Multi-factor time sequence prediction method, device and equipment for coal mine gas emission quantity

Abstract

The embodiment of the invention provides a coal mine gas emission quantity multi-factor time sequence prediction method, device and equipment, and the method comprises the steps: calculating the feature importance degree of each initial gas emission quantity influence factor based on a preset recursive feature elimination method-based model, and obtaining a feature importance degree sorting result; by adopting a recursive feature elimination method, based on the feature importance ranking result, eliminating the influence factor with the lowest feature importance from the plurality of initial gas emission quantity influence factors to obtain residual gas emission quantity influence factors, and repeatedly eliminating the influence factor with the lowest feature importance to obtain a target gas emission quantity influence factor; calculating feature comprehensive scores of the target gas emission quantity influence factors, comparing the feature comprehensive scores of the target gas emission quantity influence factors, and determining the feature combination with the highest feature comprehensive score as a prediction feature combination; and on the basis of a pre-constructed multi-factor gas emission quantity time sequence prediction model, predicting the coal mine gas water inflow to obtain a gas emission quantity prediction result.

Description

Technical Field

[0001] The present invention relates to the technical field of coal mine gas control, and in particular to, but not limited to, a method, device and equipment for predicting the multi-factor time series of coal mine gas emission. Background Art

[0002] Gas, as a by-product of coal, is also a major source of danger in coal mine production. Gas disasters are extremely destructive, not only threatening the lives of underground workers, but also bringing instability to society. The primary goal of gas control is to control the gas concentration within a safe range and take corresponding extraction measures according to the real-time gas concentration to achieve the purpose of safe production. In recent years, with the substantial increase in my country's coal demand, the intensity and scale of mine mining have continued to expand. Higher requirements are placed on the accuracy and timeliness of gas emission prediction. Correctly grasping the law of underground gas emission changes over time and predicting the gas emission of the working face in advance are of great reference significance for gas extraction and other work.

[0003] With the development of computer technology, grey system theory, multivariate linear regression technology, support vector machine technology, artificial neural network technology, machine learning, deep learning and other methods have been widely used in the prediction of gas emission. In related technologies, a comprehensive analysis method based on empirical mode decomposition and ARMA time series is proposed. The original gas emission time series is decomposed and predicted, and the changing trend of gas emission is obtained, but it does not take into account that gas emission is affected by multiple factors. In related technologies, it is proposed to construct a multi-factor data set of gas emission, and use feature selection to achieve dimensionality reduction of influencing factors, which reduces the difficulty of model training and improves the model prediction accuracy, but does not take into account that gas emission changes over time.

[0004] However, the existing gas outflow prediction methods still have the following shortcomings: (1) Some prediction algorithms select a single influencing factor, ignoring the fact that gas outflow is affected by many factors and the influencing factors are highly correlated, resulting in low prediction accuracy; (2) Some prediction algorithms do not take into account that gas outflow changes over time, ignoring the fact that gas outflow in the future is affected by past moments, and the constructed prediction models have problems such as data leakage. Summary of the invention

[0005] In order to solve the above technical problems existing in the prior art, the present invention provides a method, device and equipment for multi-factor time series prediction of coal mine gas emission.

[0006] The technical method of the embodiment of the present invention is implemented as follows:

[0007] In a first aspect, an embodiment of the present invention provides a method for multi-factor time series prediction of coal mine gas emission, the method comprising:

[0008] Obtain multiple influencing factors of initial gas emission with time series characteristics;

[0009] Based on a preset recursive feature elimination method model, the characteristic importance of each of the initial gas emission influencing factors is calculated to obtain a characteristic importance ranking result of each of the initial gas emission influencing factors;

[0010] A recursive feature elimination method is adopted. Based on the feature importance ranking result, the influencing factors with the lowest feature importance are eliminated from the multiple initial gas emission influencing factors to obtain the remaining gas emission influencing factors. The feature importance of the remaining gas emission influencing factors is recalculated and sorted. The influencing factors with the lowest feature importance are eliminated again from the remaining gas emission influencing factors. The influencing factors with the lowest feature importance are repeatedly eliminated to obtain the target gas emission influencing factors. The number of features of the target gas emission influencing factors meets the preset feature quantity threshold.

[0011] Calculating the feature comprehensive scores of the factors affecting the target gas emission volume, comparing the feature comprehensive scores of each of the factors affecting the target gas emission volume, and determining the feature combination with the highest feature comprehensive score as the prediction feature combination;

[0012] Based on the prediction feature combination and the pre-constructed multi-factor gas emission time series prediction model, the coal mine gas emission is predicted to obtain a gas emission prediction result.

[0013] In a second aspect, an embodiment of the present invention provides a multi-factor time series prediction device for coal mine gas emission, the device comprising:

[0014] An acquisition module, used for obtaining a plurality of influencing factors of initial gas emission with time series characteristics;

[0015] A calculation module, used for calculating the characteristic importance of each of the initial gas emission influencing factors based on a preset recursive feature elimination method model, and obtaining a characteristic importance ranking result of each of the initial gas emission influencing factors;

[0016] A selection module is used to adopt a recursive feature elimination method, based on the feature importance ranking result, eliminate the influencing factor with the lowest feature importance from the multiple initial gas emission influencing factors to obtain the remaining gas emission influencing factors, recalculate the feature importance of the remaining gas emission influencing factors and sort them, eliminate the influencing factor with the lowest feature importance from the remaining gas emission influencing factors again, repeatedly eliminate the influencing factor with the lowest feature importance, and obtain the target gas emission influencing factor; the number of features of the target gas emission influencing factor meets the preset feature number threshold;

[0017] The calculation module is further used to calculate the feature comprehensive score of the target gas emission volume influencing factors, compare the feature comprehensive score of each target gas emission volume influencing factor, and determine the feature combination with the highest feature comprehensive score as the prediction feature combination;

[0018] The prediction module is used to predict the coal mine gas inrush volume based on the prediction feature combination and the pre-built multi-factor gas outburst time series prediction model to obtain the gas outburst volume prediction result.

[0019] In some embodiments, the multiple factors affecting the gas outburst volume with time series characteristics include at least: average daily advance, return air channel air supply volume, upper corner pipe extraction concentration, upper corner buried pipe extraction concentration, coal seam extraction concentration, high-position extraction concentration, upper corner gas concentration, return air flow gas concentration, working face gas concentration, and average mine pressure.

[0020] In some embodiments, the pre-constructed multi-factor gas emission time series prediction model is trained through the following steps: based on the multiple initial gas emission influencing factors with time series characteristics, a gas emission prediction index system is established; based on the gas emission prediction index system, the original data set of each initial gas emission influencing factor is obtained; the data samples of the target gas emission influencing factors are divided into a training set and a test set; a long short-term memory network model is constructed as a basic model for gas emission prediction, and the basic model is adjusted to obtain an optimized multi-factor long short-term memory network model; the training set is input into the optimized multi-factor long short-term memory network model for training to obtain a trained multi-factor gas emission time series prediction model, and the time series prediction model is evaluated through multiple evaluation indicators; the test set is input into the trained multi-factor gas emission time series prediction model for verification to obtain the pre-constructed multi-factor gas emission time series prediction model.

[0021] In some embodiments, the multiple evaluation indicators include at least: goodness of fit R 2 , mean absolute error MAE, root mean square error RMSE; where R 2 It is used to measure the degree of fit between the predicted value and the true value. The closer its value is to 1, the better the fit effect is. MAE and RMSE measure the prediction accuracy from the error size and deviation degree respectively. The smaller its value is, the higher the prediction accuracy is. It is calculated by the following formula:

[0022]

[0023]

[0024]

[0025] In a third aspect, an embodiment of the present invention provides an electronic device, comprising: a memory for storing executable instructions; and a processor for implementing the above-mentioned coal mine gas emission multi-factor time series prediction method when executing the executable instructions stored in the memory.

[0026] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium storing executable instructions for causing a processor to execute the executable instructions to implement the above-mentioned method for multi-factor time series prediction of coal mine gas emission.

[0027] The multi-factor time series prediction method, device and equipment for coal mine gas emission provided by the embodiments of the present invention, on the one hand, by analyzing the relevant factors of coal mine gas emission, and according to the time series characteristics of gas emission, multiple factors are selected to construct a gas emission time series prediction index system, which lays a foundation for the multi-factor time series prediction of coal mine gas emission; on the other hand, in view of the coupling and nonlinear characteristics among multiple influencing factors of gas emission, the recursive feature elimination method is used to select the features of the influencing factors, thereby further simplifying the data, reducing the redundant information in the original data, and improving the model training efficiency; thirdly, in view of the characteristic that gas emission has time series characteristics, the LSTM model is selected as the basic model for multi-factor time series prediction of gas emission, which can learn the change law of gas emission over time from historical gas emission data, so as to make full use of existing data, improve the model interpretability and model prediction accuracy, and thus effectively improve the prediction precision and accuracy of coal mine gas emission. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a structural schematic diagram of a multi-factor time series prediction system for coal mine gas emission provided by an embodiment of the present invention;

[0029] Figure 2 It is a flow chart of a method for multi-factor time series prediction of coal mine gas emission provided by an embodiment of the present invention;

[0030] Figure 3 It is a flow chart of another method for multi-factor time series prediction of coal mine gas emission provided by an embodiment of the present invention;

[0031] Figure 4 It is a gas emission prediction index system diagram provided by an embodiment of the present invention;

[0032] Figure 5 is a schematic diagram showing how the feature combination score of a test set provided by an embodiment of the present invention changes with the number of features;

[0033] Figure 6 is a gas emission prediction result diagram of a test set provided by an embodiment of the present invention;

[0034] Figure 7 is a gas emission prediction error result diagram of the test set provided by an embodiment of the present invention;

[0035] Figure 8 It is a schematic diagram of the composition structure of a multi-factor time series prediction device for coal mine gas emission provided by an embodiment of the present invention;

[0036] Fig. 9 The present invention is a schematic diagram of the composition structure of a multi-factor time series prediction device for coal mine gas emission provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present invention. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present invention.

[0038] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict. Unless otherwise defined, all technical and scientific terms used in the embodiments of the present invention have the same meaning as those commonly understood by those skilled in the art to which the embodiments of the present invention pertain. The terms used in the embodiments of the present invention are only for the purpose of describing the embodiments of the present invention and are not intended to limit the present invention.

[0039] The following describes an exemplary application of a multi-factor time series prediction device for coal mine gas emission according to an embodiment of the present invention. The multi-factor time series prediction device for coal mine gas emission provided by an embodiment of the present invention can be implemented as a terminal or a server. In one implementation, the multi-factor time series prediction device for coal mine gas emission provided by an embodiment of the present invention can be implemented as various types of terminals such as laptop computers, tablet computers, desktop computers, and mobile devices; in another implementation, the multi-factor time series prediction device for coal mine gas emission provided by an embodiment of the present invention can also be implemented as a server, wherein the server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content distribution networks (CDN, Content Delivery Network), and big data and artificial intelligence platforms. The terminal and the server can be directly or indirectly connected by wired or wireless communication, which is not limited in the embodiment of the present invention. Below, an exemplary application of the multi-factor time series prediction device for coal mine gas emission when it is implemented as a server will be described.

[0040] See also Figure 1 , Figure 1 It is a structural schematic diagram of a coal mine gas emission multi-factor time series prediction system 10 provided by an embodiment of the present invention. In order to realize the multi-factor time series prediction of coal mine gas emission, an embodiment of the present invention can provide a coal mine gas emission multi-factor time series prediction platform, and the coal mine gas emission multi-factor time series prediction platform can be implemented as a coal mine gas emission multi-factor time series prediction application. The coal mine gas emission multi-factor time series prediction system 10 provided by an embodiment of the present invention includes a terminal 110, a network 120 and a server 130, wherein the server 130 is a server of the coal mine gas emission multi-factor time series prediction application. The server 130 can constitute a coal mine gas emission multi-factor time series prediction device of an embodiment of the present invention. The terminal 110 is connected to the server 130 via the network 120, and the network 120 can be a wide area network or a local area network, or a combination of the two.

[0041] In some embodiments, please refer to Figure 1When performing multi-factor time series prediction of coal mine gas emission, the terminal 110 sends the gas emission prediction task to the server 130 through the network 120. The server 130 responds to the gas emission prediction task initiated by the terminal 110 and obtains multiple initial gas emission influencing factors with time series characteristics; based on the preset recursive feature elimination method model, the feature importance of each initial gas emission influencing factor is calculated to obtain the feature importance ranking result of each initial gas emission influencing factor; the recursive feature elimination method is used to eliminate the influencing factor with the lowest feature importance from the multiple initial gas emission influencing factors based on the feature importance ranking result, to obtain the remaining gas emission influencing factors, and recalculate the feature importance of each initial gas emission influencing factor. Calculate the feature importance of the remaining gas emission influencing factors and sort them, remove the influencing factors with the lowest feature importance from the remaining gas emission influencing factors again, and repeatedly remove the influencing factors with the lowest feature importance to obtain the target gas emission influencing factors; the number of features of the target gas emission influencing factors meets the preset feature quantity threshold; calculate the feature comprehensive score of the target gas emission influencing factors, compare the feature comprehensive score of each target gas emission influencing factor, and determine the feature combination with the highest feature comprehensive score as the predicted feature combination; predict the coal mine gas emission based on the predicted feature combination and the pre-built multi-factor gas emission time series prediction model to obtain the gas emission prediction result. After obtaining the gas emission prediction result, the server 130 sends the gas emission prediction result to the terminal 110 through the network 120.

[0042] The multi-factor time series prediction method for coal mine gas emission provided by the embodiment of the present invention can also be implemented based on a cloud platform and through cloud technology. For example, the server 300 can be a cloud server. The preset recursive feature elimination method base model and the multi-factor gas emission time series prediction model are determined by the cloud server to realize the multi-factor time series prediction of coal mine gas emission.

[0043] It should be noted here that cloud technology refers to a hosting technology that unifies hardware, software, network and other resources in a wide area network or local area network to achieve data computing, storage, processing and sharing. Cloud technology is a general term for network technology, information technology, integration technology, management platform technology, application technology, etc. based on the cloud computing business model. It can form a resource pool that can be used on demand and is flexible and convenient. Cloud computing technology will become an important support. The backend services of the technical network system require a large amount of computing and storage resources, such as video websites, picture websites and more portal websites. With the high development and application of the Internet industry, in the future, each item may have its own identification mark, and all need to be transmitted to the backend system for logical processing. Data of different levels will be processed separately. All kinds of industry data require strong system backing support, which can only be achieved through cloud computing.

[0044] In some embodiments, there may also be a cloud storage device, in which a preset recursive feature elimination method-based model pre-constructed by the server may be stored in the cloud storage device. Then, when the server receives a gas inrush prediction task sent by a terminal, the server may first accurately and efficiently eliminate the feature importance of the initial gas outrush influencing factors based on the stored preset recursive feature elimination method-based model, thereby obtaining the target gas outrush influencing factors, and then predict the coal mine gas outrush based on the target gas outrush influencing factors and the pre-constructed multi-factor gas outrush time series prediction model.

[0045] The embodiment of the present invention provides a method for predicting the multi-factor time series of coal mine gas emission. Figure 2 , Figure 2 is a flow chart of a multi-factor time series prediction method for coal mine gas emission provided by an embodiment of the present invention, which is combined with Figure 2 The steps shown are explained.

[0046] Step S210, obtaining a plurality of influencing factors of initial gas outflow rate with time series characteristics.

[0047] In some embodiments, multiple influencing factors of the initial gas outburst volume with time series characteristics include average daily advance, return air channel air supply volume, upper corner pipe extraction concentration, upper corner buried pipe extraction concentration, coal seam extraction concentration, high-position extraction concentration, upper corner gas concentration, return air flow gas concentration, working face gas concentration, and average mine pressure.

[0048] Here, the average daily advance refers to the average distance that a coal mine advances within a certain period of time (such as one day, one week, one month, etc.), which is used to measure the progress of the project or the speed of mining. The air volume of the return air lane refers to the air volume allocated in the return air lane to meet the ventilation needs of the mine. The upper corner pipe extraction concentration refers to the gas concentration measured by pipe extraction at the upper corner of the return air of the coal mining face (the triangular area near the upper wall of the return air lane and the edge of the goaf). The upper corner buried pipe extraction concentration refers to the gas concentration measured by buried pipe extraction at the upper corner. The extraction concentration of this coal seam refers to the gas concentration extracted by drilling or other means when mining the coal seam. The high-level extraction concentration refers to the gas concentration extracted by high-level drilling or high extraction lanes in the fracture zone or caving zone above the coal seam. The upper corner gas concentration refers to the gas concentration in the upper corner area. The return air flow gas concentration refers to the gas concentration in the return air flow of the mine. The gas concentration in the working face refers to the gas concentration in the air of the coal mining working face. The average mine pressure refers to the average ground pressure or roof pressure in a certain area in the mine.

[0049] Step S220, based on a preset recursive feature elimination method-based model, calculate the feature importance of each of the initial gas outflow influencing factors, and obtain a feature importance ranking result of each of the initial gas outflow influencing factors.

[0050] In some embodiments, the preset recursive feature elimination method is a feature selection method that recursively reduces the feature set to find the most important features. In each step, the model is trained based on the current feature set and the importance of each feature is evaluated. The least important features are removed and the process is repeated until a predetermined number of features is reached.

[0051] In some embodiments, the base model refers to the underlying machine learning or statistical model used for feature selection and evaluation. This model can be linear regression, decision tree, random forest, support vector machine, etc. During the RFE process, the base model is trained according to the current feature set and outputs the importance score or weight of the feature. In the present invention, the base model can be a random forest model.

[0052] Step S230, using a recursive feature elimination method, based on the feature importance ranking result, eliminate the influencing factors with the lowest feature importance from the multiple initial gas outflow influencing factors to obtain the remaining gas outflow influencing factors, recalculate the feature importance of the remaining gas outflow influencing factors and sort them, eliminate the influencing factors with the lowest feature importance from the remaining gas outflow influencing factors again, repeatedly eliminate the influencing factors with the lowest feature importance to obtain the target gas outflow influencing factors; the number of features of the target gas outflow influencing factors meets the preset feature number threshold.

[0053] Step S240, calculating the feature comprehensive scores of the factors influencing the target gas emission volume, comparing the feature comprehensive scores of each of the factors influencing the target gas emission volume, and determining the feature combination with the highest feature comprehensive score as the prediction feature combination.

[0054] Step S250, based on the prediction feature combination and the pre-constructed multi-factor gas emission time series prediction model, the coal mine gas emission is predicted to obtain a gas emission prediction result.

[0055] The multi-factor time series prediction method, device and equipment for coal mine gas emission provided by the embodiments of the present invention, on the one hand, by analyzing the relevant factors of coal mine gas emission, and according to the time series characteristics of gas emission, multiple factors are selected to construct a gas emission time series prediction index system, which lays a foundation for the multi-factor time series prediction of coal mine gas emission; on the other hand, in view of the coupling and nonlinear characteristics among multiple influencing factors of gas emission, the recursive feature elimination method is used to select the features of the influencing factors, thereby further simplifying the data, reducing the redundant information in the original data, and improving the model training efficiency; thirdly, in view of the characteristic that gas emission has time series characteristics, the LSTM model is selected as the basic model for multi-factor time series prediction of gas emission, which can learn the change law of gas emission over time from historical gas emission data, so as to make full use of existing data, improve the model interpretability and model prediction accuracy, and thus effectively improve the prediction precision and accuracy of coal mine gas emission.

[0056] In some embodiments, the pre-constructed multi-factor gas emission time series prediction model in the above step S250 is trained through the following steps: establishing a gas emission prediction index system based on the multiple initial gas emission influencing factors with time series characteristics; obtaining the original data set of each initial gas emission influencing factor based on the gas emission prediction index system; dividing the data samples of the target gas emission influencing factors into a training set and a test set; constructing a long short-term memory network model as a basic model for gas emission prediction, and adjusting the parameters of the basic model to obtain an optimized multi-factor long short-term memory network model; inputting the training set into the optimized multi-factor long short-term memory network model for training to obtain a trained multi-factor gas emission time series prediction model, and evaluating the time series prediction model through multiple evaluation indicators; inputting the test set into the trained multi-factor gas emission time series prediction model for verification to obtain the pre-constructed multi-factor gas emission time series prediction model.

[0057] In some embodiments, the multiple evaluation indicators in the above model building step include at least: goodness of fit R 2 , mean absolute error MAE, root mean square error RMSE; where R 2 It is used to measure the degree of fit between the predicted value and the true value. The closer its value is to 1, the better the fit effect is. MAE and RMSE measure the prediction accuracy from the error size and deviation degree respectively. The smaller its value is, the higher the prediction accuracy is. It is calculated by the following formula:

[0058]

[0059]

[0060]

[0061] The following describes an exemplary application of an embodiment of the present invention in a practical application scenario.

[0062] This embodiment provides another method for multi-factor time series prediction of coal mine gas emission. Figure 3 As shown, the specific steps are as follows:

[0063] Step 1: Analyze the factors affecting gas emission and establish a gas emission prediction index system.

[0064] In this embodiment, the gas emission influencing factors include gas emission influencing factors with time series characteristics, including average daily advance (F1, m), return air lane air volume (F2, m 3 / min), upper corner pipe extraction concentration (F3, %), upper corner buried pipe extraction concentration (F4, %), extraction concentration of the coal seam (F5, %), high-position extraction concentration (F6, %), upper corner gas concentration (F7, %), return air flow gas concentration (F8, %), working face gas concentration (F9, %), and average mine pressure (F10, kPa).

[0065] In specific implementation, each influencing factor has different effects on the gas emission of the working face, some have a high degree of influence, some have a low degree of influence, and some influencing factors are coupled. Therefore, there are many factors affecting gas emission and the correlation is complex. After carefully analyzing the law of coal mine gas emission, 10 influencing factors with time series characteristics were identified, and a gas emission prediction index system was established, such as Figure 4 shown.

[0066] Step 2: Obtain the original data set of factors affecting gas emission based on the established gas emission prediction index system.

[0067] During the specific implementation, some original sample data of factors affecting gas emission are shown in Table 1. The time interval for collecting each factor is 1 day, including average daily advance (F1), return air lane air volume (F2), upper corner pipe extraction concentration (F3), upper corner buried pipe extraction concentration (F4), coal seam extraction concentration (F5), high-position extraction concentration (F6), upper corner gas concentration (F7), return air flow gas concentration (F8), working face gas concentration (F9), average mine pressure (F10). The prediction target is the absolute gas emission volume Y.

[0068] Table 1 Part of the original sample data

[0069]

[0070] In this embodiment, the original sample data includes 269 groups of samples, and each group of samples includes 10 factors affecting the gas emission.

[0071] Step 3: Use the recursive feature elimination method to select the features of the factors affecting the gas emission and reduce the nonlinearity and coupling between the factors.

[0072] In some embodiments, the above step three may include steps 301 to 305:

[0073] Step 301: Determine a recursive feature elimination base model: a random forest model.

[0074] Step 302: Calculate the feature importance of each factor according to the recursive feature elimination method based model to obtain the feature importance ranking of each factor.

[0075] Step 303: Eliminate the influencing factors with the lowest feature importance, calculate the scores of the remaining feature combinations, and recalculate and sort the feature importances of the remaining factors.

[0076] Step 304, loop step 303 until the number of remaining features is 2.

[0077] Step 305: compare the scores of each feature combination and select the combination with the highest score as the final prediction feature combination.

[0078] In the specific implementation, in order to more comprehensively consider the influencing factors of gas outflow, 10 influencing factors of gas outflow were selected as prediction indicators. Since there are many influencing factors selected, if these 10 influencing factors are directly used as input variables of the prediction model, the complexity of the model will increase, affecting the calculation rate and prediction performance of machine learning. Moreover, the correlation between the influencing factors of gas outflow is complex, with the characteristics of nonlinearity, coupling, and overlap. Therefore, in order to minimize the impact of redundancy between influencing factors, it is necessary to perform feature selection on the original influencing factor indicator system. Feature selection can not only select the optimal combination of prediction influencing factors, reduce the complexity of input data, and improve the model training rate, but also to a certain extent ensure the reliability of the multi-factor time series prediction model of gas outflow. The feature combination score varies with the number of features, such as Figure 5 Therefore, the number of influencing factors selected after the recursive feature elimination method is 4, including: upper corner pipe extraction concentration, coal seam extraction concentration, high-position extraction concentration, and return air flow gas concentration.

[0079] Step 4: Divide the data samples after feature selection into training set and test set.

[0080] In the specific implementation, the data set after feature selection is divided into training set and test set in a ratio of 7:3, with 186 sets of training sets and 83 sets of prediction sets.

[0081] Step 5: Construct an LSTM model as the basic model for gas emission prediction and set the parameters in the model.

[0082] In the specific implementation, the input of the long short-term memory network is the factors affecting the gas emission, including the upper corner pipe extraction concentration, the extraction concentration of the coal seam, the high-level extraction concentration, and the return air flow gas concentration; through forward propagation, the time series information of the gas emission is obtained. The model parameters are set to have 3 hidden layer units, 128 network layers, 0.001 learning rate, 12 batch size, and 200 iterations. Finally, the prediction result of the gas emission is obtained.

[0083] Step 6: Input the training set described in step 4 into the optimized multi-factor LSTM model for training to obtain a trained multi-factor gas emission time series prediction model, and evaluate the model through multiple evaluation indicators.

[0084] Step 7: Input the verification set described in step 4 into the trained gas emission prediction model for testing to obtain the final gas emission prediction model.

[0085] In the specific implementation, the prediction results and error results of the test set were obtained through experimental simulation. The prediction results of gas emission are as follows: Figure 6 As shown, the absolute error of the prediction is Figure 7 As shown in the prediction results, it can be seen that the absolute error of the model prediction is between -1.53 ​​and 0.89, the average absolute error is 0.54, and the goodness of fit R 2 The root mean square error (RMSE) is 0.9897 and 0.2455, indicating that the model has a good fitting result and a small prediction error, which indicates that the prediction of the model is relatively stable and can provide a certain theoretical basis for gas extraction and other work.

[0086] In some embodiments, the original sample data in the above step 2 includes multiple groups of time series samples, and each group of samples includes 10 factors affecting the gas outflow rate.

[0087] In some embodiments, the model parameters in the above step five mainly include: the number of hidden layer units, the number of network layers, the learning rate, the batch size, and the number of iterations.

[0088] In some embodiments, the model evaluation index in step 6 mainly includes: goodness of fit R 2 , mean absolute error MAE, root mean square error RMSE. Among them, R 2 It is used to measure the degree of fit between the predicted value and the true value. The closer its value is to 1, the better the fit effect is. MAE and RMSE measure the prediction accuracy from the error size and deviation degree respectively. The smaller its value is, the higher the prediction accuracy is. The calculation formula is:

[0089]

[0090]

[0091]

[0092] Figure 8 Schematic diagram of the composition structure of the multi-factor time series prediction device for coal mine gas emission provided by an embodiment of the present invention. Figure 8 As shown, the coal mine gas emission multi-factor time series prediction device 800 includes: an acquisition module 801, which is used to obtain multiple initial gas emission influencing factors with time series characteristics; a calculation module 802, which is used to calculate the feature importance of each of the initial gas emission influencing factors based on a preset recursive feature elimination method base model, and obtain the feature importance ranking result of each of the initial gas emission influencing factors; a selection module 803, which is used to use the recursive feature elimination method, based on the feature importance ranking result, eliminate the influencing factor with the lowest feature importance from the multiple initial gas emission influencing factors, obtain the remaining gas emission influencing factors, recalculate the feature importance of the remaining gas emission influencing factors and sort them, and select the remaining gas emission influencing factors. The influencing factors with the lowest feature importance are eliminated again from the remaining gas emission influencing factors, and the influencing factors with the lowest feature importance are repeatedly eliminated to obtain the target gas emission influencing factors; the number of features of the target gas emission influencing factors meets the preset feature quantity threshold; the calculation module 802 is also used to calculate the feature comprehensive score of the target gas emission influencing factors, compare the feature comprehensive score of each target gas emission influencing factor, and determine the feature combination with the highest feature comprehensive score as the predicted feature combination; the prediction module 804 is used to predict the coal mine gas influencing volume based on the predicted feature combination and a pre-constructed multi-factor gas emission time series prediction model to obtain a gas emission prediction result.

[0093] In some embodiments, the multiple factors affecting the gas outburst volume with time series characteristics include at least: average daily advance, return air channel air supply volume, upper corner pipe extraction concentration, upper corner buried pipe extraction concentration, coal seam extraction concentration, high-position extraction concentration, upper corner gas concentration, return air flow gas concentration, working face gas concentration, and average mine pressure.

[0094] In some embodiments, the pre-constructed multi-factor gas emission time series prediction model is trained through the following steps: based on the multiple initial gas emission influencing factors with time series characteristics, a gas emission prediction index system is established; based on the gas emission prediction index system, the original data set of each initial gas emission influencing factor is obtained; the data samples of the target gas emission influencing factors are divided into a training set and a test set; a long short-term memory network model is constructed as a basic model for gas emission prediction, and the basic model is adjusted to obtain an optimized multi-factor long short-term memory network model; the training set is input into the optimized multi-factor long short-term memory network model for training to obtain a trained multi-factor gas emission time series prediction model, and the time series prediction model is evaluated through multiple evaluation indicators; the test set is input into the trained multi-factor gas emission time series prediction model for verification to obtain the pre-constructed multi-factor gas emission time series prediction model.

[0095] In some embodiments, the multiple evaluation indicators include at least: goodness of fit R 2 , mean absolute error MAE, root mean square error RMSE; where R 2 It is used to measure the degree of fit between the predicted value and the true value. The closer its value is to 1, the better the fit effect is. MAE and RMSE measure the prediction accuracy from the error size and deviation degree respectively. The smaller its value is, the higher the prediction accuracy is. It is calculated by the following formula:

[0096]

[0097]

[0098]

[0099] It should be noted that the description of the device of the embodiment of the present invention is similar to the description of the above method embodiment, and has similar beneficial effects as the same method embodiment, so it will not be repeated. For technical details not disclosed in the embodiment of the device, please refer to the description of the method embodiment of the present invention for understanding.

[0100] It should be noted that, in the embodiment of the present invention, if the above-mentioned coal mine gas emission multi-factor time series prediction method is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention can be essentially or partly reflected in the form of a software product, which is stored in a storage medium and includes a number of instructions for a terminal to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a U disk, a mobile hard disk, a read-only memory (ROM), a magnetic disk or an optical disk. In this way, the embodiment of the present invention is not limited to any specific combination of hardware and software.

[0101] Correspondingly, an embodiment of the present invention provides a multi-factor time series prediction device for coal mine gas emission. Fig. 9 Schematic diagram of the composition structure of a multi-factor time series prediction device for coal mine gas emission provided by an embodiment of the present invention. Fig. 9 As shown, the coal mine gas emission multi-factor time series prediction device 900 at least includes: a processor 901 and a computer-readable storage medium 902 configured to store executable instructions, wherein the processor 901 generally controls the overall operation of the coal mine gas emission multi-factor time series prediction device 900. The computer-readable storage medium 902 is configured to store instructions and applications executable by the processor 901, and can also cache data to be processed or processed by each module in the processor 901 and the coal mine gas emission multi-factor time series prediction device 900, which can be implemented through flash memory (FLASH) or random access memory (RAM, Random Access Memory).

[0102] An embodiment of the present invention provides a storage medium storing executable instructions, wherein the executable instructions are stored. When the executable instructions are executed by a processor, the processor will be caused to execute the method provided by the embodiment of the present invention, for example, Figure 2 The method shown.

[0103] In some embodiments, the storage medium can be a computer-readable storage medium, for example, a ferroelectric random access memory (FRAM), a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a flash memory, a magnetic surface memory, an optical disk, or a compact disk read-only memory (CD-ROM), etc.; it can also be various devices including one or any combination of the above memories.

[0104] In some embodiments, executable instructions may be in the form of a program, software, software module, script or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including as a stand-alone program or as a module, component, subroutine or other unit suitable for use in a computing environment.

[0105] As an example, executable instructions may, but need not necessarily, correspond to a file in a file system, may be stored as part of a file storing other programs or data, for example, in one or more scripts in a Hyper Text Markup Language (HTML) document, in a single file dedicated to the program in question, or in multiple coordinated files (e.g., files storing one or more modules, subroutines, or code portions). As an example, executable instructions may be deployed to be executed on one electronic device, or on multiple electronic devices located at one location, or on multiple electronic devices distributed at multiple locations and interconnected by a communication network.

[0106] The above description is only an embodiment of the present invention and is not intended to limit the protection scope of the present invention. Any modification, equivalent replacement and improvement made within the spirit and scope of the present invention are included in the protection scope of the present invention.

[0107] It should be understood that "one embodiment" or "an embodiment" mentioned throughout the specification means that specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present invention. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that in various embodiments of the present invention, the size of the serial number of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention. The serial numbers of the above-mentioned embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments.

[0108] It should be noted that, in this article, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method or device. In the absence of further restrictions, an element defined by the statement "comprises one..." does not exclude the presence of other identical elements in the process, method, article or device including the element. In several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored, or not executed.

[0109] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A multi-factor time series prediction method for coal mine gas emission, characterized in that: The method comprises: Obtain multiple influencing factors of initial gas emission with time series characteristics; Based on a preset recursive feature elimination method model, the characteristic importance of each of the initial gas emission influencing factors is calculated to obtain a characteristic importance ranking result of each of the initial gas emission influencing factors; A recursive feature elimination method is adopted. Based on the feature importance ranking result, the influencing factors with the lowest feature importance are eliminated from the multiple initial gas emission influencing factors to obtain the remaining gas emission influencing factors. The feature importance of the remaining gas emission influencing factors is recalculated and sorted. The influencing factors with the lowest feature importance are eliminated again from the remaining gas emission influencing factors. The influencing factors with the lowest feature importance are repeatedly eliminated to obtain the target gas emission influencing factors. The number of features of the target gas emission influencing factors meets the preset feature quantity threshold. Calculating the feature comprehensive scores of the factors affecting the target gas emission volume, comparing the feature comprehensive scores of each of the factors affecting the target gas emission volume, and determining the feature combination with the highest feature comprehensive score as the prediction feature combination; Based on the prediction feature combination and the pre-constructed multi-factor gas emission time series prediction model, the coal mine gas emission is predicted to obtain a gas emission prediction result.

2. The method according to claim 1, characterized in that: The multiple factors affecting the gas outburst volume with time series characteristics include at least: average daily advance, return air channel air supply volume, upper corner pipe extraction concentration, upper corner buried pipe extraction concentration, extraction concentration of the coal seam, high-position extraction concentration, upper corner gas concentration, return air flow gas concentration, working face gas concentration, and average mine pressure.

3. The method according to claim 1, characterized in that The pre-built multi-factor gas emission time series prediction model is trained by the following steps: Based on the multiple initial gas emission influencing factors with time series characteristics, a gas emission prediction index system is established; Based on the gas emission prediction index system, an original data set of each of the initial gas emission influencing factors is obtained; Dividing the data samples of the influencing factors of the target gas emission volume into a training set and a test set; Constructing a long short-term memory network model as a basic model for predicting gas emission, and adjusting the parameters of the basic model to obtain an optimized multi-factor long short-term memory network model; Inputting the training set into the optimized multi-factor long short-term memory network model for training to obtain a trained multi-factor gas emission time series prediction model, and evaluating the time series prediction model through multiple evaluation indicators; The test set is input into the trained multi-factor gas emission time series prediction model for verification to obtain the pre-constructed multi-factor gas emission time series prediction model.

4. The method according to claim 3, characterized in that The multiple evaluation indicators include at least: goodness of fit R 2 , mean absolute error MAE, root mean square error RMSE; where R 2 It is used to measure the degree of fit between the predicted value and the true value. The closer its value is to 1, the better the fit effect is. MAE and RMSE measure the prediction accuracy from the error size and deviation degree respectively. The smaller its value is, the higher the prediction accuracy is. It is calculated by the following formula:

5. A multi-factor time series prediction device for coal mine gas emission, characterized in that: The device comprises: An acquisition module, used for obtaining a plurality of influencing factors of initial gas emission with time series characteristics; A calculation module, used for calculating the characteristic importance of each of the initial gas emission influencing factors based on a preset recursive feature elimination method model, and obtaining a characteristic importance ranking result of each of the initial gas emission influencing factors; A selection module is used to adopt a recursive feature elimination method, based on the feature importance ranking result, eliminate the influencing factor with the lowest feature importance from the multiple initial gas emission influencing factors to obtain the remaining gas emission influencing factors, recalculate the feature importance of the remaining gas emission influencing factors and sort them, eliminate the influencing factor with the lowest feature importance from the remaining gas emission influencing factors again, repeatedly eliminate the influencing factor with the lowest feature importance, and obtain the target gas emission influencing factor; the number of features of the target gas emission influencing factor meets the preset feature number threshold; The calculation module is further used to calculate the feature comprehensive score of the target gas emission volume influencing factors, compare the feature comprehensive score of each target gas emission volume influencing factor, and determine the feature combination with the highest feature comprehensive score as the prediction feature combination; The prediction module is used to predict the coal mine gas inrush volume based on the prediction feature combination and the pre-built multi-factor gas outburst time series prediction model to obtain the gas outburst volume prediction result.

6. An electronic device, characterized in that: include: A memory for storing executable instructions; A processor is used to implement the multi-factor time series prediction method for coal mine gas emission as described in any one of claims 1 to 4 when executing the executable instructions stored in the memory.

Images