Intelligent dynamic prediction method for coal and gas outburst dangerous area of working face

Abstract

The application discloses a kind of intelligent working face coal and gas outburst dangerous area dynamic prediction method, this method passes through establishing an integrated early warning intelligence analysis platform, dynamically acquires the key parameter in mine, and constructs dynamic dataset.Utilize data fusion technology, combine historical data and regression analysis, calculate coal seam firmness factor and gas diffusion factor.Further, combine static dataset, such as geological and ventilation system distribution data, construct multi-parameter coupling prediction model.The model is trained and verified by machine learning method to realize real-time prediction of coal and gas outburst risk in working face.The system automatically triggers an early warning according to the model prediction result and preset threshold, provides a scientific basis for coal mine safety management, effectively prevents coal and gas outburst accident.

Description

Technical Field

[0001] This invention relates to coal mine safety production technology, and in particular to a dynamic prediction method for coal and gas outburst hazard zones in intelligent working faces. Background Technology

[0002] In the prevention and control of coal and gas outburst accidents, coal and gas outburst early warning technology, as a feasible preventive measure before an accident occurs, is one of the most effective means to reduce and curb the occurrence of coal and gas outburst accidents, and has been effectively applied in the prevention and control of coal and gas outbursts in mines in recent years.

[0003] In existing technologies, the prediction and monitoring of coal mine gas often rely on limited sensors and monitoring equipment. These devices typically can only monitor gas concentrations and related parameters, such as ventilation volume and temperature, in specific areas. This limitation leads to incomplete data collection, failing to fully reflect the overall working conditions of the coal mine, especially under dynamically changing working face conditions. Furthermore, existing technologies are also inadequate in real-time data processing and analysis, making it difficult to achieve real-time early warning of gas outburst hazards.

[0004] Currently, in actual early warning processes, theoretical modeling is mainly used for coal and gas outburst warnings. However, the formation mechanism of coal and gas outbursts is complex and influenced by numerous factors. Using a single theory to describe the actual mechanism of coal and gas outbursts often fails to accurately analyze the factors affecting them. In terms of early warning technology, although traditional manual experience methods have high accuracy, they cannot achieve real-time early warnings and cannot accurately predict dangerous areas of coal and gas outbursts at the working face.

[0005] For example, Chinese patent CN118410316A provides a method for predicting and analyzing gas concentration in coal mining faces based on time series analysis. This method mainly relies on historical data and time series characteristics to predict gas concentration. However, this method may not be able to reflect the dynamic changes in working face conditions in real time because it lacks rapid processing and analysis of real-time data.

[0006] On the other hand, traditional gas prediction models are usually based on static data and empirical formulas, lacking adaptability to complex geological conditions and dynamic mining activities in coal mines. These models often fail to accurately predict gas emission rates and outburst risks, especially under complex geological conditions and frequent changes in mining activities. Furthermore, existing models often neglect the actual operating status of gas drainage equipment when considering gas prediction, leading to discrepancies between prediction results and actual conditions.

[0007] For example, Chinese patent CN111156048A proposes an intelligent coal mining face gas prediction and equipment linkage safety assurance system and method. Although the system considers equipment operating parameters, it mainly focuses on equipment control rather than adaptability to complex geological conditions. This may lead to insufficient accuracy and adaptability of the prediction model in variable geological environments.

[0008] Moreover, existing technologies often fail to fully leverage the advantages of multi-source information fusion when addressing gas prediction issues. For example, geological data, mining activity data, and gas monitoring data from coal mines are not effectively integrated, and intelligent data processing and analysis methods, such as machine learning and data mining techniques, are lacking to improve the accuracy and reliability of predictions. Furthermore, gas early warning systems also suffer from shortcomings in real-time performance, failing to achieve rapid response to gas outburst hazards.

[0009] In some existing coal mine monitoring systems, such as the multi-factor-based methane concentration prediction method for return airway in coal mining faces described in Chinese patent CN116484323A, the method typically relies on periodic data acquisition and analysis. This method may not provide real-time early warnings because it lacks immediate linkage with field equipment, resulting in a slow response in emergency situations.

[0010] Furthermore, the lack of an effective linkage mechanism between the early warning system and the actual production equipment in the coal mine, as well as the lack of an automatic response early warning intelligence analysis platform, means that even if an early warning occurs, it is impossible to adjust production equipment and operating procedures in a timely manner to reduce the risk of gas outbursts. Summary of the Invention

[0011] The purpose of this invention is to provide an intelligent method for dynamic prediction of coal and gas outburst hazard areas in working faces, thereby overcoming the shortcomings of existing technologies.

[0012] Based on the first main aspect of the present invention, a method for dynamic prediction of coal and gas outburst hazard zones in working faces is provided, comprising the following steps:

[0013] S1. Establish an intelligent early warning and intelligence analysis platform for coal and gas outbursts under mining conditions, dynamically collecting data on coal seam gas content W, coal damage type factor D, original coal seam gas pressure P0, and residual coal seam gas pressure P in the mine. r , drill cuttings quantity S, drill cuttings gas desorption index K1, drill cuttings gas desorption index Δh2, coal firmness coefficient f, and initial gas emission velocity index V;

[0014] S2, the collected data is integrated using a data fusion method to construct a dynamic dataset, and the dynamic dataset is uploaded to the coal and gas outburst early warning intelligence analysis platform under the intelligent mining conditions;

[0015] S3, Based on the dynamic dataset, construct functions in the coal and gas outburst early warning intelligence analysis platform under the intelligent mining conditions to calculate the coal seam firmness factor and gas emission factor;

[0016] S4. Input the mine geological data, mining engineering plan data, and mine ventilation system distribution data obtained from the design data into the coal and gas outburst early warning intelligence analysis platform under the intelligent mining conditions to construct a static dataset;

[0017] S5. A multi-parameter coupled prediction model is constructed using the coal seam firmness factor and gas emission factor and the static dataset as input features.

[0018] S6. Using the coal seam firmness factor and gas emission factor calculated in step S3, combined with the static dataset in step S4, a multi-parameter coupled prediction model is trained using machine learning methods, and the model is validated using cross-validation.

[0019] S7. The trained multi-parameter coupled prediction model is deployed to the coal and gas outburst early warning intelligence analysis platform under intelligent mining conditions to realize real-time prediction of coal and gas outburst risk at the working face; based on historical data and expert experience, an early warning threshold is set in the coal and gas outburst early warning intelligence analysis platform under the intelligent mining conditions; when the risk value predicted by the model exceeds the threshold, the system automatically triggers an early warning.

[0020] As a further preferred embodiment, in some embodiments, the integration of the collected data through a data fusion method includes data cleaning to remove missing and outlier values; data standardization to scale all parameters to a uniform numerical range; selecting and retaining parameters most relevant to the risk of coal and gas outbursts; and merging the processed data into a dynamic dataset.

[0021] As a further preferred embodiment, in some embodiments, the integration of the collected data through a data fusion method further includes one or a combination of the following methods: smoothing the time series data using a Kalman filter, reducing the dimensionality of high-dimensional data using principal component analysis (PCA), identifying the correlation between parameters using association rule mining techniques, and achieving distributed processing and integration of data through federated learning or a multi-agent system.

[0022] As a further preferred embodiment, in some embodiments, the formula for calculating the coal seam firmness factor is:

[0023]

[0024] The formula for calculating the gas emission factor is as follows:

[0025]

[0026] Where W is the coal seam gas content, P0 is the original coal seam gas pressure, and P max To predict the maximum possible gas content, D is the coal damage type factor, with values ​​of 0, 0.2, 0.4, 0.6, and 0.8 corresponding to the five coal types. S is the amount of drill cuttings, and P... r denoted as the residual gas pressure in the coal seam, f as the coal firmness coefficient, V as the initial velocity index of gas emission, and K1 and Δh2 as two desorption indices for gas from drill cuttings.

[0027] As a further preferred embodiment, in some embodiments, the multi-parameter coupled prediction model is expressed as follows:

[0028]

[0029] In the formula, R represents the predicted risk of coal and gas outbursts; T represents the number of decision trees in the random forest; R t It is the prediction result of the t-th decision tree;

[0030] Each decision tree R t It is represented as:

[0031]

[0032] K t It is the number of split nodes in the t-th decision tree;

[0033] ω t,k It is the weight associated with the k-th split node of the t-th decision tree;

[0034] I k (x) is an indicator function that returns 1 if the input feature x satisfies the condition of the k-th split node, and 0 otherwise.

[0035] As a further preferred embodiment, in some embodiments, the multi-parameter coupled prediction model is constructed according to the following steps:

[0036] The coal seam firmness factor and gas emission factor calculated in step S3, and the static dataset in step S4 are combined into a feature matrix X;

[0037] The historical risk level of coal and gas outbursts is used as the target variable y;

[0038] A random forest model is trained using the feature matrix X and the target variable y.

[0039] Determine the number T of decision trees in the random forest and the depth or other hyperparameters of each decision tree;

[0040] For a new input feature x, a trained random forest model is used to make a prediction to obtain the risk level R of coal and gas outburst.

[0041] As a further preferred embodiment, some embodiments also include integrating GIS software into the coal and gas outburst early warning intelligence analysis platform under the intelligent mining conditions, creating a geographic information database of the mine, associating the results of the multi-parameter coupled prediction model with the geographic information data of the mine, dividing the area into regions with different risk levels based on the output of the multi-parameter coupled prediction model, and using GIS spatial analysis tools to further identify and visualize the risk areas.

[0042] Based on a second key aspect of the present invention, an intelligent early warning system for coal and gas outbursts under mining conditions is provided for use in the aforementioned method. This system includes at least one or a combination of the following modules:

[0043] The data acquisition module is used to dynamically collect data on the coal seam gas content W, coal damage type factor D, original coal seam gas pressure P0, and residual coal seam gas pressure P in the mine. r , drill cuttings quantity S, drill cuttings gas desorption index K1, drill cuttings gas desorption index Δh2, coal firmness coefficient f, and initial gas emission velocity index V;

[0044] The data fusion module is used to integrate the collected data, construct a dynamic dataset, and upload the dynamic dataset to the early warning intelligence analysis platform.

[0045] The factor calculation module is used to construct functions in the early warning intelligence analysis platform to calculate the coal seam stability factor and gas emission factor.

[0046] The static dataset construction module is used to input mine geological data, mining engineering plan data, and mine ventilation system distribution data to construct a static dataset.

[0047] A multi-parameter coupled prediction model construction module is used to construct a multi-parameter coupled prediction model using the coal seam firmness factor and gas emission factor and the static dataset as input features;

[0048] The model training and validation module is used to train multi-parameter coupled prediction models using machine learning methods and to validate the models using cross-validation.

[0049] The real-time prediction and early warning module is used to deploy the trained multi-parameter coupled prediction model to the early warning intelligence analysis platform to realize real-time prediction of coal and gas outburst risks at the working face, and automatically trigger early warnings according to the set early warning thresholds.

[0050] Based on a third key aspect of the present invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed, implements the aforementioned intelligent method for dynamic prediction of coal and gas outburst hazard zones in working faces.

[0051] According to a fourth principal aspect of the present invention, an electronic device is provided, comprising:

[0052] At least one processor;

[0053] A memory that is communicatively connected to the at least one processor;

[0054] The memory stores a computer program that, when executed by the at least one processor, enables the at least one processor to implement the aforementioned intelligent method for dynamic prediction of coal and gas outburst hazard areas in working faces.

[0055] Advantages and beneficial effects of the present invention:

[0056] First, this invention dynamically assesses the risk of coal and gas outbursts by real-time monitoring and analysis of key parameters at the coal mine working face, such as coal seam gas pressure and drill cuttings gas desorption index. This dynamic monitoring method improves the timeliness and accuracy of risk assessment, enabling coal mine safety management to respond more quickly to potential safety threats.

[0057] Secondly, through data fusion technology and historical data analysis, this invention can comprehensively consider multiple influencing factors, including the physical properties of the coal seam and the engineering conditions of the mine, thereby constructing more accurate coal seam stability factors and gas emission factors. This method not only improves the reliability of the prediction model but also enhances the understanding of coal and gas outburst behavior under complex geological conditions.

[0058] Furthermore, this invention employs machine learning techniques to train a multi-parameter coupled prediction model and performs cross-validation to ensure the model's generalization ability and prediction accuracy. This data-driven approach can adapt to constantly changing mine conditions, update risk assessment results in real time, and provide coal mines with more flexible and accurate safety warnings.

[0059] Finally, the early warning system of this invention can automatically trigger early warnings based on the risk values ​​predicted by the model and preset thresholds, promptly notifying mine managers to take corresponding safety measures. This automated early warning mechanism greatly reduces delays and errors caused by human factors, improves the coal mine's ability to prevent coal and gas outburst accidents, thereby effectively reducing the accident rate and protecting the lives of miners and the production efficiency of the coal mine. Attached Figure Description

[0060] Figure 1 The workflow of one embodiment of the present invention is illustrated;

[0061] Figure 2 A system composition diagram of one embodiment of the present invention is shown. Detailed Implementation

[0062] The preferred embodiments of the present invention will be described in detail below to provide a clearer understanding of the purpose, features, and advantages of the invention. It should be understood that the following embodiments are not intended to limit the scope of the invention, but are merely illustrative of the essential spirit of the technical solution of the invention.

[0063] In the following description, certain specific details are set forth for the purpose of illustrating various disclosed embodiments in order to provide a thorough understanding of the various disclosed embodiments. However, those skilled in the art will recognize that embodiments may be practiced without one or more of these specific details. In other instances, well-known techniques associated with this application may not have been shown or described in detail to avoid unnecessarily obscuring the description of the embodiments.

[0064] Throughout this specification, references to "an embodiment" or "an embodiment" indicate that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. Therefore, the appearance of "in an embodiment" or "an embodiment" in various places throughout the specification does not necessarily refer to the same embodiment. Furthermore, a particular feature, structure, or characteristic may be combined in any manner in one or more embodiments.

[0065] like Figure 1 According to one embodiment of the present invention, an intelligent method for dynamic prediction of coal and gas outburst hazard zones in working faces includes the following steps:

[0066] S1. Establish an intelligent early warning and intelligence analysis platform for coal and gas outbursts under mining conditions, dynamically collecting data on coal seam gas content W, coal damage type factor D, original coal seam gas pressure P0, and residual coal seam gas pressure P in the mine. r The parameters are: drill cuttings quantity S, drill cuttings gas desorption index K1, drill cuttings gas desorption index Δh2, coal firmness coefficient f, and initial gas emission velocity index V.

[0067] In practical implementation, the first step is to establish an intelligent coal and gas outburst early warning and intelligence analysis platform under intelligent mining conditions, which forms the foundation of the entire prediction method. This platform needs to have the capability for real-time data acquisition, processing, and analysis to ensure accurate capture of various key parameters in the mine. To this end, a series of borehole sampling test devices, sensors, or monitoring equipment are deployed in the mine to measure key indicators such as coal seam gas pressure, original coal seam gas pressure, and residual coal seam gas pressure.

[0068] Secondly, the platform needs to have high data compatibility, capable of receiving and processing data from different sensors and monitoring devices. This data includes indicators such as drill cuttings gas desorption index K1, drill cuttings gas desorption index Δh2, coal firmness coefficient, and initial gas release velocity. This data is crucial for predicting the risk of coal and gas outbursts; therefore, the platform must be able to accurately parse and store this data for subsequent analysis and prediction.

[0069] Finally, to ensure the real-time nature and accuracy of the data, the platform also needs to have data verification and anomaly detection capabilities. Therefore, the platform should be able to identify and exclude erroneous or abnormal data points, ensuring that only high-quality data is used for subsequent analysis. Furthermore, the platform should provide a user interface that allows operators to monitor data acquisition in real time and intervene manually when necessary.

[0070] In practice, this platform can adopt a hybrid cloud architecture to achieve functions such as big data analysis, risk and hazard identification, and hazard classification and early warning for gas-related data from a single coal mine or multiple coal mines in a region. Through the cloud platform's automatic comparison, analysis, and judgment, it can issue early warnings of data anomalies, achieving advanced early warning of gas emissions.

[0071] S2, the collected data is integrated using a data fusion method to construct a dynamic dataset, and the dynamic dataset is uploaded to the coal and gas outburst early warning intelligence analysis platform under the intelligent mining conditions.

[0072] The key to implementing step S2 lies in the application of data fusion methods. In the coal and gas outburst early warning intelligence analysis platform under intelligent mining conditions, it is necessary to integrate data from different sensors and monitoring equipment. This data includes coal seam gas pressure, original coal seam gas pressure, residual coal seam gas pressure, drill cuttings gas desorption index K1, drill cuttings gas desorption index Δh2, coal firmness coefficient, and initial gas emission velocity index, etc. Data fusion is the process of merging information from these heterogeneous data sources into a consistent, accurate, and usable dataset. In this embodiment, it is achieved by using an application programming interface (API) to call hardware resources, acquiring and integrating multi-source heterogeneous data collected by various monitoring systems in the coal mine, and performing necessary data cleaning.

[0073] Secondly, constructing a dynamic dataset is the core part of step S2. Based on data fusion, the integrated data needs to be constructed into a dynamic dataset. This dynamic dataset can be updated in real time to reflect the latest situation in the mine. The construction of the dynamic dataset can be achieved by establishing a standardized, distributed database based on the needs of coal mine gas disaster prevention and control. Such a database can provide a high-quality data foundation for coal mine gas disaster risk analysis and early warning.

[0074] Finally, the ultimate goal of step S2 is to upload the constructed dynamic dataset to the intelligent mining conditions coal and gas outburst early warning intelligence analysis platform. This platform can monitor the usage of cloud platform storage and computing resources in real time, and achieve dynamic allocation of resources through resource operation management to ensure the normal operation of the cloud platform. After uploading the dynamic dataset, the platform can use this data for further analysis and prediction, providing scientific data support for coal mine safety.

[0075] In some embodiments, acoustic and electromagnetic gas monitoring and early warning technologies and equipment can be used to monitor the coal and gas outburst risk throughout the tunneling process in real time. A dynamic dataset is constructed by integrating data from different sensors, such as acoustic emission, electromagnetic radiation, and gas signals. This data is integrated using data fusion methods to ensure consistency and accuracy. The cloud platform automatically processes and analyzes the acoustic and electromagnetic gas monitoring data from the backend, and automatically identifies and issues early warning alerts based on trend-based early warning algorithms to indicate potential outburst risks.

[0076] In some embodiments, the coal and gas outburst early warning intelligence analysis platform under intelligent mining conditions can also be designed based on big data technology. The data transmission layer consists of a mine-side data transmission service and a Kafka cluster, responsible for encapsulating the processed data into Kafka message objects and sending them to the cloud platform Kafka cluster. During this process, the data is compressed and encapsulated into data packets to improve transmission efficiency, and is locally cached in case of network anomalies, with batch uploads occurring after network recovery. This design ensures the real-time nature and integrity of the data, providing a solid foundation for subsequent data analysis and prediction.

[0077] In some embodiments, a Kalman filter can be used to smooth time series data to reduce noise and improve data accuracy. For example, by collecting time series data on gas concentration, a Kalman filter can predict and correct future data points, resulting in smoother and more reliable time series data. The Kalman filter works by estimating the system's state and covariance, and then updating these estimates using observed data to smooth the time series data.

[0078] When processing high-dimensional monitoring data in coal mines, Principal Component Analysis (PCA) can be applied to reduce the dimensionality of the data while retaining the most important information. For example, PCA can extract several key components from large amounts of data collected by multiple sensors or experiments, which can explain most of the data variability. In this way, the complexity of the model can be simplified, processing efficiency can be improved, and the consumption of computational resources can be reduced.

[0079] Furthermore, association rule mining techniques can be used to identify correlations between different parameters in coal mine monitoring data. For example, by analyzing parameters such as gas concentration, ventilation volume, and geological conditions, correlation rules can be discovered among them, thereby predicting the risk of coal and gas outbursts. This technique can help us understand how different parameters jointly affect coal mine safety and provide a basis for preventive measures.

[0080] In coal mine safety monitoring, federated learning or multi-agent systems can be used to achieve distributed data processing and integration. For example, each coal mine can act as a node, processing its own data locally and collaboratively training a global model through a federated learning algorithm without sharing the original data. This approach not only protects data privacy but also improves the model's generalization ability. In a multi-agent system, each agent can represent a coal mine, and they communicate and collaborate to solve common problems, such as predicting the risk of coal and gas outbursts.

[0081] S3. Based on the dynamic dataset, a function is constructed in the coal and gas outburst early warning intelligence analysis platform under the intelligent mining conditions to calculate the coal seam firmness factor and gas emission factor.

[0082] In the coal and gas outburst early warning intelligence analysis platform under intelligent mining conditions, the first step is to construct a function to calculate the coal seam firmness factor and gas emission factor. Specifically, by analyzing the physical parameters of the coal seam and the chemical characteristics of the gas, the key factors affecting coal and gas outbursts can be identified, and a mathematical model for calculating the coal seam firmness factor and gas emission factor can be constructed accordingly.

[0083] In this embodiment, the formula for calculating the coal seam firmness factor is as follows:

[0084]

[0085] The formula for calculating the gas emission factor is as follows:

[0086]

[0087] Where W is the coal seam gas content, P0 is the original coal seam gas pressure, and P max To predict the maximum possible gas content, D is the coal damage type factor, with values ​​of 0, 0.2, 0.4, 0.6, and 0.8 corresponding to the five coal types. S is the amount of drill cuttings, and P... r denoted as the residual gas pressure in the coal seam, f as the coal firmness coefficient, V as the initial velocity index of gas emission, and K1 and Δh2 as two desorption indices for gas from drill cuttings.

[0088] The main purpose of constructing the above functions to calculate the coal seam firmness factor and gas emission factor is to quantify the physical properties of the coal seam and the dynamic changes in gas, thereby providing a scientific basis for the risk assessment of coal and gas outbursts. The coal seam firmness factor reflects the strength of the coal body itself, that is, its ability to resist outbursts, while the gas emission factor reflects the gas emission capacity of the coal body, that is, the energy source supply capacity for outbursts. These two factors are important indicators in coal and gas outburst prediction, and their calculation results directly affect the accuracy and reliability of the prediction model.

[0089] The coal firmness coefficient reflects the physical strength and structural integrity of the coal, and is generally related to the coal type, maturity, and geological history. A high firmness coefficient means the coal seam is relatively hard and not easily broken. Gas pressure is a key factor affecting coal seam stability. High gas pressure may lead to coal seam fracturing, increasing the risk of gas release and outbursts.

[0090] In the formula, the maximum possible gas pressure is a reference value used to standardize gas pressure for comparing gas pressure levels in different coal seams. The type of coal failure directly affects its firmness and the risk of gas outbursts. The failure type factor is a dimensionless logical variable used to quantify the impact of the degree of coal failure on firmness. The calculation principle of the coal seam firmness factor is to comprehensively consider the physical properties of the coal (firmness coefficient), the influence of gas pressure (the original gas pressure of the coal seam relative to the maximum possible gas pressure), and the type of coal failure (failure type factor) to assess the overall firmness of the coal seam.

[0091] On the other hand, the calculation of the gas emission factor mainly considers indicators such as coal seam gas content, drill cuttings volume S, drill cuttings gas desorption index K1, and Δh2, and comprehensively considers the dimensions of each indicator to construct the factor. Among these, coal seam gas content is a key indicator for measuring the amount of gas in the coal seam and directly affects the gas emission potential. Drill cuttings volume reflects the amount of debris generated during drilling and is related to the degree of coal seam fragmentation and gas release. Drill cuttings gas desorption indices K1 and Δh2 reflect the desorption characteristics of gas in the drill cuttings; K1 represents the desorption rate, while Δh2 represents the pressure drop during the desorption process. The initial gas emission velocity describes the initial rate at which gas is released from the coal seam and is closely related to the risk of gas outbursts.

[0092] The calculation principle of the gas emission factor is to assess the gas emission potential of the coal seam by comprehensively considering the gas content, drill cuttings quantity, gas desorption characteristics of drill cuttings, and initial gas emission velocity. This factor quantifies the potential risk of gas emission by comparing the gas generation and release rate (numerator) with the gas pressure limit (denominator).

[0093] In step S3, a model self-correction mechanism and a cause-tracing function for early warnings can also be considered. By combining association rule algorithms and evidence theory algorithms, automatic multi-indicator fusion analysis and decision-making can be achieved, realizing the goals of self-correction of the early warning model and traceability of early warning causes. This method allows the system to automatically adjust the early warning model based on new data, improving the accuracy of early warnings, and in the event of a significant incident, it can trace the specific cause of the early warning, providing a basis for subsequent prevention and response measures.

[0094] S4. Input the mine geological data, mining engineering plan data, and mine ventilation system distribution data obtained from the design data into the coal and gas outburst early warning intelligence analysis platform under the intelligent mining conditions to construct a static dataset.

[0095] In intelligent mining conditions, the construction of static datasets is a fundamental task in the coal and gas outburst early warning intelligence analysis platform. This requires obtaining mine geological data, mining engineering plan data, and mine ventilation system distribution data from design documents. Unlike dynamic data, these data are determined in the design documents, so their values ​​do not change with the production process. Geological data includes information such as coal seam thickness, dip angle, and coal quality; mining engineering plan data involves the layout of the working face, the direction of the roadways, and the development progress; and ventilation system distribution data includes the ventilation network and equipment layout. These data provide necessary background information for the prediction model and are an important basis for assessing the risk of coal and gas outbursts.

[0096] The collected static data needs to be integrated and standardized to ensure consistency and comparability. For example, all length units should be standardized to meters, all pressure units to Pascals, and missing or outlier data should be interpolated or corrected. Furthermore, the data needs to be classified and coded to facilitate identification and processing by computer systems.

[0097] The integrated and standardized static dataset will be input into the early warning intelligence analysis platform as one of the input features for constructing a multi-parameter coupled prediction model. This data, along with the coal seam firmness factor and gas emission factor calculated in step S3, will be used to train and validate the multi-parameter coupled prediction model. Before inputting the data, it is necessary to ensure its accuracy and completeness to avoid biases during model training. Furthermore, appropriate machine learning algorithms and tools, such as decision trees, random forests, and support vector machines, need to be prepared for model training and optimization. Through these steps, the static dataset will provide solid data support for predicting coal and gas outburst risks.

[0098] S5. A multi-parameter coupled prediction model is constructed using the coal seam firmness factor and gas emission factor, along with the static dataset, as input features.

[0099] In step S5, a multi-parameter coupled prediction model is constructed using the coal seam firmness factor and gas emission factor along with a static dataset as input features. The model is built based on machine learning algorithms; in this embodiment, the random forest algorithm is used. Random forest is an ensemble learning algorithm that improves the accuracy and robustness of the model by constructing multiple decision trees and combining their prediction results. In coal and gas outburst prediction, random forest can handle high-dimensional data, automatically handle missing and outlier values, and evaluate the importance of features.

[0100] In this embodiment, the multi-parameter coupled prediction model is expressed as follows:

[0101]

[0102] In the formula, R represents the predicted risk of coal and gas outbursts; T represents the number of decision trees in the random forest; R t It is the prediction result of the t-th decision tree;

[0103] Each decision tree R t It is represented as:

[0104]

[0105] K t It is the number of split nodes in the t-th decision tree;

[0106] ω t,k It is the weight associated with the k-th split node of the t-th decision tree;

[0107] I k (x) is an indicator function that returns 1 if the input feature x satisfies the condition of the k-th split node, and 0 otherwise.

[0108] The construction process is as follows: First, the coal seam firmness factor and gas emission factor calculated in step S3, along with the static dataset from step S4, are merged into a feature matrix X; the historical risk level of coal and gas outbursts is used as the target variable y; a random forest model is trained using the feature matrix X and the target variable y. The number T of decision trees in the random forest and the depth or other hyperparameters of each decision tree are determined; for a new input feature x, the trained random forest model is used to make a prediction to obtain the risk level R of coal and gas outbursts.

[0109] S6. Using the coal seam firmness factor and gas emission factor calculated in step S3, combined with the static dataset in step S4, a multi-parameter coupled prediction model is trained using machine learning methods, and the model is validated using cross-validation.

[0110] In step S6, the present invention uses the coal seam firmness factor and gas emission factor calculated in step S3, combined with the static dataset in step S4, to apply machine learning methods to train a multi-parameter coupled prediction model. First, the data is divided into training and test sets. Then, the training set is further divided into training and validation sets to perform cross-validation during model training. This can be achieved using the `train_test_split` function, ensuring the model's generalization ability on unseen data. Next, machine learning algorithms, such as random forests and support vector machines, are used to train the model, and cross-validation is used to evaluate the model's performance to ensure its stability and accuracy.

[0111] To optimize the input of the grey relational model, Analytic Network Analysis (ANP) can be applied. ANP is a priority-based analysis method that can filter out indicators with a significant impact on coal and gas outbursts from a large number of prediction indicators and quantify the sensitivity of these indicators. This method can identify which factors have the most significant impact on coal and gas outbursts and use these factors as the key input features of the model.

[0112] Before model training, input features can be appropriately preprocessed, including standardization, normalization, categorical feature processing, and textual feature processing. For example, StandardScaler and MinMaxScaler from the scikit-learn library can be used for feature scaling to ensure the consistency of data distribution. For categorical variables, one-hot encoding or label encoding can be used. These preprocessing steps help improve the training efficiency and prediction accuracy of the model. Through these steps, a high-quality training dataset can be built, laying a solid foundation for training multi-parameter coupled prediction models.

[0113] Using cross-validation methods to validate the model maximizes data utilization, such as K-fold cross-validation and leave-one-out cross-validation. Furthermore, feature selection is a crucial step in improving model performance. By training a random forest model and obtaining feature importance, the features most influential on the model's predictions can be identified. Based on feature importance, the most important features can be selected to be retained, or a threshold can be set to filter features, thereby simplifying the model and potentially improving its performance. This process can be visualized by plotting a bar chart of feature importance, and features to be retained can be selected based on the bar chart. These steps ensure the accuracy and efficiency of the model in predicting coal and gas outburst risks.

[0114] S7. The trained multi-parameter coupled prediction model is deployed to the coal and gas outburst early warning intelligence analysis platform under intelligent mining conditions to realize real-time prediction of coal and gas outburst risk at the working face; based on historical data and expert experience, an early warning threshold is set in the coal and gas outburst early warning intelligence analysis platform under the intelligent mining conditions; when the risk value predicted by the model exceeds the threshold, the system automatically triggers an early warning.

[0115] In step S7, the multi-parameter coupled prediction model trained and validated in step S6 will be deployed to the coal and gas outburst early warning intelligence analysis platform under intelligent mining conditions. This deployment process integrates the model into the existing monitoring system, ensuring that the model can receive the latest monitoring data in real time and perform risk prediction. After deployment, the model will run automatically, continuously analyze the input data, and output the risk assessment results of coal and gas outbursts.

[0116] Based on historical data and expert experience, early warning thresholds need to be set in the early warning intelligence analysis platform. These thresholds are the critical points that trigger an early warning when the risk assessment results are reached or exceeded. The setting of thresholds needs to comprehensively consider the historical frequency, severity, and potential consequences of coal and gas outbursts. In practice, the technical or safety manager of the coal mine sets the thresholds on the platform, specifically choosing from multiple preset thresholds. For example, if the platform presets five risk thresholds (1, 2, 3, 4, and 5), the technical or safety manager of the coal mine selects one as the early warning threshold for their specific coal mine and enters it into the system.

[0117] Once the risk value predicted by the model exceeds the set threshold, the system will automatically trigger an early warning. This warning can be an audible and visual alarm, an SMS notification, or an automatic prompt within the system, ensuring that relevant personnel receive the warning information in a timely manner and take appropriate preventive or emergency measures. Simultaneously, the system should also have recording and feedback functions to collect actual event data after the warning is triggered, for subsequent model optimization and threshold adjustment. This closed-loop early warning and response mechanism can effectively improve the safety of coal mine operations and reduce the occurrence of coal and gas outburst accidents.

[0118] In some embodiments, the method further includes integrating GIS software into the coal and gas outburst early warning intelligence analysis platform under the intelligent mining conditions, creating a geographic information database for the mine, associating the results of the multi-parameter coupled prediction model with the geographic information data of the mine, dividing the area into regions with different risk levels based on the output of the multi-parameter coupled prediction model, and using GIS spatial analysis tools to further identify and visualize the risk areas.

[0119] For example, using GIS software, a database containing geographic information such as mine geological structure, roadway layout, and ventilation system is created. High-resolution geological exploration data of the mine, including coal seam thickness, dip angle, and coal quality, is collected and imported into the GIS database. Mining engineering plan data and mine ventilation system distribution data are integrated to ensure accurate correspondence of all data within the GIS. The output of multi-parameter coupled prediction models, such as coal seam firmness factors and gas emission factors, are correlated with geographic information data in the GIS database. An interface is designed to allow the model's output to be directly imported into the GIS system and matched with corresponding geographic locations. Based on the risk assessment results output by the model, GIS spatial analysis tools are used to divide the mine into areas with different risk levels. A risk level color coding system is designed to visually display risk levels on the GIS map using different colors. An interactive risk map is implemented, allowing users to click on specific areas to view detailed risk assessment data and prediction results.

[0120] For example, a 3D model of the mine can be created using oblique photogrammetry and 3D laser scanning technology and integrated into GIS software. This 3D model is then combined with geological, mining, and ventilation data to form a complete 3D geographic information database. The results of a multi-parameter coupled prediction model are correlated with the 3D GIS model to achieve 3D visualization of risk assessment. In the 3D model, areas with different risk levels are highlighted using varying transparency or color, allowing users to observe and analyze risk distribution from multiple perspectives. GIS spatial analysis tools, such as buffer analysis and overlay analysis, are used to further identify potential risk areas. An early warning system is designed to automatically highlight high-risk areas in the 3D model and send warning notifications to relevant personnel when the prediction model identifies them.

[0121] Alternatively, develop a web-based GIS platform that allows users to access coal and gas outburst risk information via a web browser. Integrate geographic information databases and multi-parameter coupled prediction models into the web GIS platform. Provide a collaboration tool that allows different departments and teams to share risk assessment results and geographic information. Implement a user-friendly interface so that non-technical users can easily view and understand risk maps. Design the platform to support dynamic data updates, ensuring that risk assessment results and geographic information are always up-to-date. Implement a real-time early warning function; when the monitored risk value exceeds a preset threshold, the system will automatically send an early warning to the user and dynamically mark the risk area on the map.

[0122] like Figure 2As shown, this invention provides an intelligent coal and gas outburst early warning system for use in the aforementioned method. This system serves as the basic architecture of an intelligent coal and gas outburst early warning intelligence analysis platform under intelligent mining conditions. The system includes at least one of the following modules or a combination thereof:

[0123] The data acquisition module is used to dynamically collect data on coal seam gas pressure, original coal seam gas pressure, residual coal seam gas pressure, drill cuttings gas desorption index K1, drill cuttings gas desorption index Δh2, coal firmness coefficient, and initial gas emission velocity in the mine. The module includes various sensors and testing devices deployed in the mine for real-time monitoring of key parameters such as coal seam gas pressure, original coal seam gas pressure, and residual coal seam gas pressure. A general-purpose controller or a dedicated data acquisition module can be used to collect physical signals from the field, convert these physical signals into electrical signals, and then convert them into readable digital quantities through an analog-to-digital converter (A / D) or digital circuitry.

[0124] The data fusion module integrates the collected data to construct a dynamic dataset, which is then uploaded to the early warning intelligence analysis platform. Multi-source information fusion technologies, such as association rule algorithms and evidence theory algorithms, can be employed to integrate the collected data and construct the dynamic dataset. This helps process information from heterogeneous data sources and improves data consistency and accuracy. The completed dynamic dataset is uploaded to the intelligent mining conditions coal and gas outburst early warning intelligence analysis platform via wired or wireless communication. During this process, a network communication unit can be used to parse the data acquisition protocol based on fieldbus, industrial Ethernet, or standard Ethernet. Alternatively, the data fusion module can be directly integrated into the intelligent mining conditions coal and gas outburst early warning intelligence analysis platform.

[0125] The factor calculation module is used to construct functions in the early warning intelligence analysis platform to calculate the coal seam stability factor and gas emission factor.

[0126] The static dataset construction module is used to construct a static dataset by inputting mine geological data, mining engineering plan data, and mine ventilation system distribution data. This data can include information such as coal seam thickness, dip angle, and coal quality, as well as the layout of the mining engineering and the distribution of the ventilation system.

[0127] The multi-parameter coupled prediction model construction module is used to construct a multi-parameter coupled prediction model using the coal seam firmness factor and gas emission factor and the static dataset as input features; this module constructs a multi-parameter coupled prediction model using the coal seam firmness factor and gas emission factor and the static dataset as input features.

[0128] The model training and validation module is used to train multi-parameter coupled prediction models using machine learning methods and to validate these models using cross-validation. This module applies machine learning methods, such as random forests or neural networks, to train multi-parameter coupled prediction models. The `train_test_split` and `KFold` functions from the sklearn library can be used for dataset splitting and cross-validation. Cross-validation is used to validate the model, ensuring its stability and accuracy. This can be achieved using the `KFold` function, which allows the model to be trained and tested on different subsets of data.

[0129] The real-time prediction and early warning module is used to deploy the trained multi-parameter coupled prediction model to the early warning intelligence analysis platform, enabling real-time prediction of coal and gas outburst risks at the working face and automatically triggering early warnings based on set early warning thresholds. Specifically, the trained multi-parameter coupled prediction model is deployed to the early warning intelligence analysis platform to achieve real-time prediction of coal and gas outburst risks at the working face. Early warnings are automatically triggered based on set early warning thresholds. When the risk value predicted by the model exceeds the threshold, the system will automatically issue an early warning signal, prompting relevant personnel to take appropriate preventative measures.

[0130] Furthermore, as a practical application, a computer-readable storage medium, such as a solid-state drive (SSD) or hard disk drive (HDD), can be designed to store computer programs for implementing intelligent early warning methods for coal and gas outbursts under mining conditions. This storage medium contains a series of computer programs, including code for modules such as data acquisition, data fusion, factor calculation, static dataset construction, multi-parameter coupled prediction model construction, model training and validation, and real-time prediction and early warning. When these programs are executed by a processor, they can implement a dynamic prediction method for coal and gas outburst hazard areas in intelligent working faces.

[0131] Alternatively, an electronic device may be designed, comprising at least one processor and a memory communicatively connected to the processor. The memory stores a computer program that, when executed by the at least one processor, enables the at least one processor to implement a dynamic prediction method for coal and gas outburst hazard zones in intelligent working faces.

[0132] The processor reads the program from the memory and executes the following processes: receiving a first request from the terminal, indicating a request for edge computing services; authenticating the terminal according to a second request to obtain first authentication information; receiving a third request from a second network element to verify the terminal's edge computing authorization information; obtaining a token according to the third request; and sending the token to the second network element. The processor is also used to obtain the terminal's token from the UDM, or obtain a stored token, or obtain the terminal's token from the first network element to provide edge computing services.

[0133] The above electronic devices can be a server, a personal computer, or a mobile terminal. By executing programs stored in the memory, they can perform functions such as data acquisition, fusion, factor calculation, model building, training and verification, and real-time prediction and early warning.

[0134] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A method for dynamic prediction of coal and gas outburst hazard zones in intelligent working faces, characterized in that, The method includes the following steps: S1. Establish an intelligent early warning and intelligence analysis platform for coal and gas outbursts under mining conditions, dynamically collecting data on coal seam gas content W, coal damage type factor D, original coal seam gas pressure P0, and residual coal seam gas pressure P in the mine. r , drill cuttings quantity S, drill cuttings gas desorption index K1, drill cuttings gas desorption index Δh2, coal firmness coefficient f, and initial gas emission velocity index V; S2, the collected data is integrated using a data fusion method to construct a dynamic dataset, and the dynamic dataset is uploaded to the coal and gas outburst early warning intelligence analysis platform under the intelligent mining conditions; S3, Based on the dynamic dataset, construct functions in the coal and gas outburst early warning intelligence analysis platform under the intelligent mining conditions to calculate the coal seam firmness factor and gas emission factor; The formula for calculating the coal seam firmness factor is as follows: ； The formula for calculating the gas emission factor is as follows: ； Among them, P max To predict the maximum possible gas content, the values ​​for the damage types of the five types of coal corresponding to D are 0, 0.2, 0.4, 0.6, and 0.8, respectively. S4. Input the mine geological data, mining engineering plan data, and mine ventilation system distribution data obtained from the design data into the coal and gas outburst early warning intelligence analysis platform under the intelligent mining conditions to construct a static dataset; S5, using the coal seam firmness factor and gas emission factor along with the static dataset as input features, construct a multi-parameter coupled prediction model; the multi-parameter coupled prediction model is constructed according to the following steps: The coal seam firmness factor and gas emission factor calculated in step S3, and the static dataset in step S4 are combined into a feature matrix X; The historical risk level of coal and gas outbursts is used as the target variable y; A random forest model is trained using the feature matrix X and the target variable y. Determine the number T of decision trees in the random forest and the depth of each decision tree; For a new input feature x, a trained random forest model is used to predict the risk level R of coal and gas outbursts. The multi-parameter coupled prediction model is expressed as follows: ； In the formula, R represents the predicted risk of coal and gas outbursts; T represents the number of decision trees in the random forest; R t It is the prediction result of the t-th decision tree; S6. Using the coal seam firmness factor and gas emission factor calculated in step S3, combined with the static dataset in step S4, a multi-parameter coupled prediction model is trained using machine learning methods, and the model is validated using cross-validation. S7. The trained multi-parameter coupled prediction model is deployed to the coal and gas outburst early warning intelligence analysis platform under intelligent mining conditions to realize real-time prediction of coal and gas outburst risk at the working face; based on historical data and expert experience, an early warning threshold is set in the coal and gas outburst early warning intelligence analysis platform under the intelligent mining conditions; when the risk value predicted by the model exceeds the threshold, the system automatically triggers an early warning.

2. The intelligent dynamic prediction method for coal and gas outburst hazard zones in working faces according to claim 1, characterized in that, The process of integrating the collected data using data fusion methods includes: data cleaning to remove missing and outlier values; data standardization to scale all parameters to a uniform numerical range; selecting and retaining parameters most relevant to coal and gas outburst risk; and merging the processed data into a dynamic dataset.

3. The intelligent dynamic prediction method for coal and gas outburst hazard zones in working faces according to claim 1, characterized in that, The integration of the collected data through data fusion methods also includes one or a combination of the following methods: using a Kalman filter to smooth the time series data, applying principal component analysis to reduce the dimensionality of high-dimensional data, using association rule mining techniques to identify the correlation between parameters, and achieving distributed processing and integration of data through federated learning or multi-agent systems.

4. The intelligent dynamic prediction method for coal and gas outburst hazard zones in working faces according to claim 1, characterized in that, Each decision tree R t Represented as: ； K t It is the number of split nodes in the t-th decision tree; It is the weight associated with the k-th split node of the t-th decision tree; It is an indicator function that returns 1 if the input feature x satisfies the condition of the k-th split node, and 0 otherwise.

5. The intelligent dynamic prediction method for coal and gas outburst hazard zones in working faces according to claim 1, characterized in that, It also includes integrating GIS software into the coal and gas outburst early warning intelligence analysis platform under the intelligent mining conditions, creating a geographic information database for the mine, linking the results of the multi-parameter coupled prediction model with the geographic information data of the mine, dividing the area into regions with different risk levels based on the output of the multi-parameter coupled prediction model, and using GIS spatial analysis tools to further identify and visualize the risk areas.

6. An intelligent early warning system for coal and gas outbursts under mining conditions for implementing the method of any one of claims 1-5, characterized in that... The system includes: The data acquisition module is used to dynamically collect data on the following parameters in the mine: coal seam gas content W, coal damage type factor D, original coal seam gas pressure P0, residual coal seam gas pressure Pr, drill cuttings volume S, drill cuttings gas desorption index K1, drill cuttings gas desorption index Δh2, coal firmness coefficient f, and initial gas emission velocity index V. The data fusion module is used to integrate the collected data, construct a dynamic dataset, and upload the dynamic dataset to the early warning intelligence analysis platform. The factor calculation module is used to construct functions in the early warning intelligence analysis platform to calculate the coal seam stability factor and gas emission factor. The static dataset construction module is used to input mine geological data, mining engineering plan data, and mine ventilation system distribution data to construct a static dataset. A multi-parameter coupled prediction model construction module is used to construct a multi-parameter coupled prediction model using the coal seam firmness factor and gas emission factor and the static dataset as input features; The model training and validation module is used to train multi-parameter coupled prediction models using machine learning methods and to validate the models using cross-validation. The real-time prediction and early warning module is used to deploy the trained multi-parameter coupled prediction model to the early warning intelligence analysis platform to realize real-time prediction of coal and gas outburst risks at the working face, and automatically trigger early warnings according to the set early warning thresholds.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed, it implements the intelligent dynamic prediction method for coal and gas outburst hazard areas in working faces as described in any one of claims 1-5.

8. An electronic device, characterized in that, include: At least one processor; A memory that is communicatively connected to the at least one processor; The memory stores a computer program that, when executed by the at least one processor, enables the at least one processor to implement the intelligent dynamic prediction method for coal and gas outburst hazard areas in working faces as described in any one of claims 1-5.

Images