Digital twinborn model and method for coal mine disaster scene based on multi-modal data

Abstract

The invention provides a digital twinborn model and method for a coal mine disaster scene based on multi-modal data. The method comprises the following steps: acquiring original data in a mine; performing cleaning, normalization and de-noising preprocessing on the original data; fusing different types of preprocessed original data by using a data fusion algorithm, and constructing a comprehensive digital twinborn model; establishing a virtual mirror image of the mine by using a digital twinning technology, and mapping the mine in the real world into a digital twinning model; real-time mapping and simulation analysis are carried out based on the established digital twinborn model, and the disaster development trend is predicted by monitoring the internal environment and the equipment state of the mine; and according to a model analysis result, optimizing prevention and control measures of coal mine disasters, establishing a real-time monitoring and early warning system of the disasters, and taking measures in time when abnormal conditions are found. The system comprises a data acquisition module, a model construction module and a model analysis module. The pertinence and effectiveness of coal mine disaster prevention and control can be improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a digital twin model and method for coal mine disaster scenarios based on multimodal data. Background Art

[0002] Coal mine disasters refer to accidents or disasters that occur during the production process of coal mines, including gas explosions, coal dust explosions, roof accidents and water seepage accidents. These disasters often lead to serious consequences such as casualties, damage to mine facilities, and waste of resources. The significance of preventing and controlling coal mine disasters is mainly reflected in the following aspects: ensuring personnel safety, protecting mine facilities, saving resources and maintaining social stability. In summary, preventing and controlling coal mine disasters is of great significance in ensuring personnel safety, protecting mine facilities, saving resources and maintaining social stability. However, the existing technology has poor accuracy in predicting the trend of disaster occurrence, and the pertinence and effectiveness of disaster prevention and control are poor.

[0003] Prior art 1, application number: CN 202310077185.5 discloses a risk monitoring method, device, electronic device and storage medium for coal mine disasters, obtains initial production data and initial static data of the area to be detected, processes the initial production data and initial static data, obtains disaster precursor information corresponding to multiple disaster types in the area to be detected, determines a coal mine disaster early warning processing model according to the mine condition information of the area to be detected, and performs joint analysis and processing on multiple disaster precursor information according to the coal mine disaster early warning processing model to obtain comprehensive disaster risk information of the area to be detected. Although it is possible to jointly analyze the predicted data of multiple disasters in the area to be detected, before the disaster occurs, the possibility and harm of the disaster are evaluated based on the obtained disaster hidden dangers and precursor information, and the real-time and accurate early warning of coal mine disasters is realized, which effectively improves the safety of coal mine production. However, there is a lack of accurate judgment on the trend of disaster occurrence, and the effectiveness of disaster prevention and control cannot be effectively improved.

[0004] Prior art 2, application number: CN 202310928019.1 discloses a multi-dimensional control method for coal mine disasters, which evaluates, warns, governs and assesses the type and degree of disasters in the exploration, design, pre-mining, mining and post-mining stages of the mine's entire life cycle; judges whether there are primary disasters in the mining process, whether it will cause secondary and derivative disasters in mining, and regenerative disasters after mining, and judges whether the regenerative disasters in the post-mining stage will generate new disasters as a disaster source. Although a multi-dimensional control method for coal mine disasters using the above steps is used to identify major coal mine disasters through important steps and key links in the entire life cycle of the mine, multi-dimensional control of major disasters is carried out according to the source and degree of disasters, thereby establishing a generalizable and replicable coal mine major disaster prevention and control model. However, the disaster prevention and control measures are not optimized, and the data processing speed is slow.

[0005] Prior art three, application number: CN 202410165967.9 discloses a method for the coordinated fusion and hierarchical prediction of heterogeneous information of typical disaster risks in coal mines, including the following steps: Step 1, establish a multi-parameter indicator system for typical disaster risks in coal mines to identify characteristic parameters of typical disaster risks, and screen key characteristic parameters of typical disaster risks based on a time-series restricted Boltzmann machine model and perform feature extraction; Step 2, construct a typical disaster risk data fusion model and a multi-source information coordinated fusion database; Step 3, construct a multi-parameter perception hierarchical prediction model for typical disaster risks based on convolutional neural networks and long short-term memory networks, and train and evaluate the model based on the collected historical data; Step 4, construct an intelligent linkage and decision-making platform for typical disaster risks based on big data technology. Although the environmental status of coal mines is quickly predicted and classified, the accuracy of coal mine disaster type discrimination is improved. However, no accurate judgment is made on the progress and trend of disasters, which reduces the effect of disaster prevention and control.

[0006] At present, the existing technologies 1, 2 and 3 lack technical means to predict the development trend of disasters, and are unable to optimize disaster prevention and control measures, which reduces the pertinence and effectiveness of coal mine disaster prevention and control, and increases the casualties and economic losses caused by disasters. Therefore, the present invention provides a digital twin model and method for coal mine disaster scenarios based on multimodal data. Summary of the invention

[0007] In order to solve the above technical problems, the present invention provides a digital twin model analysis method for coal mine disaster scenarios based on multimodal data, comprising the following steps:

[0008] Collect raw data in the mine, including environmental monitoring data, equipment operation data and personnel behavior data; pre-process the raw data;

[0009] The pre-processed raw data of different types are fused using data fusion algorithms to build a comprehensive digital twin model. The digital twin technology is used to build a virtual mirror image of the mine and map the mine in the real world into the digital twin model.

[0010] Based on the established digital twin model, real-time mapping and simulation analysis are carried out to predict the development trend of disasters; the prevention and control measures for coal mine disasters are optimized, and a real-time monitoring and early warning system for disasters is established.

[0011] Preferably, the environmental monitoring data includes temperature, humidity and gas concentration, the equipment operation data includes sensor data and equipment status, and the personnel behavior data includes personnel location and walking trajectory.

[0012] Preferably, a data set of environmental monitoring data, equipment operation data and personnel behavior data in historical mines is collected, and the data set is divided into a training set and a validation set.

[0013] Preferably, the preprocessing process of the raw data comprises the following steps:

[0014] Several environmental monitoring devices, equipment operation monitoring systems and personnel behavior monitoring devices are installed inside the coal mine to collect environmental data, equipment operation data and personnel behavior data inside the mine;

[0015] Clean the collected raw data, remove outliers, and then normalize the raw data of different types and dimensions; then denoise the raw data;

[0016] Get the preprocessed raw data and extract the target features; create data sets according to the different types of raw data as identifiers.

[0017] Preferably, the average value of each type of historical raw data is calculated, the average value is the critical value of abnormal data, and the value of the raw data in each acquisition cycle is compared with the critical value of abnormal data. When the value is less than or equal to the critical value of abnormal data and greater than or equal to 1 / 2 of the critical value, the raw data is marked as an abnormal value; when the value is less than 1 / 2 of the critical value of abnormal data, the raw data is marked as a normal value.

[0018] Preferably, the process of building a digital twin model includes the following steps:

[0019] Align the preprocessed raw data according to timestamps and spatial locations, associate equipment operation data with personnel behavior data, and analyze the interaction between multiple variables; generate fused data using weighted average of environmental monitoring data, equipment operation data, and personnel behavior data;

[0020] According to the actual situation of the mine, a physical model is established, which includes various physical processes in the mine, air flow, temperature field and humidity field; the fused data is analyzed to obtain the rules in the fused data, and machine learning is used to train the data-driven model;

[0021] The parameters of the physical model are calibrated and adjusted with those of the data-driven model. The physical model and the data-driven model are combined to build a digital twin model, which provides predictions and optimization suggestions for the operating status of the mine.

[0022] Preferably, the process of generating fused data by weighted averaging comprises the following steps:

[0023] Determine the data sources that need to be integrated, including environmental monitoring data, equipment operation data, and personnel behavior data;

[0024] Different weights are given to each data source based on reliability and importance. The weights are determined based on the number of devices that collect environmental monitoring data, equipment operation data, and personnel behavior data.

[0025] For each timestamp and spatial location, the raw data of each data source is weighted averaged according to the weight, that is, the value of each data source is multiplied by its corresponding weight, all products are added up, and divided by the sum of the weights.

[0026] Preferably, the process of calibrating and adjusting the parameters in the digital twin model comprises the following steps:

[0027] Define a mean square error objective function to measure the difference between the output of the digital twin model and the actual observation result; initialize the parameters in the digital twin model to an initial value;

[0028] Using the initialized parameters, the output of the model is calculated through the digital twin model; the output of the digital twin model is compared with the real observation results, and the difference between them, i.e., the error, is calculated;

[0029] Based on the error, the back propagation algorithm is used to calculate the gradient of the model parameters, and gradient descent is used to update the parameter values ​​of the digital twin model according to the gradient of the parameters until the error reaches the predetermined number of iterations.

[0030] The present invention provides a digital twin model of a coal mine disaster scene based on multimodal data, comprising:

[0031] The data acquisition module is responsible for collecting the raw data in the mine, including environmental monitoring data, equipment operation data and personnel behavior data; cleaning, normalizing and denoising the raw data;

[0032] The model building module is responsible for fusing different types of pre-processed raw data using data fusion algorithms to build a comprehensive digital twin model. It uses digital twin technology to build a virtual mirror image of the mine and map the mine in the real world to the digital twin model.

[0033] The model analysis module is responsible for real-time mapping and simulation analysis based on the established digital twin model to predict the development trend of disasters; optimize the prevention and control measures for coal mine disasters and establish a real-time monitoring and early warning system for disasters.

[0034] Preferably, the environmental monitoring data includes temperature, humidity and gas concentration, the equipment operation data includes sensor data and equipment status, and the personnel behavior data includes personnel location and walking trajectory;

[0035] A dataset of environmental monitoring data, equipment operation data, and personnel behavior data from historical mines is collected and divided into a training set and a validation set.

[0036] The present invention can ensure the accuracy and reliability of the input data of the digital twin model. The information of different data sources is integrated together to improve the comprehensiveness and integration of the model. Potential disaster risks can be discovered in advance, and prevention and control measures can be taken in time to reduce the occurrence of disasters. It can improve the pertinence and effectiveness of coal mine disaster prevention and control, reduce casualties and economic losses caused by disasters; it can realize real-time monitoring and early warning of the internal environment and equipment conditions of the mine, predict disaster trends in advance, provide scientific basis for decision makers, optimize prevention and control measures, and improve the level of coal mine safety management.

[0037] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings.

[0038] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0040] Figure 1 This is a flow chart of a digital twin model analysis method for coal mine disaster scenarios based on multimodal data in Example 1 of the present invention;

[0041] Figure 2 This is a diagram of the preprocessing process of raw data in Example 2 of the present invention;

[0042] Figure 3 A process diagram of mapping a mine in the real world into a digital twin model in Example 3 of the present invention;

[0043] Figure 4 This is a process diagram for constructing a digital twin model in Example 4 of the present invention;

[0044] Figure 5 This is a process diagram of generating fused data by weighted averaging in Embodiment 5 of the present invention;

[0045] Figure 6 A process diagram of combining a physical model and a data-driven model in Embodiment 6 of the present invention;

[0046] Figure 7This is a process diagram for calibrating and adjusting parameters in a digital twin model in Example 7 of the present invention;

[0047] Figure 8 A process diagram of real-time mapping and simulation analysis in Example 8 of the present invention;

[0048] Fig. 9 This is a block diagram of a digital twin model of a coal mine disaster scenario based on multimodal data in Example 9 of the present invention. DETAILED DESCRIPTION

[0049] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0050] The terms used in the embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to limit the embodiments of the present application. The singular forms of "a", "said" and "the" used in the embodiments of the present application and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0051] When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. On the contrary, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the attached claims. In the description of the present application, it should be understood that the terms "first", "second", "third", etc. are only used to distinguish similar objects, and do not have to be used to describe a specific order or sequence, nor can they be understood as indicating or implying relative importance. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to the specific circumstances.

[0052] Example 1: Figure 1 As shown, an embodiment of the present invention provides a digital twin model analysis method for coal mine disaster scenarios based on multimodal data, comprising the following steps:

[0053] S100: Collecting raw data including environmental monitoring data, equipment operation data and personnel behavior data in the mine; cleaning, normalizing and denoising the raw data;

[0054] Among them, environmental monitoring data includes temperature, humidity, and gas concentration, equipment operation data includes sensor data and equipment status, and personnel behavior data includes personnel location and walking trajectory;

[0055] S200: Use data fusion algorithms to fuse different types of pre-processed raw data to build a comprehensive digital twin model; use digital twin technology to build a virtual mirror image of the mine and map the mine in the real world into the digital twin model;

[0056] The data set of environmental monitoring data, equipment operation data and personnel behavior data in historical mines is collected and divided into a training set and a validation set;

[0057] S300: Based on the established digital twin model, real-time mapping and simulation analysis are carried out to predict the development trend of disasters by monitoring the internal environment of the mine and the status of equipment. According to the results of the model analysis, the prevention and control measures of coal mine disasters are optimized, and a real-time monitoring and early warning system for disasters is established to detect abnormal situations and take timely measures.

[0058] The working principle and beneficial effects of the above technical solution are as follows: First, the original data including environmental monitoring data, equipment operation data and personnel behavior data in the mine are collected in this embodiment; the original data is cleaned, normalized and pre-processed for denoising; secondly, the original data after pre-processing of different types are fused using a data fusion algorithm to build a comprehensive digital twin model; the virtual mirror of the mine is established using digital twin technology, and the mine in the real world is mapped to the digital twin model; finally, based on the established digital twin model, real-time mapping and simulation analysis are performed, and the development trend of disasters is predicted by monitoring the internal environment and equipment status of the mine; according to the model analysis results, the prevention and control measures of coal mine disasters are optimized, a real-time monitoring and early warning system for disasters is established, abnormal situations are discovered, and measures are taken in time. Step S100 of the above scheme can ensure the accuracy and reliability of the input data of the digital twin model. Step S200 integrates information from different data sources to improve the comprehensiveness and comprehensiveness of the model. Step S300 discovers potential disaster risks in advance, takes prevention and control measures in time, and reduces the occurrence of disasters.

[0059] To sum up, through the above steps, this embodiment can improve the pertinence and effectiveness of coal mine disaster prevention and control, reduce casualties and economic losses caused by disasters; it can realize real-time monitoring and early warning of the internal environment and equipment conditions of the mine, predict disaster trends in advance, provide scientific basis for decision makers, optimize prevention and control measures, and improve the level of coal mine safety management.

[0060] Example 2: Figure 2 As shown, based on Example 1, the preprocessing process of the raw data provided by the embodiment of the present invention includes the following steps:

[0061] S101: Install a number of environmental monitoring equipment, equipment operation monitoring systems and personnel behavior monitoring equipment inside the coal mine to collect environmental data, equipment operation data and personnel behavior data inside the mine;

[0062] S102: Clean the collected raw data, remove outliers, and then normalize the raw data of different types and dimensions; then perform denoising on the raw data;

[0063] The average value of each type of historical raw data is calculated, and the average value is the critical value of abnormal data. The value of the raw data in each collection cycle is compared with the critical value of abnormal data. When the value is less than or equal to the critical value of abnormal data and greater than or equal to 1 / 2 of the critical value, the raw data is marked as an abnormal value; when the value is less than 1 / 2 of the critical value of abnormal data, the raw data is marked as a normal value.

[0064] S103: Obtain the preprocessed raw data and extract target features; create a data set according to different types of raw data as identifiers.

[0065] The working principle and beneficial effects of the above technical scheme are as follows: In this embodiment, a number of environmental monitoring equipment, equipment operation monitoring system and personnel behavior monitoring equipment are first set up inside the coal mine to collect environmental data, equipment operation data and personnel behavior data in the mine; secondly, the collected raw data is cleaned, outliers are removed, and raw data of different types and dimensions are normalized; then the raw data is denoised; wherein, the average value of each type of historical raw data is calculated, and the average value is the critical value of the abnormal data. The value of the raw data in each acquisition cycle is compared with the critical value of the abnormal data. When the value is less than or equal to the critical value of the abnormal data and greater than or equal to 1 / 2 of the critical value, the raw data is marked as an abnormal value; when the value is less than 1 / 2 of the critical value of the abnormal data, the raw data is marked as a normal value; finally, the pre-processed raw data is obtained, and the target features are extracted; and a data set is made according to the different types of raw data as identifiers. Step S101 of the above scheme obtains various data inside the coal mine, including relevant information on the environment, equipment and personnel. Step S102: The cleaning process removes outliers to eliminate errors or outliers in the data, making the data more accurate and reliable. Normalization converts data of different types and dimensions into a unified range to facilitate subsequent data processing and analysis. De-noising can reduce noise interference in the data and improve the quality and credibility of the data. Step S103: Target feature extraction is to extract key features related to specific tasks or problems from the original data for subsequent model training and analysis. Making data sets is to integrate the processed data into data sets according to different types to facilitate subsequent data analysis and modeling.

[0066] To sum up, this embodiment improves the accuracy and reliability of data, reduces noise interference in data, and makes data more standardized and easier to process; the preprocessed data can be better applied to fields such as data analysis, modeling and decision-making, which helps to improve the safety, production efficiency and management level of coal mines.

[0067] Example 3: Figure 3 As shown, based on Example 1, the process of mapping a mine in the real world into a digital twin model provided by the embodiment of the present invention includes the following steps:

[0068] S201: Using data fusion algorithm, using data association, interaction and spatiotemporal relationship to fuse the pre-processed raw data, and using the fused raw data to build a digital twin model;

[0069] S202: Calibrate and adjust the parameters in the digital twin model based on real-world mine data and situations;

[0070] S203: Use the validation set to validate the digital twin model and compare the consistency between the output results of the digital twin model and the actual observation results.

[0071] The working principle and beneficial effects of the above technical solution are as follows: First, the present embodiment uses a data fusion algorithm to fuse the preprocessed raw data using data association, interaction and spatiotemporal relationship, and uses the fused raw data to build a digital twin model; secondly, the parameters in the digital twin model are calibrated and adjusted according to the mine data and context in the real world; finally, the digital twin model is verified using a validation set to compare the consistency between the output results of the digital twin model and the actual observation results. Step S201 of the above solution fuses multimodal data such as environmental monitoring data, equipment operation data and personnel behavior data together to obtain more comprehensive and accurate information, providing a more reliable basis for subsequent analysis and prediction. Step S202 can ensure that the digital twin model can accurately reflect the characteristics and behaviors of the real mine through calibration and adjustment, and improve the credibility and prediction accuracy of the model. Step S203 evaluates the accuracy and reliability of the digital twin model. Through verification, the availability of the digital twin model can be determined, and the model can be further optimized and improved.

[0072] In summary, this embodiment establishes an accurate and reliable digital twin model based on real-world data, which can accurately reflect the environment and equipment status of the mine; provides a feasible analysis and prediction tool for optimizing the prevention and control measures of coal mine disasters and establishing a real-time monitoring and early warning system; through the digital twin model, the development trend of disasters can be effectively predicted, and timely measures can be taken to improve the safety and production efficiency of mines.

[0073] Example 4: Figure 4 As shown, based on Example 3, the process of building a digital twin model provided by the embodiment of the present invention includes the following steps:

[0074] S2011: Align the pre-processed raw data according to timestamps and spatial positions, associate the equipment operation data with the personnel behavior data, and analyze the interaction between multiple variables; generate fused data by weighted average of the environmental monitoring data, equipment operation data, and personnel behavior data;

[0075] S2012: Based on the actual situation of the mine, a physical model is established, which includes various physical processes in the mine, such as air flow, temperature field and humidity field. The fused data is analyzed to obtain the rules in the fused data, and machine learning is used to train the data-driven model.

[0076] S2013: Calibrate and adjust the parameters of the physical model with the parameters of the data-driven model, combine the physical model and the data-driven model to build a digital twin model, which provides predictions and optimization suggestions for the mine operation status.

[0077] The working principle and beneficial effects of the above technical solution are as follows: This embodiment first aligns the pre-processed raw data according to timestamps and spatial positions, associates equipment operation data with personnel behavior data, and analyzes the interaction between multiple variables; uses weighted average to generate fused data for environmental monitoring data, equipment operation data, and personnel behavior data; secondly, according to the actual situation of the mine, a physical model is established, and the physical model includes various physical processes in the mine, such as air flow, temperature field, and humidity field; the fused data is analyzed to obtain the rules in the fused data, and machine learning is used to train the data-driven model; finally, the parameters of the physical model are calibrated and adjusted with the parameters of the data-driven model, and the physical model and the data-driven model are combined to construct a digital twin model, which provides predictions and optimization suggestions for the operation status of the mine. Step S2011 of the above solution obtains more accurate and comprehensive data to provide for subsequent steps, integrates data from different sources, eliminates data noise and inconsistency, and provides a reliable data foundation for building a digital twin model. Step S2012 extracts key features and rules from the data to build a model to simulate and predict the physical processes in the mine, which can simulate the actual situation in the mine and provide an accurate physical basis for the digital twin model. Step S2013 combines the physical model with the data-driven model, and uses the mutual complementation and synergy between the models to improve the accuracy and reliability of the digital twin model; provide decision support for mine operations, optimize operational efficiency, reduce accident risks, and improve safety and economic benefits.

[0078] In summary, through the above steps, this embodiment can use the fused raw data to build a digital twin model, thereby realizing the modeling and simulation of mines in the real world; the digital twin model can be used to optimize mine operations, predict faults and optimize decisions, which has important practical significance.

[0079] Example 5: Figure 5 As shown, based on Example 4, the process of generating fused data by weighted average provided in the embodiment of the present invention includes the following steps:

[0080] S20111: Determine the data sources that need to be integrated, including environmental monitoring data, equipment operation data, and personnel behavior data;

[0081] S20112: For each data source, different weights are given according to reliability and importance. The weights are determined according to the number of devices that collect environmental monitoring data, equipment operation data, and personnel behavior data;

[0082] S20113: For each timestamp and spatial location, the original data of each data source is weighted averaged according to the weight, that is, the value of each data source is multiplied by its corresponding weight, all products are added up, and divided by the sum of the weights.

[0083] The working principle and beneficial effects of the above technical solution are as follows: This embodiment first determines the data sources that need to be fused, including environmental monitoring data, equipment operation data and personnel behavior data; secondly, for each data source, different weights are given according to reliability and importance, and the weights are determined according to the number of devices that collect environmental monitoring data, equipment operation data and personnel behavior data; finally, for each timestamp and spatial position, the original data of each data source is weighted averaged according to the weight, that is, the value of each data source is multiplied by its corresponding weight, all products are added up, and divided by the sum of the weights. Step S20111 of the above solution determines the data sources that need to be fused to ensure that key sources of information are included. Step S20112 assigns weights to different data sources according to actual conditions to ensure that more reliable and important data have a greater impact on the fusion results. Step S20113 fuses the information of different data sources together to generate fused data, making the fused data more comprehensive and representative.

[0084] To sum up, when constructing a digital twin model in this embodiment, the accuracy and reliability of the model can be improved by fusing information from different data sources; weighted average generation of fused data can consider the contributions of different data sources and combine them, so that the digital twin model can more comprehensively reflect the actual situation and provide more accurate predictions and optimization suggestions.

[0085] Example 6: Figure 6As shown, based on Example 4, the process of combining the physical model and the data-driven model provided in the embodiment of the present invention includes the following steps:

[0086] S20131: Use the established physical model to simulate the system and obtain the output of the physical model, which may be the system status, parameters and predicted values; at the same time, collect actual data for use in the data-driven model;

[0087] S20132: Use the output of the physical model as the input of the data-driven model. During the model integration process, provide the results generated by the physical model and the actual data to the data-driven model; use the data set of the integrated physical model output to train parameters for the data-driven model and optimize the model; during the training process, the data-driven model will learn the complex relationships and patterns of the integrated data;

[0088] S20133: The data-driven model training is completed and verified for prediction and analysis. At this point, the output of the physical model is combined with the prediction results of the data-driven model to obtain a digital twin model, which predicts and optimizes the mine operation status.

[0089] The working principle and beneficial effects of the above technical solution are as follows: First, the present embodiment uses the established physical model to simulate the system and obtain the output result of the physical model, which may be the state, parameters and predicted values ​​of the system; at the same time, actual data is collected for use by the data-driven model; secondly, the output of the physical model is used as the input of the data-driven model, and in the process of model integration, the results generated by the physical model and the actual data are provided to the data-driven model; the data set output by the integrated physical model is used to train parameters for the data-driven model and optimize the model; in the training process, the data-driven model will learn the complex relationships and patterns of the integrated data; finally, the data-driven model training is completed and verified for prediction and analysis. At this time, the output of the physical model is combined with the predicted results of the data-driven model to obtain a digital twin model, which predicts and optimizes the operating status of the mine. Step S20131 of the above solution obtains the output results and actual data of the physical model. Step S20132 combines the physical model with the data-driven model, and improves the accuracy and reliability of the model through learning and optimization of the data-driven model. Step S20133 integrates the physical model and the data-driven model to provide more accurate prediction results and optimization suggestions to help decision makers make better decisions.

[0090] In summary, this embodiment combines the physical model and the data-driven model, making full use of the knowledge of the physical model and the learning ability of the data-driven model to improve the accuracy and reliability of the model. Through the digital twin model, the operating status of the system can be better understood and analyzed, and prediction and optimization can be performed to improve the efficiency and safety of the system, which can make up for the shortcomings of the physical model and the data-driven model, making the model more comprehensive and reliable.

[0091] Example 7: Figure 7 As shown, based on Example 3, the process of calibrating and adjusting the parameters in the digital twin model provided in the embodiment of the present invention includes the following steps:

[0092] S2021: Define a mean square error objective function to measure the difference between the output of the digital twin model and the actual observation result; initialize the parameters in the digital twin model to an initial value;

[0093] S2022: Using the initialized parameters, the output result of the model is calculated through the digital twin model; the output result of the digital twin model is compared with the real observation result, and the difference between them, i.e., the error, is calculated;

[0094] S2023: Based on the error, use the back propagation algorithm to calculate the gradient of the model parameters, and use gradient descent to update the parameter values ​​of the digital twin model according to the gradient of the parameters until the error reaches a predetermined number of iterations.

[0095] The working principle and beneficial effects of the above technical solution are as follows: This embodiment first defines a mean square error objective function to measure the difference between the output result of the digital twin model and the actual observation result; initializes the parameters in the digital twin model to an initial value; secondly, uses the initialized parameters to calculate the output result of the model through the digital twin model; compares the output result of the digital twin model with the real observation result, and calculates the difference between them, that is, the error; finally, according to the error, the gradient of the model parameters is calculated using the back propagation algorithm, and the parameter value of the digital twin model is updated according to the gradient of the parameter using gradient descent until the error reaches a predetermined number of iterations. Step S2021 of the above scheme measures the difference between the output result of the digital twin model and the actual observation result. The objective function can provide a numerical indicator for evaluating the accuracy and precision of the model and determining a measurable standard for guiding the process of parameter adjustment. Step S2022 can determine the error size of the model by comparing the model output with the real observation result, determine the performance of the model under the current parameter setting, and provide a reference for the next step of parameter adjustment. Step S2023 determines the degree of influence of each parameter on the error. By calculating the gradient, we can understand the contribution of each parameter change to the error and determine the direction of parameter adjustment. The technical effect of using the gradient descent method to update the model parameters is to adjust the parameters in the direction of reducing the error, so that the model is closer to the actual observation results; by iteratively updating the parameters, the model is gradually optimized so that it can more accurately predict the real-world situation.

[0096] In summary, this embodiment continuously calibrates and adjusts the parameters of the digital twin model to enable it to more accurately simulate real-world mine scenarios; by comparing the consistency of the model output results with the actual observation results, the accuracy and reliability of the digital twin model can be verified; it can be used to predict and optimize mine operations and improve production efficiency and safety.

[0097] Example 8: Figure 8 As shown, based on Example 1, the process of real-time mapping and simulation analysis provided by the embodiment of the present invention includes the following steps:

[0098] S301: input the collected real-time data into the digital twin model and match it with the virtual environment and device status in the digital twin model;

[0099] S302: Based on the real-time mapped data, simulation analysis and prediction are carried out. The digital twin model predicts the development trend of coal mine disasters, including geological disasters, gas explosions and fires, based on historical data and current data;

[0100] S303: The digital twin model performs calculations to obtain the possibility and impact of the disaster.

[0101] The working principle and beneficial effects of the above technical solution are as follows: First, the real-time data collected is input into the digital twin model in this embodiment, and matched with the virtual environment and equipment status in the digital twin model; secondly, simulation analysis and prediction are performed based on the real-time mapped data, and the digital twin model predicts the development trend of coal mine disasters, including geological disasters, gas explosions and fires, etc., according to historical data and current data. Finally, the digital twin model performs calculations to obtain the possibility and impact of the disaster. The matching of real-time data with the virtual environment and equipment status in step S301 of the above solution can ensure the accuracy and authenticity of the digital twin model. Step S302 Through simulation analysis and prediction, the digital twin model can provide the trend and possibility of the development of coal mine disasters, and help predict and warn related disaster events. Step S303 The calculation of the digital twin model can quantify the possibility and impact of coal mine disasters, provide a scientific basis for decision-making, and help formulate corresponding early warning measures and safety recommendations.

[0102] In summary, this embodiment can timely monitor the environment and equipment status inside the mine, predict the development trend of coal mine disasters, and provide corresponding warnings and suggestions through real-time mapping and simulation analysis; it helps to improve the safety and production efficiency of coal mines, reduce the occurrence of disasters and accidents, and protect the lives and property interests of miners. At the same time, the application of digital twin models can also optimize mine operation management and improve the production efficiency and sustainable development of coal mines.

[0103] Example 9: Fig. 9 As shown, on the basis of Embodiment 1 to Embodiment 8, the digital twin model of coal mine disaster scene based on multimodal data provided by the embodiment of the present invention includes:

[0104] The data acquisition module is responsible for collecting the raw data in the mine, including environmental monitoring data, equipment operation data and personnel behavior data; cleaning, normalizing and denoising the raw data;

[0105] Among them, environmental monitoring data includes temperature, humidity, and gas concentration, equipment operation data includes sensor data and equipment status, and personnel behavior data includes personnel location and walking trajectory;

[0106] The model building module is responsible for fusing different types of pre-processed raw data using data fusion algorithms to build a comprehensive digital twin model. It uses digital twin technology to build a virtual mirror image of the mine and map the mine in the real world to the digital twin model.

[0107] The data set of environmental monitoring data, equipment operation data and personnel behavior data in historical mines is collected and divided into a training set and a validation set;

[0108] The model analysis module is responsible for real-time mapping and simulation analysis based on the established digital twin model, and predicts the development trend of disasters by monitoring the internal environment of the mine and the status of equipment. According to the model analysis results, it optimizes the prevention and control measures of coal mine disasters, establishes a real-time monitoring and early warning system for disasters, discovers abnormal situations, and takes timely measures.

[0109] The working principle and beneficial effects of the above technical solution are as follows: the data acquisition module of this embodiment collects raw data including environmental monitoring data, equipment operation data and personnel behavior data in the mine; the raw data is cleaned, normalized and denoised; wherein the environmental monitoring data includes temperature, humidity and gas concentration, the equipment operation data includes sensor data and equipment status, and the personnel behavior data includes personnel location and walking trajectory; the model construction module fuses different types of pre-processed raw data using a data fusion algorithm to construct a comprehensive digital twin model; the digital twin technology is used to establish a virtual mirror of the mine, and the mine in the real world is mapped to the digital twin model; wherein a data set of environmental monitoring data, equipment operation data and personnel behavior data in historical mines is collected, and the data set is divided into a training set and a verification set; the model analysis module performs real-time mapping and simulation analysis based on the established digital twin model, and predicts the development trend of disasters by monitoring the internal environment and equipment status of the mine; according to the model analysis results, the prevention and control measures for coal mine disasters are optimized, a real-time monitoring and early warning system for disasters is established, and abnormal situations are discovered and timely measures are taken. The data acquisition module of the above scheme cleans, normalizes and denoises the raw data to ensure the accuracy and availability of the data; through this module, high-quality processed data can be obtained to provide a reliable data basis for subsequent model construction and analysis. The model construction module integrates different types of pre-processed raw data to build a comprehensive digital twin model; through digital twin technology, the mine in the real world is mapped into the virtual environment to achieve real-time monitoring and analysis of the mine. The model analysis module performs real-time mapping and simulation analysis based on the established digital twin model; by monitoring the internal environment and equipment status of the mine, the development trend of coal mine disasters, including geological disasters, gas explosions and fires, is predicted. At the same time, according to the results of the model analysis, the prevention and control measures of coal mine disasters are optimized, and a real-time monitoring and early warning system for disasters is established to detect abnormal situations in a timely manner and take measures to minimize the occurrence and impact of disasters.

[0110] In summary, these modules in this embodiment improve the level and efficiency of coal mine safety management. Through the digital twin model, real-time monitoring and prediction of the internal environment and equipment status of the mine can be achieved, potential disaster risks can be discovered in advance, and corresponding prevention and control measures can be taken to ensure the safety of miners and reduce economic losses caused by disasters. In addition, the digital twin model can also be used to simulate and optimize the operation and management strategies of the mine, and improve the production efficiency and resource utilization efficiency of the mine.

[0111] Embodiment 10: Based on Embodiment 9, the data acquisition module provided in the embodiment of the present invention comprises:

[0112] The data collection submodule is responsible for setting up a number of environmental monitoring equipment, equipment operation monitoring systems and personnel behavior monitoring equipment inside the coal mine to collect environmental data, equipment operation data and personnel behavior data inside the mine;

[0113] The data processing submodule is responsible for cleaning the collected raw data, removing outliers, and normalizing the raw data of different types and dimensions; and then denoising the raw data;

[0114] The average value of each type of historical raw data is calculated, and the average value is the critical value of abnormal data. The value of the raw data in each collection cycle is compared with the critical value of abnormal data. When the value is less than or equal to the critical value of abnormal data and greater than or equal to 1 / 2 of the critical value, the raw data is marked as an abnormal value; when the value is less than 1 / 2 of the critical value of abnormal data, the raw data is marked as a normal value.

[0115] The data output submodule is responsible for obtaining the preprocessed raw data and extracting the target features; it uses the different types of raw data as identifiers to create data sets.

[0116] The working principle and beneficial effects of the above technical solution are as follows: the data acquisition submodule of this embodiment sets up a number of environmental monitoring equipment, equipment operation monitoring systems and personnel behavior monitoring equipment inside the coal mine to collect environmental data, equipment operation data and personnel behavior data in the mine; the data processing submodule cleans the collected raw data, removes outliers, and then normalizes the raw data of different types and dimensions; then denoises the raw data; wherein, the average value of each type of historical raw data is calculated, and the average value is the critical value of the abnormal data. The value of the raw data in each acquisition cycle is compared with the critical value of the abnormal data. When the value is less than or equal to the critical value of the abnormal data and greater than or equal to 1 / 2 of the critical value, the raw data is marked as an abnormal value; when the value is less than 1 / 2 of the critical value of the abnormal data, the raw data is marked as a normal value; the data output submodule obtains the preprocessed raw data and extracts the target features; and a data set is produced according to the different types of raw data as identifiers. The data acquisition submodule of the above scheme can set up various monitoring equipment inside the coal mine to collect environmental data, equipment operation data and personnel behavior data in real time, and can obtain real-time data inside the coal mine, providing a data basis for the subsequent establishment and analysis of the digital twin model. The data processing submodule cleans, normalizes and denoises the collected raw data. Through cleaning, outliers and noise can be removed to improve the quality and accuracy of the data; through normalization, data of different types and dimensions can be unified into the same range, which is convenient for subsequent data fusion and analysis; through denoising, the accuracy and reliability of the data can be further improved. The data output submodule extracts the target features from the preprocessed raw data and creates a data set according to the data type, which can provide input data for the subsequent construction of the digital twin model. At the same time, the extracted target features can be used to establish the input features of the model, further improving the accuracy and effect of the model.

[0117] In summary, this embodiment provides high-quality multimodal data, and provides a reliable data foundation for the establishment and analysis of digital twin models of coal mine disaster scenarios based on multimodal data. Through real-time monitoring and analysis of the internal environment, equipment, and personnel behavior of coal mines, the development trend of disasters can be predicted, and the prevention and control measures for coal mine disasters can be optimized, a real-time monitoring and early warning system can be established, and timely measures can be taken to improve the safety and efficiency of coal mines.

[0118] Embodiment 11: Based on Embodiment 9, the model building module provided in the embodiment of the present invention comprises:

[0119] The data fusion submodule is responsible for using the data fusion algorithm to fuse the pre-processed raw data using data association, interaction and spatiotemporal relationship, and building a digital twin model using the fused raw data;

[0120] The parameter calibration submodule is responsible for calibrating and adjusting the parameters in the digital twin model based on real-world mine data and situations;

[0121] The model verification submodule is responsible for verifying the digital twin model using the verification set and comparing the consistency between the output results of the digital twin model and the actual observation results.

[0122] The working principle and beneficial effects of the above technical solution are as follows: the data fusion submodule of this embodiment uses the data fusion algorithm to fuse the preprocessed raw data using data association, interaction and spatiotemporal relationship, and uses the fused raw data to build a digital twin model; the parameter calibration submodule calibrates and adjusts the parameters in the digital twin model according to the mine data and context in the real world; the model verification submodule verifies the digital twin model using the verification set, and compares the consistency between the output results of the digital twin model and the actual observation results. The data fusion submodule of the above scheme fuses the preprocessed raw data, and by considering the association, interaction and spatiotemporal relationship between the data, the comprehensive information volume of the data is improved, so that the constructed digital twin model is more accurate and comprehensive. The parameter calibration submodule calibrates and adjusts the parameters in the digital twin model according to the mine data and context in the real world, so that the model can better reflect the real situation and improve the accuracy and reliability of the model. The model verification submodule verifies the digital twin model using the verification set, and by comparing the consistency between the model output results and the actual observation results, the prediction ability and credibility of the model are evaluated, and the trust in the model is further enhanced.

[0123] In summary, this embodiment improves the accuracy and reliability of the model, so that the model can be better applied in actual scenarios, providing effective decision support and early warning capabilities for coal mine safety management. At the same time, through the construction and verification of the digital twin model, the behavior and change laws of the coal mine system can be better understood and explained, providing reference and guidance for optimizing coal mine operations.

[0124] Embodiment 12: Based on Embodiment 9, the model analysis module provided in the embodiment of the present invention comprises:

[0125] The content matching submodule is responsible for inputting the collected real-time data into the digital twin model and matching it with the virtual environment and device status in the digital twin model;

[0126] The trend prediction submodule is responsible for simulation analysis and prediction based on real-time mapped data. The digital twin model predicts the development trend of coal mine disasters, including geological disasters, gas explosions and fires, based on historical data and current data.

[0127] The model calculation submodule is responsible for calculating the digital twin model to obtain the possibility and impact of disasters.

[0128] The working principle and beneficial effects of the above technical solution are as follows: the content matching submodule of this embodiment inputs the collected real-time data into the digital twin model, and matches it with the virtual environment and equipment status in the digital twin model; the trend prediction submodule performs simulation analysis and prediction based on the real-time mapped data, and the digital twin model predicts the development trend of coal mine disasters, including geological disasters, gas explosions and fires, etc. according to historical data and current data; the model calculation submodule digital twin model performs calculations to obtain the possibility and impact of disasters. The content matching submodule of the above scheme matches the real-time collected data with the virtual environment and equipment status in the digital twin model to achieve simulation and comparison of the actual situation. The trend prediction submodule uses the digital twin model for simulation analysis and prediction based on the real-time mapped data, and predicts the development trend of coal mine disasters, including geological disasters, gas explosions and fires, etc. through historical data and current data, which helps to give early warning and take corresponding countermeasures to reduce disaster risks. The model calculation submodule uses the digital twin model to calculate and obtain the possibility and impact of disasters. By calculating various parameters and conditions in the model, the potential risks and possible consequences of disasters can be evaluated. This helps guide mine managers and related personnel to develop effective safety measures and emergency response plans.

[0129] In summary, the entire model analysis module of this embodiment provides a prediction and evaluation of the trend of coal mine disasters through the combination of digital twin models and real-time data, which can help mine managers take measures in advance to prevent and reduce the occurrence of disasters and accidents, and ensure the safety of miners and the normal operation of mines. At the same time, the model analysis module can also provide a reference for emergency response and improve the ability and effect of responding to disasters and accidents.

[0130] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A digital twin model analysis method for coal mine disaster scenarios based on multimodal data, characterized in that: The following steps are involved: Collect raw data in the mine, including environmental monitoring data, equipment operation data and personnel behavior data; pre-process the raw data; The pre-processed raw data of different types are fused using data fusion algorithms to build a comprehensive digital twin model. The digital twin technology is used to build a virtual mirror image of the mine and map the mine in the real world into the digital twin model. Based on the established digital twin model, real-time mapping and simulation analysis are carried out to predict the development trend of disasters; Optimize the prevention and control measures for coal mine disasters and establish a real-time monitoring and early warning system for disasters.

2. The digital twin model analysis method for coal mine disaster scenarios based on multimodal data according to claim 1, characterized in that: in, Environmental monitoring data includes temperature, humidity and gas concentration, equipment operation data includes sensor data and equipment status, and personnel behavior data includes personnel location and walking trajectory.

3. The digital twin model analysis method for coal mine disaster scenarios based on multimodal data according to claim 1, characterized in that: A dataset of environmental monitoring data, equipment operation data, and personnel behavior data from historical mines is collected and divided into a training set and a validation set.

4. The digital twin model analysis method for coal mine disaster scenarios based on multimodal data according to claim 1, characterized in that: The preprocessing process of raw data includes the following steps: Several environmental monitoring devices, equipment operation monitoring systems and personnel behavior monitoring devices are installed inside the coal mine to collect environmental data, equipment operation data and personnel behavior data inside the mine; Clean the collected raw data, remove outliers, and then normalize the raw data of different types and dimensions; then denoise the raw data; Get the preprocessed raw data and extract the target features; Create data sets based on the different types of original data.

5. The digital twin model analysis method for coal mine disaster scenarios based on multimodal data according to claim 4, characterized in that: in, Calculate the average value of each type of historical raw data. The average value is the critical value of abnormal data. Compare the value of the raw data in each collection cycle with the critical value of abnormal data. When the value is less than or equal to the critical value of abnormal data and greater than or equal to 1 / 2 of the critical value, mark the raw data as an abnormal value. When the value is less than 1 / 2 of the critical value of abnormal data, the original data is marked as normal.

6. The digital twin model analysis method for coal mine disaster scenarios based on multimodal data according to claim 1, characterized in that: The process of building a digital twin model includes the following steps: Align the preprocessed raw data according to timestamps and spatial locations, associate equipment operation data with personnel behavior data, and analyze the interaction between multiple variables; use weighted average to generate fused data for environmental monitoring data, equipment operation data, and personnel behavior data; According to the actual situation of the mine, a physical model is established, which includes various physical processes in the mine, air flow, temperature field and humidity field; Analyze the fused data, obtain the patterns in the fused data, and use machine learning to train the data-driven model; The parameters of the physical model are calibrated and adjusted with those of the data-driven model, and the physical model and the data-driven model are combined to build a digital twin model, which provides predictions and optimization suggestions for the operating status of the mine.

7. The digital twin model analysis method for coal mine disaster scenarios based on multimodal data according to claim 6, characterized in that: The process of generating fused data using weighted averaging includes the following steps: Determine the data sources that need to be integrated, including environmental monitoring data, equipment operation data, and personnel behavior data; Different weights are given to each data source based on reliability and importance. The weights are determined based on the number of devices that collect environmental monitoring data, equipment operation data, and personnel behavior data. For each timestamp and spatial location, the raw data of each data source is weighted averaged according to the weight, that is, the value of each data source is multiplied by its corresponding weight, all products are added up, and divided by the sum of the weights.

8. The digital twin model analysis method for coal mine disaster scenarios based on multimodal data according to claim 1, characterized in that: The process of calibrating and adjusting the parameters in the digital twin model includes the following steps: Define the mean square error objective function to measure the difference between the output of the digital twin model and the actual observation results; Initialize the parameters in the digital twin model to an initial value; Using the initialized parameters, the output of the model is calculated through the digital twin model; Compare the output of the digital twin model with the real observations and calculate the difference between them, i.e. the error; Based on the error, the back propagation algorithm is used to calculate the gradient of the model parameters, and gradient descent is used to update the parameter values ​​of the digital twin model according to the gradient of the parameters until the error reaches the predetermined number of iterations.

9. A digital twin model of coal mine disaster scenarios based on multimodal data, characterized in that: Include: The data acquisition module is responsible for collecting the raw data in the mine, including environmental monitoring data, equipment operation data and personnel behavior data; cleaning, normalizing and denoising the raw data; The model building module is responsible for fusing different types of pre-processed raw data using data fusion algorithms to build a comprehensive digital twin model. It uses digital twin technology to build a virtual mirror image of the mine and map the mine in the real world to the digital twin model. The model analysis module is responsible for real-time mapping and simulation analysis based on the established digital twin model to predict the development trend of disasters; Optimize the prevention and control measures for coal mine disasters and establish a real-time monitoring and early warning system for disasters.

10. The digital twin model of coal mine disaster scenarios based on multimodal data according to claim 9, characterized in that: in, Environmental monitoring data includes temperature, humidity and gas concentration, equipment operation data includes sensor data and equipment status, and personnel behavior data includes personnel location and walking trajectory; A dataset of environmental monitoring data, equipment operation data, and personnel behavior data from historical mines is collected and divided into a training set and a validation set.

Images