A coal rock fracture trajectory dynamic identification and real-time prediction method

Abstract

The application provides a coal rock fracture track dynamic identification and real-time prediction method. The method establishes a coal rock internal and external fracture development track identification model, tracks the coal rock internal and external fracture development track in real time, and realizes the dynamic identification of the fracture development track combined with the coal rock internal and external fractures. A real-time self-correction prediction model of the coal rock internal and external fractures is constructed, the prediction results of the coal rock internal and external fracture development track are corrected in real time, and the dynamic forward prediction of the coal rock internal and external fracture development path is realized. The coal rock fracture identification and prediction model constructed by the application can master the coal rock mass fracture development in real time, and the self-correction prediction model has the characteristics of wide applicability, high reliability and high precision, and is suitable for coal rock fracture development in different prediction scenes. Based on the real-time identification and self-correction prediction model of the coal rock internal and external fractures, the coal rock stability can be determined and predicted in time.

Description

Technical Field

[0001] This invention relates to the field of rock mechanics technology, and in particular to a method for dynamic identification and real-time prediction of coal and rock fracture trajectories. Background Technology

[0002] In underground engineering projects such as tunnel construction and coal mining, geotechnical engineering disasters caused by the expansion and connection of rock fissures occur frequently. Most of the instability and failure of geotechnical engineering is related to the expansion and connection of fissures, mainly manifested as the initiation, expansion and connection of rock fissures, which further leads to macroscopic rupture phenomena.

[0003] To address this, scholars employed digital image correlation technology (DICT) to study the crack propagation and deformation processes of coal and rock samples under load. They analyzed the initiation, propagation, and interconnection of external cracks in coal and rock. While significant progress has been made in rock crack research, current studies largely focus on identifying and predicting the development of internal or external cracks in single coal and rock samples, failing to simultaneously observe both internal and external cracks. Since internal and external cracks often influence each other, both must be considered in practical engineering. Furthermore, observations of the development of pre-existing cracks in rock specimens under load have revealed differences in the development of internal and external cracks. Currently, few studies effectively combine the development processes of internal and external cracks. Moreover, achieving real-time prediction under large-scale real-time data still faces many challenges. Different working conditions or engineering requirements may necessitate different levels of model prediction accuracy; however, existing prediction models typically only predict under specific working conditions and lack self-correction mechanisms. This means that the accuracy and precision of prediction models are limited across different scenarios, and the prediction results often fail to meet practical engineering needs. To address these challenges, it is necessary to propose more accurate and stable identification and prediction models, and further modify and optimize the prediction models for different operating conditions.

[0004] Therefore, there is an urgent need for a dynamic identification method that combines the development trajectory of internal and external fractures in coal and rock. Summary of the Invention

[0005] The purpose of this invention is to provide a method for dynamic identification and real-time prediction of coal and rock fracture trajectories, so as to solve the problems existing in the prior art.

[0006] The technical solution adopted to achieve the purpose of this invention is as follows: a method for dynamic identification and real-time prediction of coal and rock fracture trajectories, comprising the following steps:

[0007] 1) Grid division of the sample surface. n acoustic emission probes are evenly arranged on the sidewall of the coal and rock sample, and a high-speed camera is arranged on the outside of the coal and rock sample.

[0008] 2) The acquisition period T is divided into multiple acquisition steps according to the time step Δt. During the mechanical experiment, acoustic emission equipment and a high-speed camera are used to collect the propagation process of internal fractures and the propagation trajectory of external fractures in the coal and rock within the acquisition period T. The coordinates of internal fractures at times t, t+1, t+2, t+3, ..., t+n are obtained, thereby establishing a dataset of internal fracture tip coordinate changes with temporal characteristics. (t)= The coordinates of the external fracture at times t, t+1, t+2, t+3, ..., t+n are obtained, thus establishing a dataset W(t) of external fracture tip coordinate changes with temporal characteristics. The data acquisition and calculation process for a single time step includes the following sub-steps:

[0009] a) Acoustic emission information during the coal and rock fracturing process is collected using acoustic emission equipment. The development trajectory of internal fractures in the coal and rock at time t is identified. Acoustic emission technology is used to record the acoustic emission information throughout the entire process of internal fracture development in the coal and rock. Then, the internal fractures are identified using a sound source localization method. Based on the acoustic emission localization technology, the fracture trajectory is inverted to determine the spatial location of the internal fracture development in the coal and rock sample.

[0010] (b) A high-speed camera is used to record the series of processes of external fracture development, and the fracture trajectory images are extracted. The trajectory of external fracture development in coal and rock at time t is identified. An intelligent image recognition algorithm is used to identify external fractures in coal and rock. The extracted coal and rock fracture images are binarized to separate the fractures from the background. The trajectory coordinates during fracture development are further extracted to obtain data on the change of external fracture location over time.

[0011] c) Fusion of the development trajectories of internal and external fractures in coal and rock at time t. Based on the internal and external coordinates at time t extracted in steps a) and b), the internal and external fracture points are matched using a point-to-point matching method to achieve real-time fusion of internal and external fractures.

[0012] 3) The dataset of coordinate changes at the tips of internal and external fractures extracted in step 2) is used as a historical dataset for learning to obtain a fracture tip coordinate prediction model. This model is then used to predict the development trajectory of internal and external fractures. By predicting the fracture tip coordinates at future moments, the dynamic extension process of internal and external fractures at the next moment is predicted, and the predicted coordinate values ​​of the dynamic development process of internal and external fractures are output. and .

[0013] 4) The prediction step size is continuously adjusted based on the prediction results of the internal and external fracture prediction model, and the model is updated in real time to achieve forward self-correction prediction of the development trajectory of internal and external fractures in coal and rock.

[0014] 5) Identify and predict the development trajectory of internal and external fractures. Based on the predicted coordinates of internal and external fractures obtained in steps 3) and 4), the internal and external fracture locations are matched using a point-to-point matching method to achieve time-series-based fusion of internal and external fracture locations. This yields the predicted forward fracture trajectory.

[0015] Furthermore, in step 1), the mechanical experiment is a tensile failure mechanical experiment.

[0016] Furthermore, in step 1), n ​​≥ 4.

[0017] Furthermore, step a) specifically includes the following sub-steps:

[0018] a1) Establish a spatial rectangular coordinate system with the center point of the coal and rock sample as the origin, and obtain the spatial coordinates (x, y, z) of the n sensors. n y n , z n The spatial coordinates corresponding to the fracture development path are marked as (x i y i , z i The time interval of the sound source signals received by the n sensors is denoted as t. n Let t be the time it takes for the signal to be emitted from the acoustic emission source. Equation (1) is established based on the time difference positioning method.

[0019] (1)

[0020] Solving equation (1) yields several sets of coordinate points. The geometric centers of these coordinate points are then determined. These geometric centers are the spatial coordinates (x, y) corresponding to the desired fracture development path. i y i , z i ).

[0021] a2) Extract the changes in internal fracture coordinates during fracture development and establish a spatial coordinate dataset. ={(x1,y1,z1) (x2,y2,z2) (x3,y3,z3)……(x n y n , z n )}.

[0022] a3) By updating the spatial coordinates of the coal and rock fracture tips in real time, a dataset of internal fracture tip coordinate changes with temporal characteristics is established. (t)= .

[0023] Furthermore, step b) specifically includes the following sub-steps:

[0024] b1) Record the trajectory image of external fracture development. Among them, the center point of the coal and rock specimen is selected as the origin of the coordinate system to draw a plane coordinate system.

[0025] b2) Perform binarization processing on the image of the coal and rock external fracture development process obtained in step b1).

[0026] b3) Extract the coordinates of the external fracture tip. Based on the fracture image after binarization in step b2), scan line by line, sequentially traversing the pixel values ​​corresponding to any point (x, y) in the fracture image. and stored in set W i In the middle, i takes the values ​​0, 1, 2, ..., n.

[0027] (2)

[0028] The tip of the crack exhibits a change in pixel value. The first point of threshold change is defined as the crack tip point, i.e., satisfying... The corresponding (x,y) is denoted as the coordinates of the crack tip.

[0029] b4) Repeat the tip search step in b3) to obtain a series of processes in the development of external fractures in coal and rock, and further derive the coordinate changes of the trajectory during fracture development, denoted as Ω={(x1, y1) (x2, y2) (x3, y3)……(x n y n )}.

[0030] b5) By updating the coordinates of the external fracture tips in coal and rock in real time, a dataset of temporal changes in the coordinates of the external fracture tips is established, denoted as W(t) = .

[0031] Furthermore, in step c), the internal coordinates reflecting the morphology and size of the fracture are combined with the external coordinates reflecting the displacement and deformation of the fracture. For the external coordinates corresponding to a certain time t, combined with the corresponding internal fracture coordinates, the fracture development trajectory of the coal and rock specimen at time t is further inverted. Based on the extracted internal and external coordinates at time t, the internal and external fracture points are matched using a point-to-point matching method to achieve real-time fusion of internal and external fractures. A three-dimensional fracture model is formed based on the fusion result, which more accurately identifies the location and morphology of internal and external fractures.

[0032] Furthermore, in step 3), the extracted dataset of coordinate changes of the tips of internal and external fractures in coal and rock is used for training to obtain a fracture development trajectory prediction model. By using this model to predict the coordinates of the tips of internal and external fractures at future moments, the development trajectory of internal and external fractures in coal and rock can be predicted. The prediction model includes a Long Short-Term Memory (LSTM) network.

[0033] Furthermore, in step 4), the real-time self-correction method collects and monitors fracture development data in real time and establishes a corresponding feedback mechanism. During implementation, based on the research accuracy, a corresponding error range threshold is set for the prediction model. When the model's predicted value exceeds the error threshold, the model needs to be self-corrected to improve the prediction accuracy.

[0034] The technical effects of this invention are beyond doubt:

[0035] A. Combining intelligent image processing technology and acoustic emission monitoring technology, a method for identifying and predicting the development trajectory of internal and external fractures in coal and rock is proposed, which is beneficial for mastering the development trajectory of internal and external fractures in coal and rock.

[0036] B. Based on the identification of the development trajectory of coal and rock mass fractures, this invention constructs a prediction model for the development trajectory of coal and rock mass fractures, and adopts a self-correcting method to continuously adjust the prediction step size according to the prediction results of the model and update the model in real time to improve the accuracy of the prediction results and realize the forward prediction of the development trajectory of coal and rock mass fractures.

[0037] C. By identifying the initiation, expansion, and connection paths of internal and external cracks in coal and rock, the coal and rock crack identification and prediction model constructed in this invention can monitor the development of internal and external cracks in coal and rock in real time, which is beneficial for timely identification and prediction of coal and rock stability. Attached Figure Description

[0038] Figure 1 Here is a flowchart of the prediction method;

[0039] Figure 2 The diagram shows the structure of the LSTM coordinate prediction network. Detailed Implementation

[0040] The present invention will be further described below with reference to embodiments, but it should not be construed that the scope of the present invention is limited to the following embodiments. Various substitutions and modifications made based on ordinary technical knowledge and common practices in the art without departing from the above-described technical concept of the present invention should be included within the scope of protection of the present invention.

[0041] Example 1:

[0042] Fracturing in rock masses typically extends in three dimensions. Identifying a single external fracture in coal or rock cannot reflect the development process of internal fractures, and identifying and predicting fractures only within the coal or rock mass cannot provide a direct view of the development path of external fractures. Therefore, there is an urgent need for methods to identify the development trajectory of coal and rock fractures, enabling the identification of the expansion and connection trajectories of external and internal fractures in coal and rock.

[0043] This embodiment provides a method for dynamic identification and real-time prediction of coal and rock fracture trajectories, including the following steps:

[0044] 1) Grid the sample surface. Distribute n acoustic emission probes evenly on the sidewall of the coal and rock sample, and place a high-speed camera on the outside of the sample. n ≥ 4.

[0045] 2) The acquisition period T is divided into multiple acquisition steps according to the time step Δt. During the mechanical experiment, acoustic emission equipment and a high-speed camera are used to collect the propagation process of internal fractures and the propagation trajectory of external fractures in the coal and rock within the acquisition period T. The coordinates of internal fractures at times t, t+1, t+2, t+3, ..., t+n are obtained, thereby establishing a dataset of internal fracture tip coordinate changes with temporal characteristics. (t)= The coordinates of the external fracture at times t, t+1, t+2, t+3, ..., t+n are obtained, thus establishing a dataset W(t) of external fracture tip coordinate changes with temporal characteristics. The data acquisition and calculation process for a single time step includes the following sub-steps:

[0046] a) Acoustic emission information during the coal and rock fracturing process is collected using acoustic emission equipment. The trajectory of internal fracture development in the coal and rock at time t is identified. Acoustic emission technology is used to record the acoustic emission information throughout the entire process of internal fracture development in the coal and rock. Then, the internal fractures are identified using a sound source localization method. Based on the acoustic emission localization technology, the fracture trajectory is retrieved, and the spatial location of the internal fracture development in the coal and rock sample is determined. Step a) specifically includes the following sub-steps:

[0047] a1) Establish a spatial rectangular coordinate system with the center point of the coal and rock sample as the origin, and obtain the spatial coordinates (x, y, z) of the n sensors. n y n , z n The spatial coordinates corresponding to the fracture development path are marked as (x i y i , z i The time interval of the sound source signals received by the n sensors is denoted as t. n Let t be the time it takes for the signal to be emitted from the acoustic emission source. Equation (1) is established based on the time difference positioning method.

[0048] (1)

[0049] Solving equation (1) yields several sets of coordinate points. The geometric centers of these coordinate points are then determined. These geometric centers are the spatial coordinates (x, y) corresponding to the desired fracture development path. i y i , z i ).

[0050] a2) Extract the changes in internal fracture coordinates during fracture development and establish a spatial coordinate dataset. ={(x1,y1,z1) (x2,y2,z2) (x3,y3,z3)……(x n y n , z n )}.

[0051] a3) By updating the spatial coordinates of the coal and rock fracture tips in real time, a dataset of internal fracture tip coordinate changes with temporal characteristics is established. (t)= .

[0052] b) A high-speed camera is used to record the series of processes of external fracture development, and the fracture trajectory images are extracted. The trajectory of external fracture development in coal and rock at time t is identified. An intelligent image recognition algorithm is used to identify external fractures in coal and rock. The extracted coal and rock fracture images are binarized to separate the fractures from the background. The trajectory coordinates during fracture development are further extracted to obtain data on the change of external fracture position over time. Step b) specifically includes the following sub-steps:

[0053] b1) Record the trajectory image of external fracture development. Among them, the center point of the coal and rock specimen is selected as the origin of the coordinate system to draw a plane coordinate system.

[0054] b2) Perform binarization processing on the image of the coal and rock external fracture development process obtained in step b1).

[0055] b3) Extract the coordinates of the external fracture tip. Based on the fracture image after binarization in step b2), scan line by line, sequentially traversing the pixel values ​​corresponding to any point (x, y) in the fracture image. and stored in set W i In the middle, i takes the values ​​0, 1, 2, ..., n.

[0056] (2)

[0057] The tip of the crack exhibits a change in pixel value. The first point of threshold change is defined as the crack tip point, i.e., satisfying... The corresponding (x,y) is denoted as the coordinates of the crack tip.

[0058] b4) Repeat the tip search step in b3) to obtain a series of processes in the development of external fractures in coal and rock, and further derive the coordinate changes of the trajectory during fracture development, denoted as Ω={(x1, y1) (x2, y2) (x3, y3)……(x n y n )}.

[0059] b5) By updating the coordinates of the external fracture tips in coal and rock in real time, a dataset of temporal changes in the coordinates of the external fracture tips is established, denoted as W(t) = .

[0060] c) Fusion of the development trajectories of internal and external fractures in the coal and rock at time t. Based on the internal and external coordinates extracted in steps a) and b), the internal and external fracture points are matched using a point-to-point correspondence method to achieve real-time fusion of internal and external fractures. The internal coordinates, reflecting fracture morphology and size information, are combined with the external coordinates, reflecting fracture displacement and deformation information. For the external coordinates corresponding to a specific time t, combined with the corresponding internal fracture coordinates, the fracture development trajectory of the coal and rock specimen at time t is further inverted. Based on the extracted internal and external coordinates at time t, the internal and external fracture points are matched using a point-to-point correspondence method to achieve real-time fusion of internal and external fractures. A three-dimensional fracture model is formed based on the fusion results, allowing for more accurate identification of the location and morphology of internal and external fractures.

[0061] 3) The dataset of coordinate changes at the tips of internal and external fractures extracted in step 2) is used as a historical dataset for learning to obtain a fracture tip coordinate prediction model. This model is then used to predict the development trajectory of internal and external fractures. By predicting the fracture tip coordinates at future moments, the dynamic extension process of internal and external fractures at the next moment is predicted, and the predicted coordinate values ​​of the dynamic development process of internal and external fractures are output. and A fracture development trajectory prediction model is obtained by training the model using the extracted dataset of coordinate changes of the tips of internal and external fractures in coal and rock. This model is then used to predict the coordinates of the tips of internal and external fractures at future moments, thereby predicting the development trajectory of internal and external fractures in coal and rock. The prediction model includes a Long Short-Term Memory (LSTM) network.

[0062] 4) The prediction step size is continuously adjusted based on the prediction results of the internal and external fracture prediction model, and the model is updated in real time to achieve forward self-correcting prediction of the development trajectory of internal and external fractures in coal and rock. The real-time self-correcting method collects and monitors fracture development data in real time and establishes a corresponding feedback mechanism. During implementation, based on the research accuracy, an appropriate error range threshold is set for the prediction model. When the model's predicted value exceeds the error threshold, the model needs to be self-corrected to improve the prediction accuracy.

[0063] 5) Identify and predict the development trajectory of internal and external fractures. Based on the predicted coordinates of internal and external fractures obtained in steps 3) and 4), the internal and external fracture locations are matched using a point-to-point matching method to achieve time-series-based fusion of internal and external fracture locations. This yields the predicted forward fracture trajectory.

[0064] This embodiment achieves real-time fusion of internal and external fractures in coal and rock, and based on the identification results, performs self-correcting prediction of fracture development trajectories to meet the needs of rock mechanics for efficient, real-time prediction models. The coal and rock fracture identification and prediction model constructed in this embodiment can monitor the development of fractures in coal and rock masses in real time. The self-correcting prediction model is characterized by wide applicability, high reliability, and high accuracy, and is applicable in various prediction scenarios. The real-time identification and self-correcting prediction model based on internal and external fractures in coal and rock facilitates timely identification and prediction of coal and rock stability.

[0065] Example 2:

[0066] The main content of this embodiment is the same as that of embodiment 1. The mechanical experiment is an experiment in which cracks on the surface of the specimen can be observed, including but not limited to three-point bending experiment, four-point bending experiment or Brazilian splitting experiment of coal and rock under single coal or different combination conditions.

[0067] Example 3:

[0068] This embodiment provides a method for dynamic identification and real-time prediction of coal and rock fracture trajectories, including the following steps:

[0069] 1) Grid the sample surface; uniformly arrange n acoustic emission probes on the sidewall of the coal and rock sample, and arrange a high-speed camera on the outside of the coal and rock sample;

[0070] 2) The acquisition period T is divided into multiple acquisition steps according to the time step Δt. During the mechanical experiment, acoustic emission equipment and high-speed cameras are used to collect the propagation process of internal fractures in coal and rock and the trajectory of external fractures within the acquisition period T. The coordinates of internal fractures at times t, t+1, t+2, t+3, ..., t+n are obtained, thereby establishing a dataset of internal fracture tip coordinate changes with temporal characteristics. (t)= The coordinates of the external fracture at times t, t+1, t+2, t+3, ..., t+n are obtained, thus establishing a dataset W(t) of external fracture tip coordinate changes with temporal characteristics. The data acquisition and calculation process for a single time step includes the following sub-steps:

[0071] a) Acoustic emission information during the coal and rock fracturing process is collected using acoustic emission equipment; the development trajectory of internal fractures in coal and rock is identified at time t; acoustic emission information of the entire process of internal fracture development in coal and rock is recorded using acoustic emission technology, and then internal fractures in coal and rock are identified by sound source localization method; based on acoustic emission localization technology, the fracture trajectory is inverted to determine the spatial location of internal fracture development in coal and rock sample.

[0072] b) A high-speed camera is used to record a series of processes of external fracture development and the fracture trajectory image is extracted; the trajectory of external fracture development in coal and rock at time t is identified; an intelligent image recognition algorithm is used to identify external fractures in coal and rock; the extracted coal and rock fracture image is binarized to separate the fractures and background in the image, and the trajectory coordinates in the fracture development process are further extracted to obtain data on the change of external fracture position over time.

[0073] c) Fusion of the development trajectories of internal and external fractures in coal and rock at time t; Based on the internal and external coordinates at time t extracted in steps a) and b), the internal and external fracture points are matched using the point-to-point method to achieve real-time fusion of internal and external fractures.

[0074] 3) The dataset of coordinate changes at the tips of internal and external fractures extracted in step 2) is used as a historical dataset for learning to obtain a fracture tip coordinate prediction model. This model is then used to predict the development trajectory of internal and external fractures. The coordinates of the fracture tips at future moments are predicted, enabling the prediction of the dynamic extension process of internal and external fractures at the next moment. The predicted coordinate values ​​of the dynamic development process of internal and external fractures are then output. and ;

[0075] 4) The prediction step size is continuously adjusted based on the prediction results of the internal and external fracture prediction model, and the model is updated in real time to realize the forward self-correction prediction of the development trajectory of internal and external fractures in coal and rock.

[0076] 5) Identify and predict the development trajectory of internal and external fractures; based on the predicted coordinate values ​​of internal and external fractures obtained in steps 3) and 4), use the point-to-point method to match the internal and external fracture points to achieve the fusion of internal and external fracture points based on time sequence; and then obtain the predicted forward fracture trajectory.

[0077] Example 4:

[0078] The main content of this embodiment is the same as that of embodiment 3, except that in step 1), the mechanical experiment is a tensile failure mechanical experiment.

[0079] Example 5:

[0080] The main content of this embodiment is the same as that of embodiment 3, except that in step 1), n≥4.

[0081] Example 6:

[0082] The main content of this embodiment is the same as that of embodiment 3, wherein step a) specifically includes the following sub-steps:

[0083] a1) Establish a spatial rectangular coordinate system with the center point of the coal and rock sample as the origin, and obtain the spatial coordinates (x, y, z) of the n sensors. ny n , z n The spatial coordinates corresponding to the fracture development path are marked as (x i y i , z i The time interval of the sound source signals received by the n sensors is denoted as t. n Let t be the time it takes for the signal to be emitted from the acoustic emission source; establish equation (1) based on the time difference positioning method;

[0084] (1)

[0085] Solving equation (1) yields several sets of coordinate points; the geometric centers of these coordinate points are then determined; these geometric centers are the spatial coordinates (x, y) corresponding to the required fracture development path. i y i , z i );

[0086] a2) Extract the changes in internal fracture coordinates during fracture development and establish a spatial coordinate dataset. ={(x1,y1,z1) (x2,y2,z2) (x3,y3,z3)……(x n y n , z n )};

[0087] a3) By updating the spatial coordinates of the coal and rock fracture tips in real time, a dataset of internal fracture tip coordinate changes with temporal characteristics is established. (t)= .

[0088] Step b) specifically includes the following sub-steps:

[0089] b1) Record the trajectory image of external fracture development; wherein, the center point of the coal and rock specimen is selected as the origin of the coordinate system and a plane coordinate system is drawn;

[0090] b2) Perform binarization processing on the image of the coal and rock external fracture development process obtained in step b1);

[0091] b3) Extract the coordinates of the external fracture tip; based on the fracture image after binarization in step b2), scan line by line, sequentially traversing the pixel values ​​corresponding to any point (x, y) in the fracture image. and stored in set W i In the middle; where i takes the values ​​0, 1, 2, ..., n;

[0092] (2)

[0093] The tip of the crack exhibits a change in pixel value. The first point of threshold change is defined as the crack tip point, i.e., satisfying... The corresponding (x,y) is denoted as the coordinates of the crack tip;

[0094] b4) Repeat the tip search step in b3) to obtain a series of processes in the development of external fractures in coal and rock, and further derive the coordinate changes of the trajectory during fracture development, denoted as Ω={(x1, y1) (x2, y2) (x3, y3)……(x n y n )};

[0095] b5) By updating the coordinates of the external fracture tips in coal and rock in real time, a dataset of temporal changes in the coordinates of the external fracture tips is established, denoted as W(t) = .

[0096] Example 7:

[0097] The main content of this embodiment is the same as that of embodiment 3. In step c), the internal coordinates reflecting the morphology and size of the crack are combined with the external coordinates reflecting the displacement and deformation of the crack. For the external coordinates corresponding to a certain time t, combined with the corresponding internal crack coordinates, the crack development trajectory of the coal and rock specimen at time t is further inverted. Based on the extracted internal and external coordinates at time t, the internal and external crack points are matched using the point-to-point method to achieve real-time fusion of internal and external cracks. A three-dimensional crack model is formed according to the fusion result, which more accurately identifies the location and morphology of internal and external cracks.

[0098] Example 8:

[0099] The main content of this embodiment is the same as that of embodiment 3. In step 3), the extracted dataset of coordinate changes of the tips of internal and external fractures in coal and rock is used for training to obtain a fracture development trajectory prediction model. By using this model to predict the coordinates of the tips of internal and external fractures at future times, the development trajectory of internal and external fractures in coal and rock can be predicted. The prediction model includes a Long Short-Term Memory (LSTM) network.

[0100] Example 9:

[0101] The main content of this embodiment is the same as that of embodiment 3. In step 4), the real-time self-correction method collects and monitors fracture development data in real time and establishes a corresponding feedback mechanism. During the implementation process, based on the research accuracy, a corresponding error range threshold is set for the prediction model. When the model prediction value exceeds the error threshold, the model needs to be self-corrected and adjusted to improve the prediction accuracy.

[0102] Example 10:

[0103] See Figure 1 and Figure 2 The main content of this embodiment is the same as that of Embodiment 1, except that this embodiment includes the following steps:

[0104] 1) Identification of internal fractures in coal and rock. Internal fracture identification is achieved through acoustic emission event localization. Based on acoustic emission localization technology, the fracture trajectory is inverted to determine the spatial location of fracture development inside the coal and rock sample.

[0105] 1.1) Before the coal and rock mechanics test, n (n≥4) acoustic emission probes are evenly arranged on the side and back of the specimen. Vaseline is applied between the specimen and the acoustic emission probes as a coupling agent to improve the accuracy of sensor signal transmission. Data is monitored by the acoustic emission equipment to collect acoustic emission information during the coal and rock fracturing process. The spatial coordinates of the internal fracture development process of the specimen are inverted using the sound source localization method.

[0106] 1.2) This invention obtains the spatial coordinates of internal crack development in a sample based on a time-difference positioning method. This acoustic emission source positioning method uses the time difference of arrival of different sensors to invert the sound source location, i.e., the crack development location. The specific steps are: establishing a spatial rectangular coordinate system with the sample center point as the origin, and obtaining the spatial coordinates (x, y, z) of n sensors. n ,y n ,x n The spatial coordinates corresponding to the fracture development path are marked as (x i ,y i ,z i Let t be the time interval of the sound source signal received by the n sensors in 1.1) above. n Let t be the time it takes for the signal to travel from the acoustic emission source. Based on the time-difference positioning method, the following system of equations is established:

[0107]

[0108] Solving the system of equations yields a set of solutions, i.e., a set of coordinate points. The geometric center of these coordinate points is then determined; this geometric center represents the spatial coordinates (x, y) corresponding to the desired fracture development path. i ,y i ,z i ).

[0109] 1.3) Based on the localization method described in 1.2), the changes in internal fracture coordinates during fracture development are extracted to establish a spatial coordinate dataset. ={(x1,y1,z1) (x2,y2,z2) (x3,y3,z3)……(x n ,y n ,z n )}.

[0110] 1.4) By updating the spatial coordinates of the coal and rock fracture tips in real time, a dataset of internal fracture tip coordinate changes with temporal characteristics is established. (t)= .

[0111] 2) Prediction of the development trajectory of internal fractures: The fracture tip coordinate prediction model uses, but is not limited to, a long short-term memory network. It learns from the fracture tip coordinate dataset extracted in 1.4) above to predict the fracture tip coordinates at future moments, and further realizes the prediction of the dynamic extension process of the fracture at the next moment.

[0112] 2.1) The Long Short-Term Memory (LSTM) network described above is an improved model based on Recurrent Neural Networks (RNNs). By adding a gating mechanism and introducing a memory module to remember important coordinate information and forget unimportant information, it solves the problems of gradient explosion and disappearance in long sequences inherent in RNNs. LSTM has hidden states h t and unit state C t It is used to control and store long-term information about coordinates. LSTM has three inputs, namely the (t-1)th hidden state h. t-1 The (t-1)th unit state C t-1 And the t-th coordinate data. The output of the LSTM is the t-th unit state C. t and the t-th hidden state h t In addition, LSTM has three "gates": the forget gate (f), the input gate (i), and the output gate (o).

[0113] The historical coordinates of the development trajectory of internal fractures in coal and rock are substituted into the LSTM intelligent prediction algorithm. First, a forgetting gate is used. The forgetting gate controls whether to forget the hidden cell state of the previous layer with a certain probability, thus filtering out useful coordinate data. The probability value ranges from (0, 1). When the value is 1, it means "retain the previous coordinate data"; when the value is 0, it means "forget the previous coordinate data". The calculation formula is as follows:

[0114]

[0115] Among them, W f and U f Corresponding to h respectively (t-1) and x (t) The weight matrix, x (t) Input W, h as the data for the crack coordinates at time t. (t-1) b is the predicted coordinate output for the previous time step. f This is the corresponding field value.

[0116] The crack coordinates at time t are obtained by using the weight matrix and the sigmoid activation function, and this output is a probability obtained by the forget gate.

[0117] The crack coordinate data filtered through the forget gate will be fed into the input gate, which is responsible for processing the input of the current sequence position coordinates. The mathematical expression is as follows:

[0118]

[0119]

[0120] Both parts relate to input gates, including W. i W C and U i U c Corresponding to h respectively (t-1) and x (t) The weight matrix, x (t) The data input for the crack coordinates at time t is h. (t-1) b is the predicted coordinate output for the previous time step. f For the corresponding field value; C (t) The output h from the previous moment （t-1） and the current input data x (t) Temporary neural units are obtained by calculating the state of neural units using the hyperbolic tangent function.

[0121] Finally, the current crack coordinates are output through the output gate. The function of the output gate is to output the hidden state h at time t-1. (t-1) and the cell state c at time t (t) The mathematical expression is as follows:

[0122]

[0123]

[0124] Among them, W o and U o Corresponding to h respectively (t-1) and x (t) The weight matrix, b o For the corresponding field value, O t The coordinate output of the output gate, h (t) The coordinate prediction value of the internal slit expansion and extension dynamic process output by the neural unit at time t. .

[0125] 3) Identification of external coal and rock fractures. External coal and rock fracture identification is achieved through intelligent image recognition algorithms, namely, binarizing the extracted coal and rock fracture images and extracting the trajectory coordinates during the fracture expansion and penetration process.

[0126] 3.1) Before the mechanical test begins, select the center point of the specimen as the origin of the coordinate system, draw a plane coordinate system, and record the trajectory image of the external fracture development to facilitate the extraction of the coordinates of the external fracture development trajectory of the coal and rock.

[0127] 3.2) The image of the external fracture development process of coal and rock extracted in 3.1) above is binarized. The grayscale image of 256 brightness levels is obtained by selecting an appropriate threshold to obtain a binarized image that still reflects the overall and local features of the image. First, the grayscale image is binarized, that is, all pixels with grayscale values ​​greater than or equal to the threshold are determined to belong to a specific object, and their grayscale value is 255. Otherwise, these pixels are excluded from the object area, and their grayscale value is 0, representing the object area other than the background or target, thus obtaining the fracture binarized image. The grayscale value of the fracture pixels is selected as 255, and the grayscale value of the remaining pixels of the coal and rock sample is 0. The threshold is selected as T according to the actual situation. The threshold calculation formula for the midpoint (x,y) of the image is further obtained as follows:

[0128]

[0129] in, This is the final image after binarization. This represents the pixel value corresponding to the point (x, y) in the grayscale image. When the value is 255, it is determined to be a specific object, i.e., a crack; otherwise, when... When the value is 0, it is determined to be coal or rock.

[0130] 3.3) Extract the coordinates of the external fracture tip. Based on the fracture image after binarization in 3.2), scan line by line, sequentially traversing the pixel values ​​corresponding to any point (x, y) in the fracture image. Store it in set W i In the case of i, i takes the values ​​0, 1, 2, ..., n;

[0131]

[0132] The crack tip exhibits pixel value changes; the first threshold change point is considered the crack tip point, i.e., satisfying... The corresponding (x,y) is denoted as the coordinates of the crack tip.

[0133] 3.4) Repeat the tip search steps described in 3.3) to obtain a series of processes in the development of external fractures in coal and rock, and further derive the coordinate changes of the trajectory during fracture development, denoted as Ω={(x1,y1) (x2,y2) (x3,y3)……(x n ,y n )}.

[0134] 3.5) By updating the coordinates of the external fracture tips in coal and rock in real time, a dataset of temporal changes in the coordinates of the external fracture tips is established, denoted as W(t) = .

[0135] 4) The LSTM algorithm is used to predict the development trajectory of external fractures in coal and rock. The coordinate dataset of the external fracture tip in 3.5) above is used as a historical dataset for learning. It is then input into the LSTM model, and the prediction process is completed according to 2.1) above. The predicted coordinate values ​​of the external fracture expansion and extension dynamic process are output. .

[0136] 4.1) Real-time fusion of coal and rock mass fractures: Based on the spatial coordinate data established by the changes in internal fracture coordinates during fracture development in 1.4) and the dataset of external fracture coordinate changes in 3.4), for the external coordinates corresponding to a certain time t, combined with the corresponding internal fracture coordinates, the fracture development trajectory of the coal and rock specimen at time t is further inverted. This step is repeated to further obtain the fracture development trajectory at times t+1, t+2, etc., so as to realize the identification of the entire process of internal and external fracture development.

[0137] 5) Prediction of self-correction of coal and rock fractures

[0138] 5.1) Self-correcting prediction of internal fractures in coal and rock. Let the spatial coordinates of the tip of the internal fracture in the coal and rock detected by the acoustic emission device at time t be... Based on the model output in section 2.1) above, the predicted coordinates at time t are... The coordinate prediction error at time t is obtained as follows:

[0139]

[0140] The allowable error limit is set to b according to actual needs, and is determined based on the specimen size and experimental accuracy requirements.

[0141] 5.1.1) When When the model prediction error is within the allowable range, the prediction result is considered good. This is true when n consecutive coordinate predictions (determined based on actual conditions) all satisfy the condition. When this happens, the prediction step size can be increased, that is, the coordinates of the development trajectory of internal fractures in the coal and rock at time t+1 can be predicted; if the conditions are still met at this time... If the prediction step size is increased, the prediction step size can be increased further; otherwise, the current time step size should be maintained.

[0142] 5.1.2) When This indicates a discrepancy between the predicted trajectory of internal fracture development and the actual propagation path. In this case, it's necessary to reduce the prediction step size, i.e., predict the coordinates of the internal fracture development trajectory at time t-1. If it remains the same... Then, further reduce the prediction step size, i.e., predict t-2 steps, when the prediction error satisfies And all n consecutive coordinate predictions (determined based on actual conditions) satisfy the following conditions. If the forecast time step is set to t-1, the forecast time step can be set to t-1; otherwise, the current time step should be maintained.

[0143] 5.1.3) In 5.1.2) above, each time the prediction step size is modified, the model is updated simultaneously, and the model is continuously trained to improve the prediction accuracy.

[0144] 5.1.4) Repeat the steps in 5.1.1), 5.1.2) and 5.1.3) above to achieve forward self-correcting prediction of the development process of internal fractures in coal and rock.

[0145] 5.2) Self-correcting prediction of the development trajectory of external fractures in coal and rock. Let the coordinates of the tip of the external fracture during the development of the external fracture at time t be... Based on the model output in 4) above, the predicted coordinates at time t are... The coordinate prediction error at time t is obtained as follows:

[0146]

[0147] The allowable error limit is set to 'a' according to actual needs, and is determined based on the specimen size and experimental accuracy requirements.

[0148] 5.2.1) By comparison and Based on the above 5.1.1), 5.1.2) and 5.1.3), the size of the fractures is used to achieve self-correcting prediction of the development trajectory of external fractures in coal and rock.

[0149] This embodiment establishes a model for identifying the development trajectories of internal and external fractures in coal and rock, enabling real-time tracking of these trajectories and achieving dynamic identification of the combined development trajectories of internal and external fractures. A real-time self-correcting prediction model for internal and external fractures is also constructed, performing real-time self-correction on the predicted development trajectories and achieving dynamic forward prediction of these trajectories.

Claims

1. A method for dynamic identification and real-time prediction of coal and rock fracture trajectories, characterized in that, Includes the following steps: 1) Grid the sample surface; uniformly arrange n acoustic emission probes on the sidewall of the coal and rock sample, and arrange a high-speed camera on the outside of the coal and rock sample; 2) Divide the acquisition period T into multiple acquisition steps according to the time step Δt; During the mechanical experiment, acoustic emission equipment and a high-speed camera were used to collect data on the propagation process of internal fractures and the trajectory of external fractures within a period T. The coordinates of internal fractures at times t, t+1, t+2, t+3, ..., t+n were obtained, thus establishing a dataset of temporally sequenced changes in the coordinates of the internal fracture tips. (t)= The coordinates of the external fracture at times t, t+1, t+2, t+3, ..., t+n are obtained, thus establishing a dataset W(t) of external fracture tip coordinate changes with temporal characteristics. The data acquisition and calculation process for a single time step includes the following sub-steps: a) Acoustic emission information during the coal and rock fracturing process is collected using acoustic emission equipment; the development trajectory of internal fractures in coal and rock is identified at time t; acoustic emission information of the entire process of internal fracture development in coal and rock is recorded using acoustic emission technology, and then internal fractures in coal and rock are identified by sound source localization method; based on acoustic emission localization technology, the fracture trajectory is inverted to determine the spatial location of internal fracture development in coal and rock sample. b) A high-speed camera is used to record a series of processes of external fracture development and the fracture trajectory image is extracted; the trajectory of external fracture development in coal and rock at time t is identified; an intelligent image recognition algorithm is used to identify external fractures in coal and rock; the extracted coal and rock fracture image is binarized to separate the fractures and background in the image, and the trajectory coordinates in the fracture development process are further extracted to obtain data on the change of external fracture position over time. c) The development trajectories of internal and external fractures in the coal and rock merge at time t; Based on the internal and external coordinates at time t extracted in steps a) and b), the internal and external fracture points are matched using the point-to-point matching method to achieve real-time fusion of internal and external fractures. 3) The dataset of coordinate changes at the tips of internal and external fractures extracted in step 2) is used as a historical dataset for learning to obtain a fracture tip coordinate prediction model. This model is then used to predict the development trajectory of internal and external fractures. The coordinates of the fracture tips at future moments are predicted, enabling the prediction of the dynamic extension process of internal and external fractures at the next moment. The predicted coordinate values ​​of the dynamic development process of internal and external fractures are then output. and ; 4) The prediction step size is continuously adjusted based on the prediction results of the internal and external fracture prediction model, and the model is updated in real time to realize the forward self-correction prediction of the development trajectory of internal and external fractures in coal and rock. 5) Identify and predict the development trajectory of internal and external fractures; Based on the predicted coordinates of internal and external fractures in coal and rock obtained in steps 3) and 4), the internal and external fracture points are matched using the point-to-point matching method to achieve the fusion of internal and external fracture points based on time sequence; then the predicted forward fracture trajectory is obtained.

2. The method for dynamic identification and real-time prediction of coal and rock fracture trajectories according to claim 1, characterized in that: In step 1), the mechanical experiment is a tensile failure mechanical experiment.

3. The method for dynamic identification and real-time prediction of coal and rock fracture trajectories according to claim 1, characterized in that: In step 1), n ​​≥ 4.

4. The method for dynamic identification and real-time prediction of coal and rock fracture trajectories according to claim 1, characterized in that, Step a) specifically includes the following sub-steps: a1) Establish a spatial rectangular coordinate system with the center point of the coal and rock sample as the origin, and obtain the spatial coordinates (x, y, z) of the n sensors. n y n , z n The spatial coordinates corresponding to the fracture development path are marked as (x i y i , z i The time interval of the sound source signals received by the n sensors is denoted as t. n Let t be the time it takes for the signal to be emitted from the acoustic emission source; establish equation (1) based on the time difference positioning method; （1） Solving equation (1) yields several sets of coordinate points; the geometric centers of these coordinate points are then determined; these geometric centers are the spatial coordinates (x, y) corresponding to the required fracture development path. i y i , z i ); a2) Extract the changes in internal fracture coordinates during fracture development and establish a spatial coordinate dataset. ={(x1,y1,z1) (x2,y2,z2) (x3,y3,z3)……(x n y n , z n )}; a3) By updating the spatial coordinates of the coal and rock fracture tips in real time, a dataset of internal fracture tip coordinate changes with temporal characteristics is established. (t)= .

5. The method for dynamic identification and real-time prediction of coal and rock fracture trajectories according to claim 1, characterized in that, Step b) specifically includes the following sub-steps: b1) Record the trajectory image of external fracture development; wherein, the center point of the coal and rock specimen is selected as the origin of the coordinate system and a plane coordinate system is drawn; b2) Perform binarization processing on the image of the coal and rock external fracture development process obtained in step b1); b3) Extract the coordinates of the external fracture tip; based on the fracture image after binarization in step b2), scan line by line, sequentially traversing the pixel values ​​corresponding to any point (x, y) in the fracture image. and stored in set W i In the middle; where i takes the values ​​0, 1, 2, ..., n; (2) The tip of the crack exhibits a change in pixel value. The first point of threshold change is defined as the crack tip point, i.e., satisfying... The corresponding (x,y) is denoted as the coordinates of the crack tip; b4) Repeat the tip search step in b3) to obtain a series of processes in the development of external fractures in coal and rock, and further derive the coordinate changes of the trajectory during fracture development, denoted as Ω={(x1, y1) (x2, y2) (x3, y3)……(x n y n )}; b5) By updating the coordinates of the external fracture tips in coal and rock in real time, a dataset of temporal changes in the coordinates of the external fracture tips is established, denoted as W(t) = .

6. The method for dynamic identification and real-time prediction of coal and rock fracture trajectories according to claim 1, characterized in that: In step c), the internal coordinates reflecting the morphology and size of the fracture are combined with the external coordinates reflecting the displacement and deformation of the fracture. For the external coordinates corresponding to a certain time t, combined with the corresponding internal fracture coordinates, the fracture development trajectory of the coal and rock specimen at time t is further inverted. Based on the extracted internal and external coordinates at time t, the internal and external fracture points are matched using the point-to-point matching method to achieve real-time fusion of internal and external fractures. A three-dimensional fracture model is formed based on the fusion result, which more accurately identifies the location and morphology of internal and external fractures.

7. The method for dynamic identification and real-time prediction of coal and rock fracture trajectories according to claim 1, characterized in that: In step 3), the extracted dataset of coordinate changes of the tips of internal and external fractures in coal and rock is used for training to obtain a fracture development trajectory prediction model. By using this model to predict the coordinates of the tips of internal and external fractures at future moments, the development trajectory of internal and external fractures in coal and rock can be predicted. The prediction model includes a Long Short-Term Memory (LSTM) network.

8. The method for dynamic identification and real-time prediction of coal and rock fracture trajectories according to claim 1, characterized in that: In step 4), the real-time self-correction method collects and monitors fracture development data in real time and establishes a corresponding feedback mechanism. During implementation, based on the research accuracy, a corresponding error range threshold is set for the prediction model. When the model's predicted value exceeds the error threshold, the model needs to be self-corrected to improve the prediction accuracy.

Images