described in conjunction with the embodiment. Thus, in this specification, the terms "include", "comprises", "has" and their variations all mean "including but not limited to", unless otherwise specifically emphasized.

[0026] In the prior art, a time series model for short-term prediction of gas sensor concentration is constructed to predict the future short-term gas concentration based on the historical data of gas concentration. If the predicted short-term gas concentration exceeds the limit, an advance prediction of gas exceeding the limit alarm is issued, and the losses caused by gas exceeding the limit and the resulting shutdown can be prevented in advance by adjusting the coal mining speed and other methods.

[0027] Common time series models include: traditional autoregressive moving average model (ARIMA), Prophet model, machine learning model, and deep learning model. The most widely used are Prophet model and deep learning model.

[0028] The ARIMA model uses statistical methods to perform differential processing on time series data to convert it into a stationary series, and uses the maximum likelihood method to estimate model parameters. Its advantages are that it has a solid statistical theoretical basis and fast analysis speed, but it has the following defects: it can only process stationary series or time series data that has been differentially processed into a stationary series, and the trend term only supports linear form. The environment of the gas monitoring point is complex, and the monitoring data changes are complex, which may not meet the requirements of the ARIMA model.

[0029] Machine learning models are represented by lightGBM (Light Gradient Boosting Machine) and XGBoost (eXtreme Gradient Boosting, distributed gradient boosting library). They generally transform time series problems into supervised learning and predict through feature engineering and machine learning. However, machine learning methods require complex manual feature processes, feature engineering requires rich professional knowledge and is time-consuming. The quality of feature engineering often determines the upper limit of prediction accuracy.

[0030] Deep learning methods are mainly based on LSTM (Long Short-Term Memory) and GRU (Gate Recurrent Unit), which have high prediction accuracy. However, deep learning methods require a large amount of data to train and adjust parameters, and there is little actual gas risk data, which makes their application difficult. In addition, deep learning methods require high computing resources, are complex to apply, and have poor interpretability, making them unsuitable for the actual situation of mining companies.

[0031] The Prophet model is based on time series decomposition (seasonal terms, trend terms, and residual terms) and machine learning fitting. It is widely used in business analysis and other fields and can handle situations with outliers and missing values. The Prophet algorithm model is a type of generalized additive statistical model (GAM) used to model the nonlinear relationship between independent variables and dependent variables, thereby performing time series prediction. In the existing Prophet model, the independent variable is time t and the dependent variable is the gas concentration of the sensor.

[0032] Specifically, the Prophet model decomposes the time series y(t) into the gas concentration growth rate trend term g(t), the seasonal trend term s(t), the holiday effect term h(t), and the error term ε t , and its calculation formula is as follows:

[0033] y(t)=g(t)+s(t)+h(t)+ε(t); (1)

[0034] Among them, the gas concentration growth rate trend term is used to represent the non-periodic changes in time series values, and the Prophet model is generally modeled through a logistic regression function or a piecewise linear function; the seasonal trend term represents the seasonal periodic changes in the time series, and the Prophet model is modeled through the Fourier series; the holiday effect represents the impact of holidays on the changes in time series, and the Prophet model is modeled by fitting the normal distribution of holiday influencing factors.

[0035] The shortcomings of the existing Prophet model are:

[0036] 1. The existing Prophet model models the trend term of the time series through logistic regression function or piecewise linear function, which has achieved good results in the field of commercial time series prediction such as software user volume change prediction and commodity sales prediction. However, before the occurrence of gas risk, the gas concentration often changes dramatically, and the trend term of logistic regression cannot describe this trend well. Therefore, in the field of gas concentration prediction, it is necessary to improve the trend term modeling of the existing Prophet model.

[0037] 2. The existing Prophet model uses the Fourier series to model the seasonal trend term of the time series. According to the characteristics of the non-stationary signal of the historical data of gas concentration, the Fourier series suitable for stationary signal modeling is not applicable to this scenario, and the modeling of seasonal trends needs to be improved.

[0038] 3. The existing Prophet model models the impact of holidays by fitting the normal distribution of holiday influencing factors. The holiday effect has a greater impact in the field of commercial forecasting. However, in addition to holidays, changes in gas concentration are also affected by other special dates. Therefore, the holiday effect term needs to be improved.

[0039] The present invention is based on an improved Prophet model and proposes an algorithm suitable for short-term prediction of gas sensor concentration by improving the modeling method of the gas concentration growth rate trend term, the seasonal trend term and the holiday effect term.

[0040] Figure 1 A flowchart of a gas concentration over-limit early warning method based on big data technology provided by an embodiment of the present invention is shown in FIG. Figure 1 As shown, the method includes:

[0041] S110, obtaining historical time series data of gas concentration in the mine;

[0042] Step 1: Collect the historical data of the gas sensors that need to be predicted in time series, clean and process the historical data, including processing missing values, outliers, and smoothing data. Convert the historical data into the input format required by the Prophet model, that is, two columns of data, one is the timestamp, and the other is the target variable, that is, the gas concentration.

[0043] S120, performing model fitting according to the gas concentration historical time series data and the improved Prophet model to obtain a fitted Prophet model, wherein the improved Prophet model includes a gas concentration growth rate trend model, a seasonal trend model and a holiday effect model, the gas concentration growth rate trend model is obtained based on a tanh function, the seasonal trend model is a wavelet expansion of the seasonal trend of gas concentration, and the holiday effect model includes the influence of statutory holidays and preset specific holidays on gas concentration;

[0044] Step 2: Construct a gas concentration growth rate trend model. The present invention improves the trend modeling method of the existing Prophet model. The specific improvement method is as follows:

[0045] The existing Prophet model modeling method for trend items is logistic regression function or piecewise linear function modeling. According to the characteristics of the distribution of historical data of gas concentration of sensors, fitting historical data through piecewise linear function requires dividing the overall time interval into more sub-intervals, which easily leads to overfitting of the model. The present invention chooses to improve the logistic regression function modeling method. Therefore, the calculation formula of the gas concentration growth rate trend model in the embodiment of the present invention is as follows:

[0046]

[0047] Wherein, g′(t) represents the gas concentration growth rate trend model, t represents time, k(t) represents the growth rate of gas concentration, m represents the offset parameter, and C(t) represents the upper limit value of gas concentration.

[0048] Step 3: Establishment of seasonal trend model. The existing Prophet model modeling method for seasonal trend terms is to use Fourier series modeling, while in the embodiment of the present invention, the trend term is improved to wavelet expansion, such as formula (3):

[0049]

[0050] Where s′(t) represents the seasonal trend model, j represents the scale parameter, k represents the translation parameter, φ j,k (t) represents the scaling function, ψ j,k (t) represents the wavelet function, c j,k represents the approximation coefficient, d j,k represents the detail coefficient, and t represents the time.

[0051] In the embodiment of the present invention, the Morlet wavelet function is selected as the basis, and the form is as follows (3):

[0052]

[0053] Step 4: Holiday effect model. The present invention improves the holiday effect modeling method of the Prophet algorithm model. The Prophet model modeling method for holidays is to fit the posterior distribution of holiday influencing factors. First, a holiday set is collected. For each holiday i, set D i ′ is the set of past and future dates of the holiday, and the indicator matrix is ​​defined as:

[0054] Z′(t)=[1(t∈D 1 ′),…,1(t∈D i ′),…,1(t∈D′ L )]; (4)

[0055] At the same time, an impact factor parameter μ is given to each holidayi ′, and μ′ obeys normal distribution, then the holiday effect term can be expressed as:

[0056] h′(t)=Z′(t)μ′; (5)

[0057] μ′=[μ 1 ′,…,μ i ′,…,μ′ L ]; (6)

[0058] Wherein, h′(t) represents the holiday effect model, Z′(t) represents the indicator matrix, and D i ′ represents the i-th holiday, L represents the total number of statutory holidays and the preset specific holidays, μ′ represents the influencing factor matrix, μ i ′ represents the impact factor of the i-th holiday, t represents the time, and i is a positive integer.

[0059] The present invention adds preset special holidays with large changes in gas sensor concentration, which are given by professionals with professional knowledge and rich practical experience, to the prescribed holiday set, such as the dates for regular maintenance and safety inspections of equipment, so as to more accurately fit the changing pattern of gas concentration.

[0060] Step 6: Combining the modeling results of steps 2, 3, and 4, the improved Prophet model of the present invention can be expressed as follows:

[0061] y′(t)=g′(t)+s′(t)+h′(t)+ε(t); (7)

[0062] Based on the parameters determined in step 5, the L-BFGS (Limited memory BFGS) algorithm is used to solve the model parameters and obtain the fitted Prophet model. L-BFGS is an algorithm for nonlinear optimization problems. It can effectively find the optimal solution for model parameters and is mainly used to solve parameter optimization problems in machine learning. It is an improved quasi-Newton method and is particularly suitable for large-scale optimization problems.

[0063] Step 7: Calculate the forecast root mean square error to evaluate the forecast performance of the fitted Prophet model.

[0064] Step 8: Save the fitted Prophet model.

[0065] S130, predicting the gas concentration within a preset time period in the future according to the fitted Prophet model, and issuing an alarm for abnormal values.

[0066] Step 9: Load the saved fitted Prophet model to predict the gas concentration in the short term in the future. If the gas concentration is abnormal, an alarm will be issued for the abnormal value. If the predicted gas concentration exceeds the limit, an advance prediction of the gas limit will be issued. The coal mining speed can be adjusted in advance to prevent the gas limit and the resulting shutdown losses.

[0067] Figure 2 A schematic diagram of a gas concentration historical time series data provided by an embodiment of the present invention, such as Figure 2 As shown in the figure, according to the analysis of the historical time series data of gas concentration, the distribution amplitude of gas concentration data varies greatly. When the feature difference is obvious and the gap is large, compared with the traditional logistic function, which is insensitive to changes near the two ends of the interval, the Tanh function in the embodiment of the present invention can better retain these differences. Through this improvement, the improved Prophet model can better simulate the trend distribution of gas concentration time series data.

[0068] like Figure 2 The gas concentration historical time series data distribution shown in the figure shows that the gas concentration has the phenomenon of instantaneous time variation. The Fourier series usually represents a signal component throughout the entire time domain. Even for instantaneous time-varying signals, their characteristics are decomposed into the entire time period to describe. Therefore, the Fourier series is not suitable for fitting the signal distribution with local characteristics such as instantaneous time variation. However, wavelet transform is very useful in analyzing instantaneous time-varying signals. It can effectively extract information from signals and perform multi-scale refinement analysis on functions or signals through operations such as scaling and translation. Unlike the Fourier series whose basis can only be a sine wave, the basis of wavelet transform can be constructed according to specific signal characteristics, thereby more accurately describing the local characteristics of signal distribution.

[0069] In actual production, changes in mine gas concentration not only have holiday effects, but are also affected by special dates unique to the coal mining industry. The present invention also takes into account the impact of these special dates and reconstructs the decomposition terms of the holiday effect terms. Through this improvement, the model can better reflect the characteristics of changes in gas concentration time series data at special nodes.

[0070] Figure 3 A schematic diagram of a gas concentration over-limit early warning system based on big data technology provided by an embodiment of the present invention is shown in FIG. Figure 3 As shown, the system includes:

[0071] The data acquisition module 310 is used to obtain the historical time series data of gas concentration in the mine;

[0072] The model fitting module 320 is used to perform model fitting according to the gas concentration historical time series data and the improved Prophet model to obtain the fitted Prophet model, wherein the improved Prophet model includes a gas concentration growth rate trend model, a seasonal trend model and a holiday effect model, the gas concentration growth rate trend model is obtained based on the tanh function, the seasonal trend model is a wavelet expansion of the seasonal trend of the gas concentration, and the holiday effect model includes the influence of statutory holidays and preset specific holidays on the gas concentration;

[0073] The prediction and alarm module 330 is used to predict the gas concentration within a preset time period in the future according to the fitted Prophet model and issue an alarm for abnormal values.

[0074] This embodiment is a system embodiment corresponding to the above method, and its specific implementation process is the same as that of the above method embodiment. For details, please refer to the above method embodiment, and this system embodiment does not make specific limitations on this.

[0075] Each module in the above-mentioned gas concentration over-limit warning system based on big data technology can be implemented in whole or in part through software, hardware and their combination. The above-mentioned modules can be embedded in or independent of the processor in the intelligent analysis device in the form of hardware, or can be stored in the memory in the intelligent analysis device in the form of software, so that the processor can call and execute the corresponding operations of each of the above modules.

[0076] Figure 4 A schematic diagram of the structure of an intelligent analysis device provided in an embodiment of the present invention. The intelligent analysis device may be a server, and its internal structure diagram may be as follows: Figure 4 As shown. The intelligent analysis device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the intelligent analysis device is used to provide computing and control capabilities. The memory of the intelligent analysis device includes a computer storage medium and an internal memory. The computer storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the computer storage medium. The database of the intelligent analysis device is used to store data generated or obtained during the execution of the gas concentration over-limit early warning method based on big data technology. The network interface of the intelligent analysis device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a gas concentration over-limit early warning method based on big data technology is implemented.

[0077] In one embodiment, an intelligent analysis device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the steps of the gas concentration over-limit early warning method based on big data technology in the above embodiment are implemented. Alternatively, when the processor executes the computer program, the functions of each module / unit in the embodiment of the gas concentration over-limit early warning system based on big data technology are implemented.

[0078] In one embodiment, a computer storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the gas concentration over-limit early warning method based on big data technology in the above embodiment are implemented. Alternatively, when the computer program is executed by a processor, the functions of each module / unit in the above embodiment of the gas concentration over-limit early warning system based on big data technology are implemented.

[0079] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0080] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0081] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A gas concentration over-limit early warning method based on big data technology, characterized in that: include: Obtain historical time series data of gas concentration in mines; According to the gas concentration historical time series data and the improved Prophet model, model fitting is performed to obtain the fitted Prophet model, wherein the improved Prophet model includes a gas concentration growth rate trend model, a seasonal trend model and a holiday effect model, the gas concentration growth rate trend model is obtained based on the tanh function, the seasonal trend model is a wavelet expansion of the seasonal trend of gas concentration, and the holiday effect model includes the influence of statutory holidays and preset specific holidays on gas concentration; According to the fitted Prophet model, the gas concentration in a preset time period in the future is predicted, and an alarm is issued for abnormal values; The calculation formula of the gas concentration growth rate trend model is as follows: ； in, represents the gas concentration growth rate trend model, t represents time, represents the growth rate of gas concentration, represents the offset parameter, Indicates the upper limit value of gas concentration; The calculation formula of the seasonal trend model is as follows: ； in, represents the seasonal trend model, represents the scale parameter, represents the translation parameter, represents the scaling function, represents the wavelet function, represents the approximate coefficient, represents the detail coefficient, t represents the time; The calculation formula of the holiday effect model is as follows: ； ； ； in, represents the holiday effect model, represents the indicator matrix, represents the i-th holiday, L represents the total number of statutory holidays and preset specific holidays, represents the impact factor matrix, represents the impact factor of the i-th holiday, t represents the time, and i is a positive integer; The step of performing model fitting based on the gas concentration historical time series data and the improved Prophet model to obtain the fitted Prophet model includes: Set the change point parameters of the improved Prophet model, the parameters for adjusting seasonal trends, and the holiday effect intensity parameters; The L-BFGS algorithm is used to solve the Prophet model parameters, determine the fitted Prophet model, and evaluate the prediction performance of the fitted Prophet model based on the prediction root mean square error; The step of obtaining historical time series data of gas concentration in the mine includes: Collect historical data of gas sensors that need to be predicted in time series, clean and process the historical data, and convert the historical data into an input format required by the Prophet model; Obtain historical time series data of gas concentration in the mine.

2. An intelligent analysis device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the gas concentration over-limit early warning method based on big data technology as claimed in claim 1 are implemented.

Images