Digital twinning driven intelligent sensing method for underground coal mine ventilation system

Abstract

The invention discloses a digital twinning driven coal mine underground ventilation system intelligent sensing method, which comprises the following steps: constructing a digital twinning platform to realize virtual and real synchronous updating; the method specifically comprises the steps of building real-time communication between a real roadway and a virtual-real roadway; constructing a roadway space information intelligent sensing model under the condition that the number of sensors is limited based on Gaussian process regression; the method specifically comprises the steps of real data collection, covariance function design, covariance matrix construction and Gaussian process regression calculation. And performing cloud picture display through the color space model to obtain an intuitive view of the space information. The method is based on Gaussian process regression and is combined with a digital twinning method, roadway space information can be comprehensively and intelligently sensed under the condition that the underground environment of a coal mine is limited, synchronous updating of a virtual model is achieved through a real-time data transmission protocol, and real-time monitoring of the roadway space information is achieved. The problem that a roadway blank area cannot be effectively monitored due to the fact that the number of sensors in an underground coal mine is limited can be solved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent ventilation in underground coal mines, and in particular to an intelligent sensing method for underground coal mine ventilation systems driven by digital twins. Background Art

[0002] Coal is an important basic energy source in my country, and it occupies a dominant position in the energy production and consumption structure, and has abundant reserves. However, most of my country's coal resources are buried deep underground, and underground mining methods are required during the mining process. In underground coal mining, the main function of mine ventilation is to provide fresh air to various locations underground, provide oxygen to meet the needs of underground workers, dilute and remove toxic and harmful gases and dust underground, improve the working environment, and regulate the temperature. It is a necessary condition for safe production in coal mines.

[0003] At present, most mines still rely on traditional manual methods and semi-automatic equipment to measure, monitor and manage underground ventilation systems. However, the above methods not only have problems such as blind spots in monitoring, low efficiency, large number of personnel, and insufficient intelligence, but also make it difficult to find and deal with temporary problems in the ventilation system in a timely and effective manner, resulting in many safety hazards.

[0004] Digital twin technology provides a new solution to the current problem. By building a digital model consistent with the physical entity in a virtual environment, the simulation, monitoring and prediction of the tunnel environment can be realized. However, in practical applications, how to fully obtain spatial node information is still a major scientific challenge that needs to be solved urgently. Summary of the invention

[0005] In view of the above-mentioned technical deficiencies, the purpose of the present invention is to provide a digital twin-driven intelligent perception method for coal mine underground ventilation systems, which can solve the problem that there are blank areas in coal mine underground tunnel monitoring and comprehensive perception information cannot be obtained.

[0006] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0007] The present invention provides a digital twin-driven intelligent sensing method for a coal mine underground ventilation system, comprising the following steps:

[0008] Step 1: Build a digital twin platform to achieve synchronous update of virtual and real, including building a physical tunnel in the coal mine, building a corresponding digital twin virtual tunnel model, and synchronizing the data of the physical tunnel and virtual tunnel model through a real-time communication protocol;

[0009] Step 2: Based on Gaussian process regression, an intelligent perception model of tunnel spatial information is constructed. Under the condition of limited number of sensors, a covariance function is designed through the collected tunnel environment data, a covariance matrix is ​​constructed, and the spatial information interpolation of the area without sensor deployment is calculated;

[0010] Step 3: The spatial information obtained in step 2 is converted into a visual cloud map through the color space model and displayed in real time in the digital twin virtual laneway model.

[0011] Preferably, the step 1 specifically includes:

[0012] S11: Build a scaled-down physical tunnel based on the real tunnel environment in the coal mine;

[0013] S12: Build a digital twin virtual laneway model based on the constructed physical laneway, and divide the virtual laneway model into spatial nodes;

[0014] S13: Establish Socket real-time communication, including building Socket server and client, designing data transmission protocol, and realizing real-time data stream transmission between physical lane and virtual lane model through asynchronous IO and multi-threading.

[0015] Preferably, the step 2 specifically includes:

[0016] S21: Deploy wind speed sensors and gas sensors underground in coal mines to collect tunnel environment data during ventilation;

[0017] S22: Integrate sensor data through a multi-sensor serial communication device; Integrate sensor data through a multi-sensor serial communication device, which automatically identifies each sensor according to the port number and device address, performs preliminary verification and format unification on the collected wind speed and gas concentration data, and removes invalid or abnormal data.

[0018] S23: Design a covariance function, using a multiplication combination form of a square exponential kernel, the covariance function is:

[0019]

[0020] where k(x,x′) is the covariance function, σ 1 and σ 2 is the amplitude scale, l 1 and l 2 is the length scale, x represents the data observation point, and x′ represents the point where spatial information needs to be calculated;

[0021] S24: constructing a covariance matrix according to the covariance function, and calculating the covariance relationship between the known data points and the unknown interpolation points;

[0022] S25: Based on the Gaussian process regression conditional distribution formula, the spatial information interpolation of the area where sensors are not deployed is calculated, including the predicted mean and the predicted variance.

[0023] Preferably, steps S24-S25 specifically include:

[0024] The covariance matrix is ​​constructed based on the covariance function. For a given spatial node X = {x 1 ,x 2 ,...,x n The covariance matrix K is expressed as:

[0025]

[0026] Calculate the target space node information, as follows:

[0027] The Gaussian process is defined as f(x)~GP(m(x),k(x,x′)), where m(x) is the mean function and k(x,x′) is the covariance function. In the regression problem, the goal is to obtain the Gaussian process based on the observed data. Interpolation calculation point x * Information f(x * );

[0028] Assume that the input y of the observation point i is the potential function f(x i ) plus noise:

[0029]

[0030] in is the variance of the observation noise;

[0031] At this time, for the training data X and the interpolation calculation point x * , the corresponding function values ​​f(x) and f(x * ) is:

[0032]

[0033] where K(X,X) is the covariance matrix between the training data points, K(X,x * ) is the covariance matrix between the training data points and the new data points, K(x * ,x * ) is the covariance matrix of the new data points;

[0034] According to the conditional distribution properties of Gaussian distribution, the interpolation distribution of new points is derived:

[0035] f(x * )|X,y~N(μ(x * ),σ 2 (x * ))(5)

[0036] in:

[0037]

[0038] Where μ(x * ) is the predicted mean, representing the possible data of the function value, σ 2 (x * ) is the uncertainty of the prediction, which reflects the confidence of the model in the new point.

[0039] Preferably, the step 3 specifically includes:

[0040] S31: normalizing the interpolated lane environment data to a hue value, and setting the saturation and brightness of the color space model to fixed values;

[0041] S32: Convert the HSV color space model to the RGB color space model available to Unity;

[0042] S33: Map the RGB color data to the mesh material of the digital twin virtual roadway model to generate a real-time updated visualization cloud map.

[0043] The beneficial effects of the present invention are:

[0044] 1. Improve the accuracy and comprehensiveness of tunnel environmental information monitoring. Under the condition of a limited number of sensors, the interpolation calculation with the help of Gaussian process regression method can effectively solve the monitoring gap problem caused by the limited distribution of sensors in coal mines, so as to obtain more comprehensive tunnel environmental information. By combining the covariance function with the data collected by existing sensors, the wind speed, gas concentration and other information in the area where sensors are not deployed can be accurately predicted, which greatly improves the accuracy of monitoring data and provides more reliable data support for safe production in coal mines.

[0045] 2. By building a digital twin platform to achieve virtual and real synchronous updates, and reflect the dynamic changes of the tunnel environment in real time, managers can intuitively view the distribution of information such as wind speed and gas concentration in the tunnel in the virtual tunnel model, and promptly discover problems in the ventilation system and make adjustments. At the same time, the digital twin model can predict the future state of the ventilation system based on historical data and real-time monitoring data, providing a decision-making basis for the intelligent management of the ventilation system.

[0046] 3. The introduction of the color space HSV model displays the tunnel environment information in the form of a visual cloud map in the digital twin virtual tunnel model, so that managers can quickly and intuitively grasp the distribution of environmental information in the tunnel. Once an abnormal situation such as excessive gas concentration occurs, it can be discovered in time and corresponding measures can be taken, effectively reducing the risk of accidents in coal mines and improving the efficiency and timeliness of safety management.

[0047] 4. The intelligent perception model of tunnel space information built based on Gaussian process regression can efficiently process and analyze the large amount of collected data. The model establishes a covariance matrix and calculates interpolation to explore the potential laws behind the data, providing a powerful tool for the prediction and analysis of tunnel environmental information, which helps to gain a deeper understanding of the operating characteristics of the underground ventilation system in coal mines. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0049] Figure 1 is a digital twin diagram of a lane provided by an embodiment of the present invention;

[0050] Figure 2 is a diagram of a Gaussian process return method provided by an embodiment of the present invention;

[0051] Figure 3 is a Gaussian process regression interpolation result diagram provided by an embodiment of the present invention;

[0052] Figure 4 It is a Gaussian process regression interpolation three-dimensional surface graph provided by an embodiment of the present invention;

[0053] Figure 5 It is a digital twin diagram of a lane provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0054] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0055] like Figure 1 to Figure 2 As shown, this embodiment provides a digital twin-driven intelligent perception method for a coal mine underground ventilation system, comprising the following steps:

[0056] Step 1: Build a digital twin platform to achieve synchronous update of virtual and real, including the following contents:

[0057] S11. Use U-shaped supports to build a real tunnel that is smaller than the actual tunnel in the coal mine;

[0058] S12. Build a digital twin virtual laneway model based on the real laneway (see Figure 1 ),

[0059] And perform spatial node division;

[0060] S13. Establish Socket real-time communication to realize data stream transmission between the physical object and the virtual model, as follows:

[0061] Build a Socket server, which is responsible for listening to the port and waiting for connection requests from the client;

[0062] Build a Socket client, which is responsible for connecting to the server and sending requests or data;

[0063] Design the data transmission protocol as TCP communication protocol and the data transmission format as JSON format to ensure that both parties can correctly understand and transmit data;

[0064] Set up asynchronous communication and multithreading. Socket communication is usually a blocking operation. In order to improve efficiency, asynchronous IO is needed to handle Socket communication.

[0065] Step 2: Based on Gaussian process regression, an intelligent perception model for perceiving complete lane spatial information is constructed under the condition of limited number of sensors; the details are as follows:

[0066] S21. Deploy wind speed and gas sensors according to the restrictions in the coal mine to collect tunnel environment information during ventilation;

[0067] S22, integrating sensor data through a multi-sensor serial communication device;

[0068] S23. Design a covariance function based on the degree of data smoothness. The covariance function can be combined in many ways, using multiplication. The specific form is as follows:

[0069]

[0070] where k(x,x′) is the covariance function, σ 1 and σ 2 is the amplitude scale, l 1 and l 2 is the length scale, x represents the data observation point, and x′ represents the point where spatial information needs to be calculated;

[0071] S24, construct a covariance matrix according to the covariance function, for a given sensor space node X = {x 1 ,x 2 ,x 3 ,x 4 The covariance matrix K can be expressed as:

[0072]

[0073] S25. Calculate the target space node information, as follows:

[0074] The Gaussian process is defined as f(x)~GP(m(x),k(x,x′)), where m(x) is the mean function and k(x,x′) is the covariance function. In regression problems, the goal is to Interpolation calculation point x * Information f(x * );

[0075] Assume that the input y of the observation point i is the potential function f(x i ) plus noise:

[0076]

[0077] in is the variance of the observation noise;

[0078] At this time, for the training data X and the interpolation calculation point x * , the corresponding function values ​​f(x) and f(x * ) is:

[0079]

[0080] where K(X,X) is the covariance matrix between the training data points, K(X,x * ) is the covariance matrix between the training data points and the new data points, K(x * ,x * ) is the covariance matrix of the new data points;

[0081] According to the conditional distribution properties of Gaussian distribution, the interpolation distribution of new points can be derived:

[0082] f(x * )|X,y~N(μ(x * ),σ 2 (x * ))(15)

[0083] in:

[0084]

[0085] Where μ(x * ) is the predicted mean, representing the possible data of the function value, σ 2 (x * ) is the uncertainty of the prediction, which reflects the confidence of the model in the new point;

[0086] Step 3: Display the cloud map through the color space model to obtain an intuitive view of the spatial information. The specific contents are as follows:

[0087] S31, using the data obtained by Gaussian process regression interpolation as hue, setting the saturation and brightness of the color space model to 1, and the hue to the normalized lane environment data;

[0088] S32. Convert the HSV color space model into RGB color available to Unity;

[0089] S33, obtaining the mesh of the virtual lane, and transferring the RGB color data to the mesh color material.

[0090] Figure 3-Figure 5 The final result of this embodiment is shown. Since the use of gas is involved, the gas is toxic and harmful, and is flammable and explosive, and is dangerous to use in experimental situations, so alcohol is used instead, and all the following gas concentrations are replaced by alcohol concentrations.

[0091] like Figure 3 As shown, it can be seen that Gaussian process regression interpolation calculates all spatial nodes of the lane division, and obtains the wind speed and gas concentration information of each node, and further distinguishes the size by different colors;

[0092] like Figure 4 As shown in Figure 2, it can be seen that through Gaussian process regression interpolation, all 316 nodes in the lane space are calculated and a three-dimensional surface diagram is drawn, which is consistent with Figure 3 The comparison revealed that the overall trend was the same;

[0093] like Figure 5 As shown, the interpolated data is normalized and then passed to the mesh color material to further display the digital twin model.

[0094] Under the condition of limited number of sensors, this method realizes comprehensive intelligent perception of lane environment information through Gaussian process regression method. The color space HSV model is introduced to realize synchronous update of digital twin virtual lanes through real-time data transmission, and the lane environment information is intuitively displayed.

[0095] 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-driven intelligent perception method for underground coal mine ventilation system, characterized in that: The following steps are involved: Step 1: Build a digital twin platform to achieve synchronous update of virtual and real, including building a physical tunnel in the coal mine, building a corresponding digital twin virtual tunnel model, and synchronizing the data of the physical tunnel and virtual tunnel model through a real-time communication protocol; Step 2: Based on Gaussian process regression, an intelligent perception model of tunnel spatial information is constructed. Under the condition of limited number of sensors, a covariance function is designed through the collected tunnel environment data, a covariance matrix is ​​constructed, and the spatial information interpolation of the area without sensor deployment is calculated; Step 3: The spatial information obtained in step 2 is converted into a visual cloud map through the color space model and displayed in real time in the digital twin virtual laneway model.

2. The method for intelligent sensing of underground coal mine ventilation system driven by digital twins according to claim 1, characterized in that: The step 1 specifically includes: S11: Build a scaled-down physical tunnel based on the real tunnel environment in the coal mine; S12: Build a digital twin virtual laneway model based on the constructed physical laneway, and divide the virtual laneway model into spatial nodes; S13: Establish Socket real-time communication, including building Socket server and client, designing data transmission protocol, and realizing real-time data stream transmission between physical lane and virtual lane model through asynchronous IO and multi-threading.

3. The method for intelligent sensing of underground coal mine ventilation system driven by digital twins according to claim 1, characterized in that: The step 2 specifically includes: S21: Deploy wind speed sensors and gas sensors underground in coal mines to collect tunnel environment data during ventilation; S22: Integrate sensor data through a multi-sensor serial communication device; S23: Design a covariance function, using a multiplication combination form of a square exponential kernel, the covariance function is: Where k(x,x′) is the covariance function, σ1 and σ2 are amplitude scales, l1 and l2 are length scales, x represents the data observation point, and x′ represents the point where spatial information needs to be calculated; S24: constructing a covariance matrix according to the covariance function, and calculating the covariance relationship between the known data points and the unknown interpolation points; S25: Based on the Gaussian process regression conditional distribution formula, the spatial information interpolation of the area where sensors are not deployed is calculated, including the predicted mean and the predicted variance.

4. The method for intelligent sensing of underground coal mine ventilation system driven by digital twins according to claim 3, characterized in that: Steps S24-S25 specifically include: According to the covariance function, the covariance matrix is ​​constructed. For a given spatial node X = {x1, x2, ..., x n The covariance matrix K is expressed as: Calculate the target space node information, as follows: The Gaussian process is defined as f(x)~GP(m(x),k(x,x′)), where m(x) is the mean function and k(x,x′) is the covariance function. In the regression problem, the goal is to obtain the Gaussian process based on the observed data. Interpolation calculation point x * Information f(x * ); Assume that the input y of the observation point i is the potential function f(x i ) plus noise: in is the variance of the observation noise; At this time, for the training data X and the interpolation calculation point x * , the corresponding function values ​​f(x) and f(x * ) is: where K(X,X) is the covariance matrix between the training data points, K(X,x * ) is the covariance matrix between the training data points and the new data points, K(x * ,x * ) is the covariance matrix of the new data points; According to the conditional distribution properties of Gaussian distribution, the interpolation distribution of new points is derived: f(x * )∣X,y~N(μ(x * ),s 2 (x * ))(5) among them: Where μ(x * ) is the predicted mean, representing the possible data of the function value, σ 2 (x * ) is the uncertainty of the prediction, which reflects the confidence of the model in the new point.

5. The method for intelligent sensing of underground coal mine ventilation system driven by digital twins according to claim 1, characterized in that: The step 3 specifically includes: S31: normalizing the interpolated lane environment data to a hue value, and setting the saturation and brightness of the color space model to fixed values; S32: Convert the HSV color space model to the RGB color space model available to Unity; S33: Map the RGB color data to the mesh material of the digital twin virtual roadway model to generate a real-time updated visualization cloud map.

Images