Coal rock fracture field imaging method and system based on multi-target joint guide architecture

Abstract

The invention discloses a coal rock fracture field imaging method and system based on a multi-target combined guide framework. The imaging method comprises the following steps: obtaining staged time CT detection and full space-time acoustic emission monitoring data in a coal rock sample loading process; cT detection and acoustic emission monitoring data are subjected to correspondence according to moments, and a fracture parameter data set is obtained; constructing a multi-target joint guidance neural network architecture through a learnable fracture distribution module, a fracture parameter prediction mechanism equation, a joint training equation, a fracture distribution reconstruction equation and a free space equation; the neural network architecture performs feature extraction on fracture and acoustic emission parameter data of a training set and a verification set in the fracture parameter data set to obtain a three-dimensional fracture field imaging model; and inputting acoustic emission parameter data of the test set into the three-dimensional coal rock fracture field imaging model to obtain a three-dimensional fracture field imaging result of the coal rock sample, and performing deviation calculation according to the imaging result and CT detection data of the test set to adjust neural network parameters.

Description

Technical Field

[0001] This invention relates to the field of coal and rock monitoring technology, and in particular to a coal and rock fracture field imaging method and system based on a multi-target joint guidance architecture. Background Technology

[0002] The vast majority of resources and energy hidden deep within the Earth are the material and energy foundation for the survival of all life. Exploring the Earth's depths for energy and space is an essential path to solving the fundamental resource, energy, and living space problems facing humanity. However, the extraction of underground resources and energy is often accompanied by highly destructive rock dynamic disasters (including rockbursts, rock bursts, and coal and gas outbursts), causing significant casualties and destroying basic engineering projects, equipment, and facilities. The occurrence of rock dynamic disasters typically involves the unstable propagation of fractures and the induction of large-scale fractures, a core issue closely related to fracture evolution. Therefore, to gain a deeper understanding of the mechanisms of coal and rock dynamic disasters and to achieve early warning of these disasters, it is necessary to conduct in-depth research on coal and rock fracture field imaging and other related aspects.

[0003] While existing rock mechanics theories and methods attempt to explain the evolution of fracture fields in coal and rock masses caused by mining processes, this complex process remains a "black box" that is difficult to measure and reveal. Many challenges remain in real-time monitoring, precise imaging, and revealing the development and evolution of fracture fields. Currently, industrial CT detection and numerical simulation methods have made some progress. These technologies can be used to reconstruct the internal structure of coal and rock masses, especially showing potential in constructing three-dimensional digital models of discontinuous structures and fracture fields. For example, patent CN114169182A discloses a method and equipment for reconstructing an ellipsoidal model of rock mass fractures; patent CN110146525A discloses a method for predicting coal porosity and permeability parameters based on fractal theory and CT scanning; and patent CN107478357A discloses an integrated monitoring system and quantitative determination method for surrounding rock stress and fracture fields. Nevertheless, these methods still have limitations. They typically rely on high-specification laboratory conditions and cumbersome model reconstruction methods, which limits their widespread application in practical engineering. Furthermore, these methods cannot provide real-time imaging of the dynamic spatiotemporal evolution of the fracture field distribution.

[0004] In recent years, experimental methods combining exploration-monitoring technology and numerical simulation have been applied in laboratories. However, current utilization of this exploration-monitoring data mainly focuses on manual analysis of its temporal variation patterns. Furthermore, relying solely on numerical simulation for fracture field imaging requires extremely high computational costs; even minor modifications to the model can significantly negatively impact calculations. Therefore, how to utilize exploration-monitoring data to achieve full-spatiotemporal fracture field imaging and overcome the limitations of traditional methods requires further in-depth research. With continuous technological advancements, breakthroughs in neural network methods have attracted attention for their application in fracture field imaging. In particular, imaging methods based on data mechanism-guided neural network architectures can effectively extract features from different types of data and combine them with mechanistic equations to achieve fracture field imaging. Compared to traditional neural network architectures, this significantly improves computational efficiency, reduces computational resource requirements, and achieves higher imaging accuracy.

[0005] This provides the possibility of using the above methods to achieve coal and rock fracture field imaging, but currently there is still a lack of specific feasible ideas and implementation methods. Therefore, it is urgent to carry out actual measurement of detection and monitoring signals during the loading process of coal and rock samples, propose a coal and rock fracture field imaging method and system based on a multi-target joint guidance architecture, and realize coal and rock fracture field imaging based on all-time and spacetime acoustic emission monitoring data. This will have a significant promoting effect on revealing the "black box" evolution process of coal and rock mass dynamic disasters and realizing the monitoring and early warning of coal and rock dynamic disasters. Summary of the Invention

[0006] This solution addresses the problems and needs raised above by proposing a coal and rock fracture field imaging method and system based on a multi-objective joint guidance architecture. The above technical objectives can be achieved by adopting the following technical features, and other technical effects are also brought about.

[0007] One objective of this invention is to propose a coal and rock fracture field imaging method based on a multi-objective joint guidance architecture, comprising the following steps:

[0008] S10: CT detection sensors and acoustic emission monitoring sensors are arranged on the coal and rock sample, and the coal and rock sample is placed in the testing machine for loading test. Acoustic emission monitoring data and CT detection data at different stages are collected by the acoustic emission monitoring sensor and CT detection sensor during the loading process of the coal and rock sample; among them, the acoustic emission monitoring data is the acoustic emission parameter data at all times and space, and the CT detection data is the fracture parameter data at different stages and space.

[0009] S20: The CT detection data and all-time acoustic emission monitoring data at different moments during the loading process of the coal and rock samples are correlated according to time to obtain the fracture parameter dataset, and the fracture parameter dataset is divided into training set, validation set and test set;

[0010] S30: Constructing a multi-objective joint guided neural network architecture based on a learnable crack distribution module;

[0011] S40: The multi-objective joint guided neural network architecture extracts features from the fracture parameter data and corresponding acoustic emission parameter data of the training set and validation set in the fracture parameter dataset to obtain a three-dimensional fracture field imaging model of coal and rock.

[0012] S50: Input the acoustic emission parameter data from the test set into the three-dimensional fracture field imaging model of coal and rock to obtain the three-dimensional fracture field imaging results of coal and rock. Determine whether the accuracy of the three-dimensional fracture field imaging results of coal and rock meets the requirements. If the accuracy of the three-dimensional fracture field imaging results of coal and rock does not meet the requirements, calculate the deviation using the CT detection data of the test set and adjust the neural network parameters until the accuracy of the three-dimensional fracture field imaging results of coal and rock meets the requirements.

[0013] Furthermore, the coal and rock fracture field imaging method based on a multi-target joint guidance architecture according to the present invention may also have the following technical features:

[0014] In one example of the present invention, step S30 further includes: constraining the training process of the multi-objective jointly guided neural network architecture through the crack parameter prediction mechanism equation and the joint training constraint equation, specifically including the following steps:

[0015] First, the training process of the multi-objective jointly guided neural network architecture is constrained by the crack parameter prediction mechanism equation, whereby the expression of the crack parameter prediction mechanism equation is as follows:

[0016]

[0017] In the formula, The equation representing the mechanism for predicting the location of a true three-dimensional crack; N represents the distance between different acoustic emission sensors. The total number of combinations, where n represents the total number of acoustic emission sensors; varx pred ,vary pred varz pred These represent the deviation parameters of the crack location predicted by the neural network architecture; x e y e , z e These represent the coordinates of the acoustic emission microcrack event in the x, y, and z directions, respectively; x i y i , z i These represent the coordinates of the acoustic emission sensor in the x, y, and z directions, respectively; v represents the P-wave velocity; Δt ij This represents the difference in positioning time between the i-th and j-th sensors;

[0018]

[0019] In the formula, The equation representing the prediction mechanism of true three-dimensional fracture parameters; n represents the total number of acoustic emission sensors; α represents the weighting coefficient of acoustic emission parameters; AE p,i F represents the acoustic emission parameters monitored by the i-th acoustic emission sensor; M,pred This represents the fracture parameters that need to be predicted.

[0020] Then, the training process of the multi-objective jointly guided neural network architecture is constrained by the joint training constraint equation, the expression of which is as follows:

[0021]

[0022] In the formula, represents the joint training constraint equation; argmin represents the objective of finding the minimum value of the joint training constraint equation. denoted by , W represents the latent vector of the three-dimensional fracture distribution corresponding to different fracture distributions in the dataset; represents the number of three-dimensional fracture distributions; represents the index of the latent vector of the three-dimensional fracture distribution; represents the total number of moments in the training set where fractures are distributed at different times; represents the total number of locations in the training set where fractures are distributed at different locations; x j Representative acoustic emission microcrack event location parameters; F j Represents three-dimensional fracture parameters; g θ (z i ,AE p,i ) represents the data-driven neural network's response to the acoustic emission parameter AE p,i The predicted parameters of the three-dimensional fracture; s j The representative neural network architecture obtains the fracture location deviation parameter through acoustic emission micro-fracture event coordinate prediction; z i θ represents the potential vector of the three-dimensional crack distribution; θ represents all trainable parameters of the data-driven neural network; f θ (z i ,x j ) represents a data-driven neural network's response to acoustic emission microcrack events x j The predicted location coordinates of the three-dimensional crack; Represents the deviation calculation function. δ represents the region of accumulation of acoustic emission microcrack events; σ 2 The prior distribution variance represents the potential vector of the three-dimensional crack distribution; This represents the squared norm of the latent vector.

[0023] In one example of the present invention, step S40 further includes: constraining the testing process of the three-dimensional fracture field imaging model of coal and rock by means of the fracture distribution reconstruction equation and the free space constraint equation, specifically including the following steps:

[0024] First, the prediction process of the three-dimensional fracture distribution in the three-dimensional fracture field imaging model of coal and rock is constrained by the fracture distribution reconstruction equation. The expression of the fracture distribution reconstruction equation is as follows:

[0025]

[0026] In the formula, X represents the three-dimensional crack distribution obtained by the multi-objective joint guided neural network architecture inference; X represents the data set of acoustic emission micro-crack event coordinates and real three-dimensional crack coordinates; z represents the global distribution attribute of the three-dimensional crack distribution;

[0027] Then, the prediction process of three-dimensional fracture parameters in the three-dimensional fracture field imaging model of coal and rock is constrained by the free space constraint equation, wherein the expression of the free space constraint equation is as follows:

[0028]

[0029] In the formula, Represents the free space constraint equations; The set representing the distribution of fractures in the three-dimensional space of a coal and rock sample; x k represent The distribution of cracks at a certain location in the image shows that the prediction process for these three-dimensional crack distributions has errors and requires additional constraints for correction; η represents the error range.

[0030] In one example of the present invention, in step S40, the multi-objective jointly guided neural network architecture extracts features from the fracture parameter data and the acoustic emission parameter data at the corresponding time points in the fracture parameter dataset, including the following steps:

[0031] S41: Transmit acoustic emission parameters AE through the data encoding layer p Data extraction yields parameter-enhanced acoustic emission features AEf se ;

[0032] S42: The encoded data is concatenated to the feature extraction layer B through a residual connection layer to obtain feature AE. b middle;

[0033] S43: Acoustic emission feature AEf enhanced by the first Normalization function se Regularization is performed to obtain the feature AE n ;

[0034] S44: Regularized Feature AE n Feature extraction AE is obtained through the residual connection formed by the second linear feature extraction layer A and the second activation function layer. a+s The feature AE spliced ​​into the output of the selective state space architecture sssm Get AE SSSM ;

[0035] S45: Feature AE output by the first linear feature extraction layer A a Higher-dimensional features (AE) are obtained through feature extraction via convolutional layers and a first activation function layer. conv+s ;

[0036] S46: High-dimensional features are gated using a selective state-space architecture and concatenated with features extracted from convolutional layers and the first activation function layer to obtain feature AE. SSSM ;

[0037] S47: Linear feature extraction layer B extracts features from the concatenated features and concatenates the extracted features with the features of the encoded data obtained by the residual connection layer to obtain feature AEf. b+se ;

[0038] S48: Regularize the concatenated features using the second Normalization function to obtain feature AEf. n ;

[0039] S49: The features extracted by the fully connected layer are used for calculation to obtain the fracture location deviation parameter, which is the difference (varx) between the acoustic emission micro-fracture event coordinates predicted by the neural network architecture and the actual three-dimensional fracture coordinates in the x, y, z directions. M,pred ,vary M,pred ,varz M,pred ) and fracture parameter F M,pred .

[0040] In one example of the present invention, in step S46, the selective state-space architecture is specifically implemented as follows:

[0041] First, a state-space architecture is constructed using continuous state-space equations, the expressions of which are as follows:

[0042] h t =Ah t-1 +Bx t

[0043] y t =Ch t

[0044] Among them, ht-1 Represents the system state; h t Represents the updated system state; matrix A represents the state transition matrix; matrix B represents the input gating matrix; matrix C represents the output mapping matrix; x t The data represents the current state of the input architecture.

[0045] Secondly, the state-space architecture is subjected to discrete data adaptation using discrete state-space equations; the equation for the state transition matrix A is as follows:

[0046]

[0047] In the formula, n is the row index related to the orthogonal basis functions in the polynomial space, and k is the column index related to the orthogonal basis functions in the polynomial space.

[0048] In one example of the present invention, the method further includes: randomly initializing the state transition matrix A and using a low-rank matrix obtained from matrix decomposition, the equation of which is as follows:

[0049]

[0050] Where P and Q are both vectors of length N; V represents the eigenvector matrix; V * Λ represents the transpose matrix; Λ represents the eigenvalue matrix; * represents the projection operation.

[0051] In one example of the present invention, the expression of the discrete state-space equation is as follows:

[0052]

[0053] In the formula, h k and y k These represent the memory control unit and the output control unit, respectively, for selecting the current state in the state-space architecture; h k-1 h0 represents the memory control unit that selects the previous state in the state space architecture; h0 represents the initialization state of the memory control unit; x k x0 represents the data of the current state input architecture; x0 represents the data of the initial state input architecture. All are matrices after discrete adaptation of A, B, and C.

[0054] In one example of the invention, the matrix It is obtained from the following derivation formula:

[0055]

[0056] In the formula, Δ represents the discretized nonlinear function, which discretizes the continuous state; Ι represents the identity matrix.

[0057] In one example of the present invention, in step S50, the formula for calculating the deviation using the test set CT detection data is as follows:

[0058]

[0059] In the formula, MSE represents the deviation calculation equation; M represents the number of samples; i represents the index of the crack parameter; F M,pred F represents the predicted value of the Mth sample; M,true The value represents the true value of the Mth sample.

[0060] Another objective of this invention is to propose a coal and rock fracture field imaging system based on a multi-target joint guidance architecture, comprising:

[0061] The information acquisition module is configured to deploy CT detection sensors and acoustic emission monitoring sensors on coal and rock samples, and place the coal and rock samples in a testing machine for loading tests. The acoustic emission monitoring sensors and CT detection sensors acquire full-time and spatiotemporal acoustic emission monitoring data and stage-by-stage CT detection data of the coal and rock samples during the loading process. Among them, the acoustic emission monitoring data is full-time and spatiotemporal acoustic emission parameter data, and the CT detection data is stage-by-stage full-space fracture parameter data.

[0062] The data partitioning module is configured to map the CT detection data and all-time acoustic emission monitoring data at different stages during the loading process of coal and rock samples according to time, obtain the fracture parameter dataset, and divide the fracture parameter dataset into training set, validation set and test set;

[0063] The neural network architecture building module is configured to build a multi-objective jointly guided neural network architecture based on a learnable gap distribution module.

[0064] The coal and rock three-dimensional fracture field imaging module is configured to use a multi-objective joint guided neural network architecture to extract features from the fracture parameter data and corresponding acoustic emission parameter data of the training set and validation set in the fracture parameter dataset, so as to obtain the coal and rock three-dimensional fracture field imaging model.

[0065] The parameter adjustment module is configured to input acoustic emission parameter data from the test set into the three-dimensional fracture field imaging model of coal and rock to obtain the three-dimensional fracture field imaging results of coal and rock, determine whether the accuracy of the three-dimensional fracture field imaging results of coal and rock meets the requirements, and when the accuracy of the three-dimensional fracture field imaging results of coal and rock does not meet the requirements, it calculates the deviation using CT detection data from the test set and adjusts the neural network parameters until the accuracy of the three-dimensional fracture field imaging results of coal and rock meets the requirements.

[0066] Compared with the prior art, the present invention has the following beneficial effects:

[0067] 1. This invention focuses on a coal and rock fracture field imaging method and system based on a multi-objective joint guidance architecture, which realizes fracture field imaging based on acoustic emission monitoring data during coal and rock sample loading tests. Compared with traditional numerical simulation and neural network methods, this coal and rock fracture field imaging method based on a multi-objective joint guidance architecture guides the extraction process of fracture distribution features of the learnable fracture distribution module by using fracture parameter prediction mechanism equations, joint training equations, fracture distribution reconstruction equations, and free space equations. This ensures that the output of the guiding model follows physical laws, fully explores the nonlinear features in multi-source experimental data, and guarantees the accuracy of coal and rock fracture imaging results, thus constructing a new paradigm for coal and rock fracture field imaging.

[0068] 2. The learnable fracture distribution module in this invention is based on a neural network with a selective state-space architecture. This architecture significantly reduces computational resource dependence through fine-grained partitioning and filtering of state and observation variables. Furthermore, the neural network uses a multi-matrix parameter method to efficiently extract high-dimensional features from CT detection and acoustic emission monitoring data during coal and rock sample loading, enabling it to effectively learn acoustic emission and fracture parameter features, providing high-value data features for coal and rock fracture imaging.

[0069] 3. This invention adds a fracture distribution reconstruction equation and a free space equation to guide the fracture imaging process in the learnable fracture distribution module, thereby improving the accuracy of coal and rock fracture imaging results. The fracture distribution reconstruction equation incorporates mechanistic equation knowledge, which not only gives the fracture field imaging method strong generalization ability but also enhances its interpretability. Simultaneously, the coal and rock fracture field imaging method based on a multi-objective joint guidance architecture constrains the fracture location prediction process through the free space equation, achieving accurate imaging of coal and rock fractures during sample loading.

[0070] The preferred embodiments of the invention will be described in more detail below with reference to the accompanying drawings, so as to facilitate an understanding of the features and advantages of the invention. Attached Figure Description

[0071] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments of the present invention will be briefly described below. The drawings are merely illustrative of some embodiments of the present invention and are not intended to limit the scope of the present invention to all embodiments.

[0072] Figure 1 This is a flowchart of a coal and rock fracture field imaging method based on a multi-target joint guidance architecture according to an embodiment of the present invention;

[0073] Figure 2 This is a schematic diagram of the learnable crack distribution module according to an embodiment of the present invention;

[0074] Figure 3 This is a schematic diagram of the arrangement of coal and rock samples in the testing machine according to an embodiment of the present invention;

[0075] List of reference numerals in the attached diagram:

[0076] Testing machine 100;

[0077] Loading device 110;

[0078] Acoustic emission sensor 120;

[0079] Coupler 130;

[0080] Preamplifier 140;

[0081] X-ray detector 150;

[0082] X-ray source 160;

[0083] 200 coal and rock samples;

[0084] Lateral direction X;

[0085] The vertical direction is Y. Detailed Implementation

[0086] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. The same reference numerals in the drawings represent the same components. It should be noted that the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0087] Unless otherwise defined, the technical or scientific terms used herein shall have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms “first,” “second,” and similar terms used in this patent application specification and claims do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Similarly, “an” or “a” and similar terms do not necessarily indicate a quantity limitation. Terms such as “comprising” or “including” mean that the element or object preceding the word encompasses the element or object listed following the word and its equivalents, without excluding other elements or objects. Terms such as “connected” or “linked” are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as “upper,” “lower,” “left,” and “right” are used only to indicate relative positional relationships; these relative positional relationships may change accordingly when the absolute position of the described object changes.

[0088] According to a first aspect of the present invention, a coal and rock fracture field imaging method based on a multi-target joint guidance architecture is provided, such as... Figure 1 As shown, it includes the following steps:

[0089] S10: CT detection sensors and acoustic emission monitoring sensors are placed on the coal and rock samples, and the samples are placed in a testing machine for loading tests. The acoustic emission monitoring sensors and CT detection sensors collect full-time and spatiotemporal acoustic emission monitoring data and stage-by-stage CT detection data during the loading process. The acoustic emission monitoring data represents full-time and spatiotemporal acoustic emission parameters, while the CT detection data represents stage-by-stage full-space fracture parameters. (The sample location is as follows...) Figure 3As shown, the testing machine 100 includes: a loading device 110, an acoustic emission sensor 120, a coupler 130, a preamplifier 140, an X-ray detector 150, an X-ray source 160, and a test specimen 170; wherein, the loading device 110, acoustic emission sensor 120, coupler 130, preamplifier 140, X-ray detector 150, and X-ray source 160 are all installed in the test specimen 170, and the coal and rock sample 200 is disposed in the loading device 110. Acoustic emission sensors 120 are installed on both sides of the coal and rock sample 200 in the transverse direction X, wherein the acoustic emission sensors 120 are electrically connected to the coupler 130 and the preamplifier 140, and X-ray detectors 150 and X-ray sources 160 are respectively disposed on both sides of the coal and rock sample 200 in the longitudinal direction Y, wherein the transverse direction X and the longitudinal direction Y are perpendicular to each other. The working process is as follows: The loading device 110 loads the coal and rock sample 200. During this process, the X-ray source 160 emits X-rays and the X-ray detector 150 receives the X-rays, thereby obtaining CT detection data during the fracturing process of the coal and rock sample 200. At the same time, the acoustic emission sensor 120 obtains all-time and all-space acoustic emission monitoring data.

[0090] S20: The CT detection data and all-time acoustic emission monitoring data at different moments during the loading process of the coal and rock samples are correlated according to time to obtain the fracture parameter dataset, and the fracture parameter dataset is divided into training set, validation set and test set;

[0091] S30: Constructing a multi-objective joint guided neural network architecture based on a learnable crack distribution module;

[0092] S40: The multi-objective joint guided neural network architecture extracts features from the fracture parameter data and corresponding acoustic emission parameter data of the training set and validation set in the fracture parameter dataset to obtain a three-dimensional fracture field imaging model of coal and rock.

[0093] S50: Input the acoustic emission parameter data from the test set into the three-dimensional fracture field imaging model of coal and rock to obtain the three-dimensional fracture field imaging results of coal and rock. Determine whether the accuracy of the three-dimensional fracture field imaging results of coal and rock meets the requirements. When the accuracy of the three-dimensional fracture field imaging results of coal and rock meets the requirements, the calculation ends. When the accuracy of the three-dimensional fracture field imaging results of coal and rock does not meet the requirements, perform deviation calculation using the CT detection data of the test set and adjust the neural network parameters until the accuracy of the three-dimensional fracture field imaging results of coal and rock meets the requirements.

[0094] This imaging method focuses on achieving fracture field imaging based on acoustic emission monitoring data during coal and rock sample loading tests through a coal and rock fracture field imaging method and system based on a multi-objective joint guidance architecture. Compared with traditional numerical simulation and neural network methods, this coal and rock fracture field imaging method based on a multi-objective joint guidance architecture guides the extraction process of fracture distribution characteristics of the learnable fracture distribution module by using fracture parameter prediction mechanism equations, joint training equations, fracture distribution reconstruction equations, and free space equations. This ensures that the output of the guiding model follows physical laws, fully explores the nonlinear characteristics in multi-source experimental data, and guarantees the accuracy of coal and rock fracture imaging results, thus constructing a new paradigm for coal and rock fracture field imaging.

[0095] The learnable fracture distribution module in this imaging method is based on a neural network with a selective state-space architecture. This architecture significantly reduces computational resource dependence through fine-grained partitioning and filtering of state and observation variables. Furthermore, the neural network utilizes a multi-matrix parametric method to efficiently extract high-dimensional features from CT detection and acoustic emission monitoring data during coal and rock sample loading. This allows the neural network to effectively learn acoustic emission and fracture parameter features, providing high-value data features for coal and rock fracture imaging.

[0096] This imaging method incorporates fracture distribution reconstruction equations and free space equations into the fracture imaging process of the learnable fracture distribution module to guide the imaging process, thereby improving the accuracy of coal and rock fracture imaging results. The fracture distribution reconstruction equations integrate mechanistic equation knowledge, which not only gives the fracture field imaging method strong generalization ability but also enhances its interpretability. Simultaneously, the coal and rock fracture field imaging method based on a multi-objective joint guidance architecture constrains the fracture location prediction process through free space equations, achieving accurate imaging of coal and rock fractures during sample loading.

[0097] In one example of the present invention, step S30 further includes: constraining the training process of the multi-objective jointly guided neural network architecture through the crack parameter prediction mechanism equation and the joint training constraint equation, specifically including the following steps:

[0098] First, the training process of the multi-objective jointly guided neural network architecture is constrained by the crack parameter prediction mechanism equation, whereby the expression of the crack parameter prediction mechanism equation is as follows:

[0099]

[0100] In the formula, The equation representing the mechanism for predicting the location of a true three-dimensional crack; N represents the distance between different acoustic emission sensors. The total number of combinations, where n represents the total number of acoustic emission sensors; varx pred ,vary pred varz pred These represent the deviation parameters of the crack location predicted by the neural network architecture; x e y e , z e These represent the coordinates of the acoustic emission microcrack event in the x, y, and z directions, respectively; x i y i , z i These represent the coordinates of the acoustic emission sensor in the x, y, and z directions, respectively; v represents the P-wave velocity; Δt ij This represents the difference in positioning time between the i-th and j-th sensors;

[0101]

[0102] In the formula, The equation representing the prediction mechanism of true three-dimensional fracture parameters; n represents the total number of acoustic emission sensors; α represents the weighting coefficient of acoustic emission parameters; AE p,i F represents the acoustic emission parameters monitored by the i-th acoustic emission sensor; M,pred This represents the fracture parameters that need to be predicted.

[0103] Then, the training process of the multi-objective jointly guided neural network architecture is constrained by the joint training constraint equation, the expression of which is as follows:

[0104]

[0105] In the formula, arg represents the joint training constraint equation; arg min represents the objective of finding the minimum value of the joint training constraint equation. denoted by , W represents the latent vector of the three-dimensional fracture distribution corresponding to different fracture distributions in the dataset; represents the number of three-dimensional fracture distributions; represents the index of the latent vector of the three-dimensional fracture distribution; represents the total number of moments in the training set where fractures are distributed at different times; represents the total number of locations in the training set where fractures are distributed at different locations; x j Representative acoustic emission microcrack event location parameters; F j Represents three-dimensional fracture parameters; g θ (z i ,AE p,i ) represents the data-driven neural network's response to the acoustic emission parameter AE p,i The predicted parameters of the three-dimensional fracture; s j The representative neural network architecture obtains the fracture location deviation parameter through acoustic emission micro-fracture event coordinate prediction; z iθ represents the potential vector of the three-dimensional crack distribution; θ represents all trainable parameters of the data-driven neural network; f θ (z i ,x j ) represents a data-driven neural network's response to acoustic emission microcrack events x j The predicted location coordinates of the three-dimensional crack; Represents the deviation calculation function. δ represents the region of accumulation of acoustic emission microcrack events; σ 2 The prior distribution variance represents the potential vector of the three-dimensional crack distribution; This represents the squared norm of the latent vector.

[0106] In one example of the present invention, step S40 further includes: constraining the testing process of the three-dimensional fracture field imaging model of coal and rock by means of the fracture distribution reconstruction equation and the free space constraint equation, specifically including the following steps:

[0107] First, the prediction process of the three-dimensional fracture distribution in the three-dimensional fracture field imaging model of coal and rock is constrained by the fracture distribution reconstruction equation. The expression of the fracture distribution reconstruction equation is as follows:

[0108]

[0109] In the formula, X represents the three-dimensional crack distribution obtained by the multi-objective joint guided neural network architecture inference; X represents the data set of acoustic emission micro-crack event coordinates and real three-dimensional crack coordinates; z represents the global distribution attribute of the three-dimensional crack distribution;

[0110] Then, the prediction process of three-dimensional fracture parameters in the three-dimensional fracture field imaging model of coal and rock is constrained by the free space constraint equation, wherein the expression of the free space constraint equation is as follows:

[0111]

[0112] In the formula, Represents the free space constraint equations; The set representing the distribution of fractures in the three-dimensional space of a coal and rock sample; x k represent The distribution of cracks at a certain location in the image shows that the prediction process for these three-dimensional crack distributions has errors and requires additional constraints for correction; η represents the error range.

[0113] In one example of the present invention, in step S40, as Figure 2 As shown, the multi-objective joint guided neural network architecture extracts features from the fracture parameter data and corresponding acoustic emission parameter data of the training and validation sets in the fracture parameter dataset, including the following steps:

[0114] S41: Transmit acoustic emission parameters AE through the data encoding layer p Data extraction yields parameter-enhanced acoustic emission features AEf se ;

[0115] S42: The encoded data is concatenated to the feature extraction layer B through a residual connection layer to obtain feature AE. b middle;

[0116] S43: Acoustic emission feature AEf enhanced by the first Normalization function se Regularization is performed to obtain the feature AE n ;

[0117] S44: Regularized Feature AE n Feature extraction AE is obtained through the residual connection formed by the second linear feature extraction layer A and the second activation function layer. a+s The feature AE spliced ​​into the output of the selective state space architecture sssm Get AE SSSM ;

[0118] S45: Feature AE output by the first linear feature extraction layer A a Higher-dimensional features (AE) are obtained through feature extraction via convolutional layers and a first activation function layer. conv+s ;

[0119] S46: High-dimensional features are gated using a selective state-space architecture and concatenated with features extracted from convolutional layers and the first activation function layer to obtain feature AE. SSSM ;

[0120] S47: Linear feature extraction layer B extracts features from the concatenated features and concatenates the extracted features with the features of the encoded data obtained by the residual connection layer to obtain feature AEf. b+se ;

[0121] S48: Regularize the concatenated features using the second Normalization function to obtain feature AEf. n ;

[0122] S49: The features extracted by the fully connected layer are used for calculation to obtain the fracture location deviation parameter, which is the difference (varx) between the acoustic emission micro-fracture event coordinates predicted by the neural network architecture and the actual three-dimensional fracture coordinates in the x, y, z directions. M,pred ,vary M,pred ,varz M,pred ) and fracture parameter F M,pred .

[0123] In one example of the present invention, in step S46, the selective state-space architecture is specifically implemented as follows:

[0124] First, a state-space architecture is constructed using continuous state-space equations, the expressions of which are as follows:

[0125] h t =Ah t-1 +Bx t

[0126] y t =Ch t

[0127] Among them, h t-1 Represents the system state; h t Represents the updated system state; matrix A represents the state transition matrix; matrix B represents the input gating matrix; matrix C represents the output mapping matrix; x t The data represents the current state of the input architecture.

[0128] Secondly, the state-space architecture is subjected to discrete data adaptation using discrete state-space equations; the equation for the state transition matrix A is as follows:

[0129]

[0130] In the formula, n is the row index related to the orthogonal basis functions in the polynomial space, and k is the column index related to the orthogonal basis functions in the polynomial space.

[0131] In one example of the present invention, the method further includes: randomly initializing the state transition matrix A and using a low-rank matrix obtained from matrix decomposition, the equation of which is as follows:

[0132]

[0133] Where P and Q are both vectors of length N; V represents the eigenvector matrix; V * Λ represents the transpose matrix; Λ represents the eigenvalue matrix; * represents the projection operation.

[0134] In one example of the present invention, the expression of the discrete state-space equation is as follows:

[0135]

[0136] In the formula, h k and y k These represent the memory control unit and the output control unit, respectively, for selecting the current state in the state-space architecture; h k-1h0 represents the memory control unit that selects the previous state in the state space architecture; h0 represents the initialization state of the memory control unit; x k x0 represents the data of the current state input architecture; x0 represents the data of the initial state input architecture. All are matrices after discrete adaptation of A, B, and C.

[0137] In one example of the invention, the matrix It is obtained from the following derivation formula:

[0138]

[0139] In the formula, Δ represents the discretized nonlinear function, which discretizes the continuous state; Ι represents the identity matrix.

[0140] In one example of the present invention, in step S50, the formula for calculating the deviation using the test set CT detection data is as follows:

[0141]

[0142] In the formula, MSE represents the deviation calculation equation; M represents the number of samples; i represents the index of the crack parameter; F M,pred F represents the predicted value of the Mth sample; M,true The value represents the true value of the Mth sample.

[0143] According to a second aspect of the present invention, a coal and rock fracture field imaging system based on a multi-target joint guidance architecture includes:

[0144] The information acquisition module is configured to deploy CT detection sensors and acoustic emission monitoring sensors on coal and rock samples, and place the coal and rock samples in a testing machine for loading tests. The acoustic emission monitoring sensors and CT detection sensors acquire full-time and spatiotemporal acoustic emission monitoring data and stage-by-stage CT detection data of the coal and rock samples during the loading process. Among them, the acoustic emission monitoring data is full-time and spatiotemporal acoustic emission parameter data, and the CT detection data is stage-by-stage full-space fracture parameter data.

[0145] The data partitioning module is configured to map the CT detection data and all-time acoustic emission monitoring data at different stages during the loading process of coal and rock samples according to time, obtain the fracture parameter dataset, and divide the fracture parameter dataset into training set, validation set and test set;

[0146] The neural network architecture building module is configured to build a multi-objective jointly guided neural network architecture based on a learnable gap distribution module.

[0147] The coal and rock three-dimensional fracture field imaging module is configured to use a multi-objective joint guided neural network architecture to extract features from the fracture parameter data and corresponding acoustic emission parameter data of the training set and validation set in the fracture parameter dataset, so as to obtain the coal and rock three-dimensional fracture field imaging model.

[0148] The parameter adjustment module is configured to input acoustic emission parameter data from the test set into the three-dimensional fracture field imaging model of coal and rock to obtain the three-dimensional fracture field imaging results of coal and rock, and determine whether the accuracy of the three-dimensional fracture field imaging results of coal and rock meets the requirements. When the accuracy of the three-dimensional fracture field imaging results of coal and rock meets the requirements, the calculation ends; when the accuracy of the three-dimensional fracture field imaging results of coal and rock does not meet the requirements, the deviation is calculated using the CT detection data of the test set and the neural network parameters are adjusted until the accuracy of the three-dimensional fracture field imaging results of coal and rock meets the requirements.

[0149] This imaging system focuses on achieving fracture field imaging based on acoustic emission monitoring data during coal and rock sample loading tests using a coal and rock fracture field imaging method and system based on a multi-objective joint guidance architecture. Compared with traditional numerical simulation and neural network methods, this coal and rock fracture field imaging method based on a multi-objective joint guidance architecture guides the extraction process of fracture distribution features from a learnable fracture distribution module by using fracture parameter prediction mechanism equations, joint training equations, fracture distribution reconstruction equations, and free space equations. This ensures that the output of the guiding model follows physical laws, fully exploits the nonlinear features in multi-source experimental data, and guarantees the accuracy of coal and rock fracture imaging results, thus constructing a new paradigm for coal and rock fracture field imaging.

[0150] The learnable fracture distribution module in this imaging system is based on a neural network with a selective state-space architecture. This architecture significantly reduces computational resource dependence through fine-grained partitioning and filtering of state and observation variables. Furthermore, the neural network utilizes a multi-matrix parametric method to efficiently extract high-dimensional features from CT detection and acoustic emission monitoring data during coal and rock sample loading. This allows the neural network to effectively learn acoustic emission and fracture parameter features, providing high-value data features for coal and rock fracture imaging.

[0151] This imaging system incorporates fracture distribution reconstruction equations and free-space equations into the fracture imaging process of the learnable fracture distribution module to guide the imaging process, thereby improving the accuracy of coal and rock fracture imaging results. The fracture distribution reconstruction equations integrate mechanistic equation knowledge, which not only gives the fracture field imaging method strong generalization ability but also enhances its interpretability. Simultaneously, the coal and rock fracture field imaging method based on a multi-objective joint guidance architecture constrains the fracture location prediction process through free-space equations, achieving accurate imaging of coal and rock fractures during sample loading.

[0152] It should be noted that the coal and rock fracture field imaging system based on the multi-target joint guidance architecture of the present invention can also perform any of the processing described in the coal and rock fracture field imaging method based on the multi-target joint guidance architecture previously, and the specific details are not repeated here.

[0153] It should be noted that the coal and rock fracture field imaging system based on the multi-target joint guidance architecture of the present invention can also perform any of the processing described in the previously described coal and rock fracture field imaging method based on the multi-target joint guidance architecture, and the specific details are not repeated here.

[0154] The foregoing description, with reference to preferred embodiments, details the exemplary implementation of the coal and rock fracture field imaging method and system based on a multi-target joint guidance architecture proposed in this invention. However, those skilled in the art will understand that various modifications and alterations can be made to the above specific embodiments without departing from the concept of this invention, and various combinations can be made to the various technical features and structures proposed in this invention without exceeding the protection scope of this invention, which is determined by the appended claims.

Claims

1. A coal and rock fracture field imaging method based on a multi-objective joint guidance architecture, characterized in that, Includes the following steps: S10: CT detection sensors and acoustic emission monitoring sensors are arranged on the coal and rock sample, and the coal and rock sample is placed in the testing machine for loading test. Acoustic emission monitoring data and CT detection data at different stages are collected by the acoustic emission monitoring sensor and CT detection sensor during the loading process of the coal and rock sample; among them, the acoustic emission monitoring data is the acoustic emission parameter data at all times and space, and the CT detection data is the fracture parameter data at different stages and space. S20: The CT detection data and all-time acoustic emission monitoring data at different moments during the loading process of the coal and rock samples are correlated according to time to obtain the fracture parameter dataset, and the fracture parameter dataset is divided into training set, validation set and test set; S30: Constructing a multi-objective joint guided neural network architecture based on a learnable crack distribution module; S40: The multi-objective joint guided neural network architecture extracts features from the fracture parameter data and corresponding acoustic emission parameter data of the training set and validation set in the fracture parameter dataset to obtain a three-dimensional fracture field imaging model of coal and rock. S50: Input the acoustic emission parameter data from the test set into the three-dimensional fracture field imaging model of coal and rock to obtain the three-dimensional fracture field imaging results of coal and rock. Determine whether the accuracy of the three-dimensional fracture field imaging results of coal and rock meets the requirements. If the accuracy of the three-dimensional fracture field imaging results of coal and rock does not meet the requirements, calculate the deviation using the CT detection data of the test set and adjust the neural network parameters until the accuracy of the three-dimensional fracture field imaging results of coal and rock meets the requirements.

2. The coal and rock fracture field imaging method based on a multi-target joint guidance architecture according to claim 1, characterized in that, Step S30 further includes: constraining the training process of the multi-objective jointly guided neural network architecture through the crack parameter prediction mechanism equation and the joint training constraint equation, specifically including the following steps: First, the training process of the multi-objective jointly guided neural network architecture is constrained by the crack parameter prediction mechanism equation, whereby the expression of the crack parameter prediction mechanism equation is as follows: In the formula, The equation representing the prediction mechanism of the true three-dimensional fracture location; Represents the interaction between different acoustic emission sensors The total number of combinations Represents the total number of acoustic emission sensors; , , These represent the crack location deviation parameters predicted by the neural network architecture; , , These represent the coordinates of the acoustic emission microcrack event in the x, y, and z directions, respectively. , , These represent the coordinate values ​​of the acoustic emission sensor in the x, y, and z directions, respectively. Represents P-wave velocity; Representing the i and the j The difference in positioning time between the sensors; In the formula, The equation representing the prediction mechanism of real three-dimensional fracture parameters; Represents the total number of acoustic emission sensors; Represents the weighting coefficients of acoustic emission parameters; Representing the i Acoustic emission parameters obtained by monitoring by an acoustic emission sensor; This represents the fracture parameters that need to be predicted. Then, the training process of the multi-objective jointly guided neural network architecture is constrained by the joint training constraint equation, the expression of which is as follows: In the formula, Represents the joint training constraint equation; The objective is to find the minimum value of the joint training constraint equations; denoted by , represents the latent vector of three-dimensional fracture distribution corresponding to different fracture distributions in the dataset; W represents the number of three-dimensional fracture distributions; i represents the index of the latent vector of three-dimensional fracture distribution. The total number of moments in the training set where the cracks are distributed at different times; This represents the total number of locations where cracks are distributed across different positions in the training set. Representative acoustic emission microcrack event location parameters; Represents three-dimensional fracture parameters; Represents data-driven neural networks for acoustic emission parameters The predicted parameters of the three-dimensional fracture; The parameter representing the fracture location deviation obtained by the neural network architecture through the prediction of acoustic emission micro-fracture event coordinates; Represents the potential vector of three-dimensional crack distribution; This represents all trainable parameters of a data-driven neural network. Representative data-driven neural network for acoustic emission microcrack events The predicted location coordinates of the three-dimensional crack; Represents the deviation calculation function. , This represents the region where acoustic emission microcrack events accumulate; The prior distribution variance represents the potential vector of the three-dimensional crack distribution; This represents the squared norm of the latent vector.

3. The coal and rock fracture field imaging method based on a multi-target joint guidance architecture according to claim 1, characterized in that, Step S40 further includes constraining the testing process of the three-dimensional fracture field imaging model of coal and rock through the fracture distribution reconstruction equation and the free space constraint equation, specifically including the following steps: First, the prediction process of the three-dimensional fracture distribution in the three-dimensional fracture field imaging model of coal and rock is constrained by the fracture distribution reconstruction equation. The expression of the fracture distribution reconstruction equation is as follows: In the formula, The three-dimensional crack distribution obtained by inference from a multi-objective jointly guided neural network architecture; A dataset representing the coordinates of acoustic emission microcrack events and the actual three-dimensional crack coordinates; Global distribution properties representing the distribution of three-dimensional cracks; Then, the prediction process of three-dimensional fracture parameters in the three-dimensional fracture field imaging model of coal and rock is constrained by the free space constraint equation, wherein the expression of the free space constraint equation is as follows: In the formula, Represents the free space constraint equations; A set representing the distribution of fractures in the three-dimensional space of a coal and rock sample; represent The distribution of cracks at a certain location in the structure is predicted using three-dimensional crack distribution. Errors in the prediction process require additional constraints for correction. This represents the range of error.

4. The coal and rock fracture field imaging method based on a multi-target joint guidance architecture according to claim 1, characterized in that, In step S40, the multi-objective joint guided neural network architecture extracts features from the fracture parameter data and corresponding acoustic emission parameter data of the training and validation sets in the fracture parameter dataset, including the following steps: S41: Transmit acoustic emission parameters AE through the data encoding layer p Data extraction yields parameter-enhanced acoustic emission features AEf se ; S42: The encoded data is concatenated to the feature extraction layer B through a residual connection layer to obtain feature AE. b middle; S43: Acoustic emission feature AEf enhanced by the first Normalization function se Regularization is performed to obtain the feature AE n ; S44: Regularized Feature AE n Feature extraction AE is obtained through the residual connection formed by the second linear feature extraction layer A and the second activation function layer. a+s The feature AE spliced ​​into the output of the selective state space architecture sssm Get AE SSSM ; S45: Feature AE output by the first linear feature extraction layer A a Higher-dimensional features (AE) are obtained through feature extraction via convolutional layers and a first activation function layer. conv+s ; S46: High-dimensional features are gated using a selective state-space architecture and concatenated with features extracted from convolutional layers and the first activation function layer to obtain feature AE. ssSM ; S47: Linear feature extraction layer B extracts features from the concatenated features and concatenates the extracted features with the features of the encoded data obtained by the residual connection layer to obtain feature AFf. b+se ; S48: Regularize the concatenated features using the second Normalization function to obtain feature AEf. n ; S49: The features extracted by the fully connected layer are used for calculation to obtain the fracture location deviation parameter, which is the difference (varx) between the acoustic emission micro-fracture event coordinates predicted by the neural network architecture and the actual three-dimensional fracture coordinates in the x, y, z directions. M,pred ,vary M,pred ,varz M,pred ) and fracture parameter F M,pred .

5. The coal and rock fracture field imaging method based on a multi-target joint guidance architecture according to claim 4, characterized in that, In step S46, the selective state-space architecture is specifically implemented as follows: First, a state-space architecture is constructed using continuous state-space equations, the expressions of which are as follows: h t =Ah t-1 +Bx t the t =Ch t Among them, h t-1 Represents the system state; h t Represents the updated system state; matrix A represents the state transition matrix; matrix B represents the input gating matrix; matrix C represents the output mapping matrix; x t The data represents the current state of the input architecture. Secondly, the state-space architecture is subjected to discrete data adaptation using discrete state-space equations; the equation for the state transition matrix A is as follows: In the formula, n is the row index related to the orthogonal basis functions in the polynomial space, and k is the column index related to the orthogonal basis functions in the polynomial space.

6. The coal and rock fracture field imaging method based on a multi-target joint guidance architecture according to claim 5, characterized in that, This also includes: randomly initializing the state transition matrix A and using a low-rank matrix obtained from matrix decomposition. The equation for the low-rank matrix is ​​as follows: Where P and Q are both vectors of length N; V represents the eigenvector matrix; V * Λ represents the transpose matrix; Λ represents the eigenvalue matrix; * represents the projection operation.

7. The coal and rock fracture field imaging method based on a multi-target joint guidance architecture according to claim 5, characterized in that, The expression for the discrete state-space equation is as follows: In the formula, h k and y k These represent the memory control unit and the output control unit, respectively, for selecting the current state in the state-space architecture; h k-1 h0 represents the memory control unit that selects the previous state in the state space architecture; h0 represents the initialization state of the memory control unit; x k x0 represents the data of the current state input architecture; x0 represents the data of the initial state input architecture. All are matrices after discrete adaptation of A, B, and C.

8. The coal and rock fracture field imaging method based on a multi-target joint guidance architecture according to claim 7, characterized in that, matrix It is obtained from the following derivation formula: In the formula, Δ represents the discretized nonlinear function, which discretizes the continuous state; Ι represents the identity matrix.

9. The coal and rock fracture field imaging method based on a multi-target joint guidance architecture according to claim 1, characterized in that, In step S50, the formula for calculating the deviation using the test set CT detection data is as follows: In the formula, RMSE represents the deviation calculation equation; M represents the number of samples; i represents the index of the crack parameter; F M,pred F represents the predicted value of the Mth sample; M,true The value represents the true value of the Mth sample.

10. A coal and rock fracture field imaging system based on a multi-target joint guidance architecture, characterized in that, include: The information acquisition module is configured to deploy CT detection sensors and acoustic emission monitoring sensors on coal and rock samples, and place the coal and rock samples in a testing machine for loading tests. The acoustic emission monitoring sensors and CT detection sensors acquire full-time and space-time acoustic emission monitoring data and stage-by-stage CT detection data of the coal and rock samples during the loading process. Among them, the acoustic emission monitoring data is full-time and space-time acoustic emission parameter data, and the CT detection data is stage-by-stage full-space fracture parameter data. The data partitioning module is configured to map the CT detection data and all-time acoustic emission monitoring data at different stages during the loading process of coal and rock samples according to time, obtain the fracture parameter dataset, and divide the fracture parameter dataset into training set, validation set and test set; The neural network architecture building module is configured to build a multi-objective jointly guided neural network architecture based on a learnable gap distribution module. The coal and rock three-dimensional fracture field imaging module is configured to use a multi-objective joint guided neural network architecture to extract features from the fracture parameter data and corresponding acoustic emission parameter data of the training set and validation set in the fracture parameter dataset, so as to obtain the coal and rock three-dimensional fracture field imaging model. The parameter adjustment module is configured to input acoustic emission parameter data from the test set into the three-dimensional fracture field imaging model of coal and rock to obtain the three-dimensional fracture field imaging results of coal and rock, determine whether the accuracy of the three-dimensional fracture field imaging results of coal and rock meets the requirements, and when the accuracy of the three-dimensional fracture field imaging results of coal and rock does not meet the requirements, it calculates the deviation using CT detection data from the test set and adjusts the neural network parameters until the accuracy of the three-dimensional fracture field imaging results of coal and rock meets the requirements.

Images