Working face gas concentration prediction model optimization method based on multi-source data

Through the preprocessing and model construction of multi-source data, the problem of gas concentration prediction caused by a single factor in the existing technology is solved, more accurate and comprehensive gas concentration prediction is achieved, and support for coal mine safety production is enhanced.

CN119004054APending Publication Date: 2024-11-22TIANDI CHANGZHOU AUTOMATION +1
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202411062461.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-05
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

The existing working face gas concentration prediction method considers a single factor, which leads to poor early warning effect and is difficult to capture the gradual process and potential risks of gas gushing out, especially in complex and changeable mine wind flow environments.

Method used

Using the preferred method of working face gas concentration prediction model based on multi-source data, the historical data of production equipment and environmental parameters are collected, the coefficient of variation is calculated, and the data is preprocessed and standardized, and the linear regression, logistic regression and support vector machine algorithm are constructed. For the model, select the model with a larger F1 score as the prediction model.

Benefits of technology

It achieves more accurately capturing the changing trends and potential risks of gas surges, improves the accuracy and comprehensiveness of predictions, provides a more comprehensive data analysis basis, and enhances the prediction capabilities and stability of the model.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119004054A_ABST
    Figure CN119004054A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of coal mine safety, and particularly relates to a working face gas concentration prediction model optimization method based on multi-source data, which comprises the following steps: acquiring historical data of working face production equipment working parameters and working face environment parameters in the same time period; after the variation parameters are removed, the remaining data are preprocessed; screening the parameter data at the same moment by taking the parameter with the maximum time interval of each parameter data as a reference to obtain the parameter data with the same quantity; performing standardization processing on the screened data to obtain a data sample set; a training set and a verification set are divided, a linear regression algorithm, a logistic regression algorithm and a support vector machine algorithm are used for training and verification work respectively, and the model with higher accuracy is taken as a final model. According to the method, multiple equipment factors are comprehensively considered, fusion processing of multi-source data is achieved, the change trend and potential risks of gas emission are accurately captured by constructing efficient data processing, and powerful technical support is provided for safe production of a coal mine.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of coal mine safety, and in particular relates to a method for optimizing a working face gas concentration prediction model based on multi-source data. Background Art

[0002] In the process of coal mine production, the prevention and control of gas disasters is the core link to ensure safe production. With the increasing complexity and dynamics of the coal mining environment, unprecedented challenges have been posed to the comprehensiveness, timeliness and accuracy of information acquisition of gas disaster early warning methods. Traditional gas disaster early warning methods, such as threshold alarms, can quickly trigger alarms when the gas concentration exceeds the preset threshold, but their single-point, single-parameter threshold judgment mechanism is difficult to capture the gradual process and potential risks of gas outbursts, and the early warning effect is limited. At the same time, although the analytical early warning method based on gas concentration data can predict gas concentration anomalies through time series analysis to a certain extent, most of the existing methods are limited to shallow processing of a single data source, and fail to fully explore and utilize multivariate and heterogeneous production process data. This limitation makes it difficult for the predictability and accuracy of gas concentration early warning to meet the needs of coal mine safety production, especially when faced with the complex and changeable situation of gas outbursts in the mine airflow, it seems to be powerless.

[0003] As an important area of ​​coal mine production, the gas emission of fully mechanized mining face is affected by many factors, including production activities such as coal cutting, transportation, and support, as well as dynamic changes in environmental factors such as wind speed, temperature, and gas concentration in goaf. The interaction of these factors makes the prediction of gas emission extremely complicated. Therefore, the traditional gas concentration early warning method is difficult to adapt to these changes in real time, resulting in a significant reduction in the early warning effect. Summary of the invention

[0004] The technical problem to be solved by the present invention is that the existing working face gas concentration prediction method considers a single factor and the gas concentration early warning effect is poor.

[0005] To this end, the present invention provides a method for optimizing a working face gas concentration prediction model based on multi-source data.

[0006] The technical solution adopted by the present invention to solve its technical problem is:

[0007] A method for optimizing a working face gas concentration prediction model based on multi-source data comprises the following steps:

[0008] Step 1: Collect historical data of working parameters of production equipment and working environment parameters of the working face within the same time period;

[0009] Step 2: Calculate the coefficient of variation of each parameter's historical data, remove the parameters with a coefficient of variation less than 10%, and then pre-process the remaining parameters' data;

[0010] Step 3: Based on the time intervals of the data of each parameter, the parameter with the largest time interval is used as the benchmark, and the data of the remaining parameters are screened at the same time to obtain the same amount of data of each parameter;

[0011] Step 4: Standardize the screened data to obtain a data sample set;

[0012] Step 5: Divide the data sample set into a training set and a validation set, and use the linear regression algorithm, logistic regression algorithm, and support vector machine algorithm to carry out model training and validation respectively;

[0013] Step 6: Evaluate the accuracy of each algorithm model in step 5, and take the one with the higher accuracy as the final model.

[0014] Furthermore, in step one, the working parameters of the working face production equipment include the traction speed of the working face coal mining machine, the chain speed of the scraper conveyor, the current of the transfer machine, the rotation speed of the crusher, the column pressure of the hydraulic support, the working face temperature, the goaf temperature, the working face wind speed, the goaf gas methane concentration, and the working face methane concentration; the working face environmental parameter data include the working face temperature, the goaf temperature, the working face wind speed, the goaf gas concentration, and the working face gas concentration.

[0015] Furthermore, in step 2, the preprocessing of all data of each parameter includes missing value processing, outlier processing and normalization processing.

[0016] Furthermore, in step three, if a parameter has no reference time data, the data closest to the reference time is used instead.

[0017] Furthermore, in step 4, the standardization process is as follows: first, the average value a of the normalized data of each parameter is calculated, and then the standard deviation b of the normalized data of each parameter is calculated. For each normalized data x* of each parameter, the formula c=(x*-a) / b is used to convert the normalized data x* of each parameter into a new data set c of each parameter.

[0018] Furthermore, in step five, the ratio of the training set to the validation set is N:10-N.

[0019] Further, in step 5, the linear regression algorithm model is y=β0+β1x1+β2x2+…+β i x i, where β0 is the intercept term, i is the number of remaining parameters after removing the coefficient of variation less than 10%, β1, β2, ..., β i It is the coefficient of the remaining parameters after eliminating the parameters with coefficient of variation less than 10% in step 2.

[0020] Further, in step 5, the logistic regression algorithm model is y is the predicted methane concentration of the working face, and P(y=1) represents the probability of belonging to the positive class.

[0021] Furthermore, the support vector machine algorithm model is f(x)=sign(β0+β1x1+β2x2+…+β i x i ), f(x) is the predicted methane concentration at the working face.

[0022] According to the method for optimizing the working face gas concentration prediction model based on multi-source data according to claim 1, it is characterized in that, in step six, the F1 scores of the linear regression algorithm model, the logistic regression algorithm model and the support vector machine algorithm model are compared, and the algorithm model with a larger F1 score is selected as the working face gas concentration prediction model. If the accuracy of the linear regression algorithm model, the logistic regression algorithm model and the support vector machine algorithm model is the same, the division ratio of the training set and the validation set is changed to (N-1): (10-N-1), and steps five to six are repeated.

[0023] The beneficial effect of the present invention is that the present invention comprehensively considers various factors, incorporates the working parameters of equipment such as coal mining machines, scraper conveyors, hydraulic supports, and environmental variables such as wind speed, temperature, and gas concentration in goafs into the working face gas outburst prediction index system, and realizes the fusion processing of multi-source data. By building an efficient data processing and analysis model, the method can more accurately capture the changing trends and potential risks of gas outbursts, and provide strong technical support for coal mine safety production.

[0024] Specifically, this application collects historical data on working parameters of production equipment and working environment parameters of the working face, comprehensively considers the impact of multiple factors, especially equipment working parameters, on gas concentration, and improves the accuracy and comprehensiveness of the prediction. Analyzing and predicting changes in gas concentration from multiple angles is more advantageous than the prediction method based on a single data source, and provides a more comprehensive data analysis basis.

[0025] In the data preprocessing stage, the coefficient of variation is calculated to eliminate parameters with less influence, and the remaining parameters are processed for missing values, outliers and normalization, which improves the data quality and model stability. By screening data based on the parameter with the largest time interval, the problem of time alignment of data of different parameters is solved, and the representativeness of model training data is improved, thereby effectively avoiding the influence of noise and outliers in the data on the prediction results. It is more scientific than simply using data directly, solves the problem of lack of comprehensive analysis, and enhances the prediction ability of the model.

[0026] By using F1 score and accuracy as model performance metrics and selecting the model with higher accuracy from the models with the highest F1 score, this optimization mechanism is more comprehensive than single indicator evaluation and can balance the accuracy and generalization ability of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0028] Figure 1 It is a schematic diagram of the implementation process of a method for optimizing a working face gas concentration prediction model based on multi-source data in the present invention. DETAILED DESCRIPTION

[0029] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0030] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. In addition, features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, unless otherwise specified, "multiple" means two or more.

[0031] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0032] A method for optimizing a working face gas concentration prediction model based on multi-source data comprises the following steps:

[0033] Step 1: Use a direct-connected storage device or call an API interface to collect historical data on the working parameters of the working face production equipment and the working face environmental parameters. Specifically, a USB flash drive can be used to connect to a computer that stores the working parameters of the working face production equipment and the working face environmental parameters through a USB interface.

[0034] The working parameters of the working face production equipment include but are not limited to the traction speed of the working face coal mining machine, the chain speed of the scraper conveyor, the current of the transfer machine, the rotation speed of the crusher, the column pressure of the hydraulic support, the working face temperature, the goaf temperature, the working face wind speed, the methane concentration of the goaf gas, and the methane concentration of the working face. Each parameter is collected for 100 days;

[0035] The working face environmental parameter data include but are not limited to working face temperature, goaf temperature, working face wind speed, goaf gas (carbon monoxide, carbon dioxide, methane, ethylene, acetylene, ethane, oxygen, nitrogen) concentration, and working face gas (methane, carbon monoxide, carbon dioxide, nitrogen) concentration.

[0036] Step 2: Calculate the coefficient of variation of the historical data of each parameter within 100 days. The coefficient of variation is the ratio of the standard deviation to the mean value. Eliminate the parameters with a coefficient of variation less than 10%, and perform missing value processing and outlier processing on the data of the remaining parameters.

[0037] Use the median of all the data of each parameter to fill the missing values ​​of the parameter; use the mean of all the data of each parameter to replace the outliers of the parameter; and use x*=(x-min) / (max-min) to normalize all the data of each parameter, where max is the maximum value of all the data of the parameter and min is the minimum value of all the data of the parameter. Take a set of data of production equipment working parameters and working surface environment parameters as an example (refer to the table below):

[0038]

[0039]

[0040] Step 3: Based on the time intervals of the data of each parameter, take the parameter with the largest time interval as the benchmark, filter the data of the remaining parameters at the same time, and obtain the same amount of data of each parameter. If a parameter has no benchmark time data, replace it with the data closest to the benchmark time.

[0041] Specifically, in this embodiment, since the methane concentration in the goaf generates one data every 15 minutes, which is the largest time interval for data generation among all parameters, the time when the methane concentration data in the goaf is generated is used as a benchmark to screen the data of other parameters at the same time to obtain the same number of parameter data. If a parameter has no data corresponding to the time when the methane concentration data in the goaf is generated, the data closest to the time when the corresponding methane concentration data in the goaf is used instead.

[0042] Step 4: Standardize the data obtained from each parameter screening. First, calculate the average value a of the normalized data of each parameter, and then calculate the standard deviation b of the normalized data of each parameter. For each parameter normalized data x*, use the formula c=(x*-a) / b to convert each parameter normalized data x* into a new data set c for each parameter.

[0043] Step 5: Divide the new data c into a training set and a validation set, the ratio of the training set to the validation set is (1:9) to (9:1). In this embodiment, the new data c of all parameters is divided into training set data and validation set data in a ratio of 7:3. The training set data is used to train the algorithm model, and the validation set data is used to select the algorithm model.

[0044] Step 6: Use the training set data to train the linear regression algorithm model, logistic regression algorithm model and support vector machine algorithm model. The linear regression algorithm model is y=β0+β1x1+β2x2+…+β i x i , the logistic regression algorithm model is The support vector machine algorithm model is f(x)=sign(β0+β1x1+β2x2+…+β i x i ), y and f(x) are the predicted methane concentrations of the working face, P(y=1) represents the probability of belonging to the positive class, β0 is the intercept term, i is the number of remaining parameters after removing the coefficient of variation less than 10%, β1, β2, ..., β i is the coefficient of the remaining parameters after removing the parameters with coefficient of variation less than 10% in step 2, x1-x i They are the corresponding values ​​of the remaining parameters after eliminating the parameters with coefficient of variation less than 10%.

[0045] In this embodiment, since the number of parameters in the table of step 2 is 9 (i.e., after eliminating the parameters with a coefficient of variation less than 10%, the number of remaining parameters is 9), i=9, β1, β2, ..., β9 are coefficients related to the traction speed of the coal mining machine, the chain speed of the scraper conveyor, the current of the transfer machine, the rotation speed of the crusher, the column pressure of the hydraulic support, the working surface temperature, the goaf temperature, the working surface wind speed, and the methane concentration of the goaf gas, respectively, and x1-x9 represent the values ​​of the traction speed of the coal mining machine, the chain speed of the scraper conveyor, the current of the transfer machine, the rotation speed of the crusher, the column pressure of the hydraulic support, the working surface temperature, the goaf temperature, the working surface wind speed, and the methane concentration of the goaf gas after standardization. The linear regression algorithm model and the support vector machine algorithm model are verified using the validation set data, and the number of true positive examples, true negative examples, false positive examples, and false negative examples is counted.

[0046] Step 7: Calculate the precision and recall of the linear regression algorithm model, logistic regression algorithm model and support vector machine algorithm model respectively. Precision = true positive / (true positive + false positive), recall = true positive / (true positive + false negative). Then calculate the F1 score of the linear regression algorithm model and the support vector machine algorithm model, F1 score = 2*(precision*recall) / (precision+recall). Compare the F1 scores of the linear regression algorithm model, logistic regression algorithm model and support vector machine algorithm model, and select the algorithm model with a larger F1 score as the working face gas concentration prediction model. If the F1 scores of the linear regression algorithm model, logistic regression algorithm model and support vector machine algorithm model are the same, take the one with higher accuracy as the final model. If the accuracy of the linear regression algorithm model, logistic regression algorithm model and support vector machine algorithm model is the same, change the division ratio of the training set and the validation set to 6:4, and repeat steps 5 to 7.

[0047] In summary, this application collects historical data on working parameters of production equipment and working environment parameters of the working face, comprehensively considers the impact of multiple factors, especially equipment working parameters, on gas concentration, and improves the accuracy and comprehensiveness of the prediction. Analyzing and predicting changes in gas concentration from multiple angles is more advantageous than the prediction method based on a single data source, and provides a more comprehensive data analysis basis.

[0048] In the data preprocessing stage, the coefficient of variation is calculated to eliminate parameters with less influence, and the remaining parameters are processed for missing values, outliers and normalization, which improves the data quality and model stability. By screening data based on the parameter with the largest time interval, the problem of time alignment of data of different parameters is solved, and the representativeness of model training data is improved, thereby effectively avoiding the influence of noise and outliers in the data on the prediction results. It is more scientific than simply using data directly, solves the problem of lack of comprehensive analysis, and enhances the prediction ability of the model.

[0049] By using F1 score and accuracy as model performance metrics and selecting the model with higher accuracy from the models with the highest F1 score, this optimization mechanism is more comprehensive than single indicator evaluation and can balance the accuracy and generalization ability of the model.

[0050] Based on the above ideal embodiments of the present invention, the relevant staff can make various changes and modifications without departing from the technical concept of the present invention through the above description. The technical scope of the present invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A method for optimizing a working face gas concentration prediction model based on multi-source data, characterized in that: The following steps are included: Step 1: Collect historical data of working parameters of production equipment and working environment parameters of the working face within the same time period; Step 2: Calculate the coefficient of variation of each parameter's historical data, remove the parameters with a coefficient of variation less than 10%, and then pre-process the remaining parameters' data; Step 3: Based on the time intervals of the data of each parameter, the parameter with the largest time interval is used as the benchmark, and the data of the remaining parameters are screened at the same time to obtain the same amount of data of each parameter; Step 4: Standardize the screened data to obtain a data sample set; Step 5: Divide the data sample set into a training set and a validation set, and use the linear regression algorithm, logistic regression algorithm, and support vector machine algorithm to carry out model training and validation respectively; Step 6: Evaluate the accuracy of each algorithm model in step 5, and take the one with the higher accuracy as the final model. According to the method for optimizing the working face gas concentration prediction model based on multi-source data described in claim 1, it is characterized in that, in step one, the working parameters of the working face production equipment include the traction speed of the working face coal mining machine, the chain speed of the scraper conveyor, the current of the transfer machine, the rotation speed of the crusher, the column pressure of the hydraulic support, the working face temperature, the goaf temperature, the working face wind speed, the methane concentration of the goaf gas, and the methane concentration of the working face; the working face environmental parameter data include the working face temperature, the goaf temperature, the working face wind speed, the goaf gas concentration, and the working face gas concentration.

2. The method for optimizing the working face gas concentration prediction model based on multi-source data according to claim 1 is characterized in that: In step 2, the preprocessing of all data of each parameter includes missing value processing, outlier processing and normalization processing.

3. The method for optimizing a working face gas concentration prediction model based on multi-source data according to claim 1 is characterized in that: In step 3, if there is no reference time data for a parameter, the data closest to the reference time is used instead.

4. The method for optimizing a working face gas concentration prediction model based on multi-source data according to claim 1 is characterized in that: In step 4, the standardization process is as follows: first, the average value a of the normalized data of each parameter is calculated, and then the standard deviation b of the normalized data of each parameter is calculated. For each normalized data x* of each parameter, the formula c=(x*-a) / b is used to convert the normalized data x* of each parameter into a new data set c of each parameter.

5. The method for optimizing a working face gas concentration prediction model based on multi-source data according to claim 1 is characterized in that: In step 5, the ratio of training set to validation set is N:10-N.

6. The method for optimizing a working face gas concentration prediction model based on multi-source data according to claim 1, characterized in that: In step 5, the linear regression algorithm model is y = β0 + β1x1 + β2x2 + ... + β i x i , where β0 is the intercept term, i is the number of remaining parameters after removing the coefficient of variation less than 10%, β1, β2, ..., β i It is the coefficient of the remaining parameters after eliminating the parameters with coefficient of variation less than 10% in step 2.

7. The method for optimizing a working face gas concentration prediction model based on multi-source data according to claim 1 is characterized in that: In step 5, the logistic regression algorithm model is y is the predicted methane concentration of the working face, and P(y=1) represents the probability of belonging to the positive class.

8. The method for optimizing a working face gas concentration prediction model based on multi-source data according to claim 1 is characterized in that: The support vector machine algorithm model is f(x)=sign(β0+β1x1+β2x2+…+β i x i ), f(x) is the predicted methane concentration at the working face.

9. The method for optimizing a working face gas concentration prediction model based on multi-source data according to claim 1, characterized in that: In step six, compare the F1 scores of the linear regression algorithm model, the logistic regression algorithm model, and the support vector machine algorithm model, and select the algorithm model with a larger F1 score as the working face gas concentration prediction model. If the accuracy of the linear regression algorithm model, the logistic regression algorithm model, and the support vector machine algorithm model is the same, change the division ratio of the training set and the validation set to (N-1): (10-N-1), and repeat steps five to six.

Citation Information

Patent Citations

  • Multi-sensor fusion gas concentration multi-step prediction method based on deep learning

    CN113689032A

  • Coal face gas concentration prediction method based on multi-factor generalized linear regression

    CN116484323A

  • Mine gas emission prediction method of integrated learning mining historical model

    CN116911456A