Coal rock mass fracture field inversion method and system capable of explaining and selecting state space architecture

Abstract

The invention discloses a coal and rock mass fracture field inversion method and system capable of explaining and selecting a state space framework. The inversion method comprises the following steps of obtaining staged time CT detection data and full space-time acoustic emission monitoring data in a coal and rock mass loaded fracture process; performing correspondence on the CT detection data at the stage moments and the whole space-time acoustic emission monitoring data according to the moments to obtain a fracture parameter data set; performing feature extraction on full-space fracture parameter data and acoustic emission monitoring data of a training set and a verification set in the fracture parameter data set based on a neural network of a selective state space architecture to obtain an inversion model; all fracture parameters in a full space-time fracture field in the loading process are obtained through full space-time acoustic emission monitoring data inversion; and inputting scalar parameter data of the middle test set and an inversion result obtained by the inversion model into an SHAP interpreter to obtain an influence factor between acoustic emission scalar parameter inversion results and fracture parameter inversion results, and further obtaining a fracture parameter real-time quantitative inversion mechanism model.

Description

Technical Field

[0001] The present invention relates to the technical field of coal-rock monitoring, and in particular to a coal-rock mass fracture field inversion method and system capable of interpreting a selected state space architecture. Background Art

[0002] During coal mining, factors such as mining activities will destroy the original stress equilibrium state of coal and rock mass. This process usually involves rapid and unstable expansion of cracks, which is the root cause of dynamic disasters such as rock burst, coal and gas outbursts, etc. Therefore, in order to deeply understand the mechanism of coal and rock dynamic disasters and realize early warning of disasters, it is necessary to carry out in-depth research on the distribution and evolution characteristics of the fracture field.

[0003] Although existing rock mechanics theories and methods attempt to explain the evolution of coal rock fracture fields caused by mining activities, this complex process is still regarded as a "black box" that is difficult to measure and reveal. There are still many challenges in real-time monitoring of fracture fields, accurate quantitative inversion, exploration of their formation mechanisms, and revealing their development and evolution processes. At present, industrial CT detection and numerical simulation methods have made certain progress. These technologies can be used to reconstruct the internal structure of coal rock masses, especially in constructing three-dimensional digital models of discontinuous structures and fracture fields. For example, patent CN108072467A discloses a method for measuring the internal stress field of a discontinuous structure, patent CN114187423A discloses a method, electronic equipment and storage medium for reconstructing surrounding rock fractures in three-dimensional simulation tests, and patent CN114169182A discloses a method and equipment for reconstructing an ellipsoid model of rock face fractures. Despite this, these methods still have some limitations. They usually rely on high-standard laboratory conditions and cumbersome model reconstruction methods, which limits their wide application in actual engineering. In addition, these methods cannot provide the real-time dynamic spatiotemporal evolution characteristics of the fracture field distribution, which may cause certain limitations in the actual production process.

[0004] In recent years, some experimental methods combining exploration-monitoring technology with numerical simulation have been applied in laboratories. However, the current use of these exploration-monitoring data is still mainly focused on manually analyzing their temporal variation patterns. In addition, relying solely on numerical simulation for fracture field inversion requires extremely high computational costs, and even minor changes in the model are likely to cause significant negative impacts during the model calculation process. Therefore, how to use exploration-monitoring data to achieve full-time and space fracture field inversion and make up for the limitations of traditional methods still requires further research. With the continuous advancement and development of science and technology, the breakthrough progress of neural network methods has attracted people's attention in fracture field inversion. In particular, the use of neural networks and SHAP interpreters based on selective state space architecture can effectively extract the features of different types of data, significantly improve computational efficiency and reduce the demand for computing resources compared with traditional neural network architectures. This provides the possibility of using neural network methods to achieve accurate inversion, real-time reproduction and mechanism interpretation of loaded coal and rock fracture parameters, but there is still a lack of specific feasible ideas and implementation methods. Therefore, it is urgently necessary to carry out actual measurement of detection and monitoring signals during the loading process of coal and rock masses, propose a coal and rock fracture field inversion method based on an interpretable selective state space architecture, realize the full-time and space fracture field inversion based on full-time and space acoustic emission monitoring data, and establish a real-time quantitative inversion mechanism model of fracture parameters based on acoustic emission data, which will play a significant role in revealing the "black box" evolution process of coal and rock dynamic disasters and realizing the monitoring and early warning of coal and rock dynamic disasters. Summary of the invention

[0005] In response to the problems and needs raised above, this solution proposes a coal rock fracture field inversion method and system that can interpret the selected state space architecture. Due to the adoption of the following technical features, it can achieve the above technical objectives and bring about many other technical effects.

[0006] An object of the present invention is to propose a coal rock mass fracture field inversion method that can interpret a selected state space framework, comprising the following steps:

[0007] S10: arranging CT detection sensors and acoustic emission monitoring sensors on the coal-rock mass, and placing the coal-rock mass in a testing machine for a load fracture test, and collecting CT detection data at staged moments and full-time and full-space acoustic emission monitoring data during the process of the coal-rock mass being loaded and fractured by the CT detection sensors and the acoustic emission monitoring sensors; wherein the CT detection data is full-space fracture parameter data at staged moments, and the acoustic emission monitoring data includes waveform data and scalar parameter data;

[0008] S20: Corresponding the CT detection data at the stage in the process of coal rock mass load fracture and the full-time and space acoustic emission monitoring data according to the time, obtaining a fracture parameter data set, and dividing the fracture parameter data set into a training set, a verification set and a test set;

[0009] S30: A neural network based on a selective state space architecture is used to extract features of the full-space fracture parameter data, waveform data at the corresponding time, and scalar parameter data at the corresponding time of the training set and the verification set in the fracture parameter data set to obtain a real-time quantitative inversion model of fracture parameters; the real-time quantitative inversion model of fracture parameters obtains the fracture parameters in the full-time and space fracture field during the loading process by inverting the full-time and space acoustic emission monitoring data, that is, the evolution process of the full-time and space fracture field parameters of the coal rock mass under loading is obtained;

[0010] S40: The full-time and space-time acoustic emission scalar parameter data of the test set in the fracture parameter data set and the full-time and space-time fracture parameter inversion results obtained by the fracture parameter real-time quantitative inversion model are input into the SHAP interpreter to obtain the influencing factors between the acoustic emission scalar parameters and the fracture parameter inversion results, and then obtain the fracture parameter real-time quantitative inversion mechanism model based on the acoustic emission scalar parameters.

[0011] In addition, the coal-rock mass fracture field inversion method of the interpretable selective state space framework according to the present invention may also have the following technical features:

[0012] In one example of the present invention, in step S30, the neural network based on the selective state space architecture performs feature extraction on the stage-by-stage full-space crack parameter data, the corresponding moment waveform data, and the corresponding moment scalar parameter data of the training set and the validation set in the crack parameter data set, including the following steps:

[0013] S31: The acoustic emission scalar parameters are embedded into the waveform data through the data coding layer to obtain the parameter-enhanced acoustic emission data, and the parameter-enhanced acoustic emission features AEf are extracted. se ;

[0014] S32: The encoded data is concatenated to the feature extraction layer B through residual connection to obtain the feature AE b middle;

[0015] S33: Enhance the acoustic emission characteristics AEf through the first Normalization function se Regularize to get feature AE n ;

[0016] S34: Regularized feature AE n The feature extraction is performed through the residual connection composed of the second linear feature extraction layer A and the second activation function layer to obtain AE a+s Feature AE concatenated to the output of the selective state-space architecture sssm Get AE SSSM ;

[0017] S35: Feature AE output by the first linear feature extraction layer Aa Feature extraction is performed through the convolution layer and the first activation function layer to obtain a higher-dimensional feature AE conv+s ;

[0018] S36: The high-dimensional features are gated by the selective state space architecture, and the features extracted by the residual connection formed by the second linear feature extraction layer A and the second activation function layer are concatenated to obtain the feature AE SSSM ;

[0019] S37: The linear feature extraction layer B extracts the concatenated features and concatenates the extracted features with the features of the encoded data by the residual connection layer to obtain the feature AEf b+se ;

[0020] S38: Regularize the concatenated features through the second Normalization function to obtain feature AEf n ;

[0021] S39: The features extracted by the fully connected layer are calculated to obtain the crack parameter F N .

[0022] In an example of the present invention, in step S36, the selective state space architecture is specifically implemented as follows:

[0023] First, a state space framework is constructed by continuous state space equations, wherein the expression of the continuous state space equations is as follows:

[0024] h t =Ah t-1 +Bx t

[0025] y t =Ch t

[0026] Among them, h t-1 Represents the system status; h t represents the updated system state; matrix A represents the state transition matrix; B represents the input gating matrix; C represents the output mapping matrix; x t Data representing the current state of the input schema.

[0027] Secondly, the state space architecture is adapted to discrete data by discrete state space equations; the equation expression of the state transfer matrix A is as follows:

[0028]

[0029] Where n is the row index associated with the orthogonal basis function in the polynomial space, and k is the column index associated with the orthogonal basis function in the polynomial space.

[0030] In one example of the present invention, the process further includes: randomly initializing the state transfer matrix A, and using a low-rank matrix after matrix decomposition. The low-rank matrix equation formula is as follows:

[0031]

[0032] Where P and Q are both vectors of length N; V represents the eigenvector matrix; V * represents the transposed matrix; Λ represents the characteristic matrix; * represents the projection operation.

[0033] In one example of the present invention, the expression of the discrete state space equation is as follows:

[0034]

[0035] In the formula, h k and k They represent the memory control unit and output control unit of the current state in the state space architecture; h k-1 represents the memory control unit of the previous state in the state space architecture; h0 represents the initialization state of the memory control unit; x k Represents the data of the current state input architecture; x0 represents the data of the initial state input architecture; They are all matrices after discrete adaptation of A, B, and C.

[0036] In one example of the present invention, the matrix It is derived from the following formula:

[0037]

[0038] Where Δ represents the discretized nonlinear function, which discretizes the continuous state; Ι represents the unit matrix.

[0039] In an example of the present invention, in step S40, the expression formula of the SHAP interpreter is as follows:

[0040]

[0041] in, represents the impact factor; M represents the number of input acoustic emission scalar parameters; represents the set of all permutations and combinations of M features, represents the set of all feature combinations before feature k in the permutation R, f x (S) = E(f(X)|Xs =x S ) represents the conditional expectation when the feature subset S is known; i represents the influencing factor number.

[0042] In one example of the present invention, the influencing factors, acoustic emission scalar parameters and crack parameters obtained by the SHAP interpreter constitute a mechanism model for real-time quantitative inversion of crack parameters, and the expression equation is as follows:

[0043]

[0044] Among them, MM represents the mechanism model for real-time quantitative inversion of fracture parameters; F N Represents crack parameters; AE si Represents the acoustic emission scalar parameter.

[0045] Another object of the present invention is to provide a coal rock mass fracture field inversion system capable of interpreting a selected state space framework, comprising:

[0046] The information acquisition module is configured to arrange CT detection sensors and acoustic emission monitoring sensors on the coal rock mass, and place the coal rock mass in a testing machine for a load fracture test, and acquire CT detection data at staged moments and full-time and space acoustic emission monitoring data during the process of coal rock mass load fracture through the CT detection sensors and acoustic emission monitoring sensors; wherein the CT detection data is full-space fracture parameter data at staged moments, and the acoustic emission monitoring data includes waveform data and scalar parameter data;

[0047] A data division module is configured to correspond the stage-by-stage CT detection data and the full-time and space acoustic emission monitoring data in the process of coal rock mass load fracture according to time, obtain a fracture parameter data set, and divide the fracture parameter data set into a training set, a verification set and a test set;

[0048] The full-time and space-time fracture parameter quantitative inversion module is configured to extract features of the full-time and space-time fracture parameter data, the waveform data at the corresponding time, and the scalar parameter data at the corresponding time of the training set and the verification set in the fracture parameter data set based on the neural network of the selective state space architecture, so as to obtain a real-time and quantitative inversion model of fracture parameters; the real-time and quantitative inversion model of fracture parameters obtains the fracture parameters of the full-time and space-time fracture field during the loading process by inverting the full-time and space-time acoustic emission monitoring data, that is, obtains the full-time and space-time fracture field parameter evolution process of the coal rock mass under loading;

[0049] The inversion mechanism model module is configured to input the full-time and space-time acoustic emission scalar parameter data of the test set in the fracture parameter data set and the full-time and space-time fracture parameter inversion results obtained by the real-time quantitative inversion model of the fracture parameters into the SHAP interpreter, obtain the influencing factors between the acoustic emission scalar parameters and the fracture parameter inversion results, and then obtain the real-time quantitative inversion mechanism model of the fracture parameters based on the acoustic emission scalar parameters.

[0050] In one example of the present invention, the full-time and space crack parameter quantitative inversion module includes: a data encoding layer, a first Normalization function layer, a first linear feature extraction layer A, a convolution layer, a first activation function layer, a selective state space architecture, a linear feature extraction layer B, a second Normalization function layer and a fully connected layer connected in series; wherein the output end of the data encoding layer and the output end of the linear feature extraction layer B are connected by a residual connection to splice the encoded data to the feature extraction layer B to obtain the feature AE b In the example, a residual connection consisting of a second linear feature extraction layer A and a second activation function layer is connected in series between the output end of the first Normalization function layer and the output end of the selective state space architecture;

[0051] The data encoding layer is configured to embed the acoustic emission scalar parameter into the waveform data to obtain the parameter-enhanced acoustic emission data, and extract the parameter-enhanced acoustic emission feature AEf se ;

[0052] The first Normalization function layer is configured to enhance the acoustic emission feature AEf se Regularize to get feature AE n ;

[0053] The residual connection configuration composed of the second linear feature extraction layer A and the second activation function is used to extract the feature AE a+s Feature AE concatenated to the output of the selective state-space architecture sssm Get AE SSSM ;

[0054] The convolution layer and the first activation function layer are configured to extract the feature AE output by the first linear feature extraction layer A a Perform feature extraction to obtain higher-dimensional feature AE conv+s ;

[0055] The selective state space architecture is configured to perform gated calculations on high-dimensional features and concatenate the features extracted by the residual connection formed by the second linear feature extraction layer A and the second activation function layer to obtain the feature AE SSSM ;

[0056] The linear feature extraction layer B is configured to extract features from the concatenated features, and concatenate the extracted features with the features of the data encoded by the residual connection layer to obtain the feature AEf b+se ;

[0057] The second Normalization function layer is configured to regularize the concatenated features to obtain the feature AEf n ;

[0058] The fully connected layer is configured to calculate the extracted features to obtain the crack parameter F N .

[0059] Compared with the prior art, the present invention has the following beneficial effects:

[0060] 1. The present invention focuses on the neural network method to invert the full-time and space crack field of the loaded coal-rock mass through the detection-monitoring information in the coal-rock mass load fracture experiment, combining the neural network based on the selective state space architecture and the SHAP interpreter to achieve the full-time and space reproduction of the loaded coal-rock mass crack field, and reveal the real-time quantitative inversion mechanism model of the crack parameters. Compared with traditional numerical simulation and neural network methods, this coal-rock spatiotemporal crack field inversion method based on interpretable deep learning has higher computational efficiency and accuracy, and can effectively make up for the limitations of traditional methods.

[0061] 2. The real-time quantitative inversion model of fracture parameters in the present invention is obtained by extracting spatiotemporal features of the data in the fracture parameter data set through a neural network based on a selective state space architecture, and a feature extraction method for fracture parameter data and acoustic emission data during the coal-rock mass loading process is proposed. Based on the proposed spatiotemporal feature extraction method, the full-time and spatiotemporal fracture parameters can be inverted through the full-time and spatiotemporal acoustic emission data obtained in the laboratory during the coal-rock mass loading process.

[0062] 3. The model inversion mechanism in the present invention extracts features from the acoustic emission scalar parameter data and the full-time and space crack parameter inversion results through an interpreter to obtain the influencing factors in the mechanism model. The interpreter can obtain a real-time quantitative inversion mechanism model of crack parameters during the coal-rock mass loading process through the acoustic emission data and the model's prediction results, revealing the black box process of crack parameter inversion during the coal-rock mass loading process.

[0063] The best embodiment for carrying out the present invention will be described in more detail below with reference to the accompanying drawings so that the features and advantages of the present invention can be easily understood. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings of the embodiments of the present invention, wherein the drawings are only used to illustrate some embodiments of the present invention, but not to limit all embodiments of the present invention thereto.

[0065] Figure 1 A flowchart of a coal rock mass fracture field inversion method based on an interpretable selective state space framework according to an embodiment of the present invention;

[0066] Figure 2 It is a schematic diagram of the arrangement structure of coal rock mass in a testing machine according to an embodiment of the present invention;

[0067] Figure 3 Schematic diagram of the structure of a real-time quantitative inversion model for fracture parameters according to an embodiment of the present invention.

[0068] List of reference numerals:

[0069] Testing machine 100;

[0070] Loading device 110;

[0071] Acoustic emission sensor 120;

[0072] Coupler 130;

[0073] Preamplifier 140;

[0074] X-ray detector 150;

[0075] X-ray source 160;

[0076] Coal and rock mass 200;

[0077] Horizontal direction X;

[0078] Longitudinal direction Y. DETAILED DESCRIPTION

[0079] In order to make the purpose, technical solution and advantages of the technical solution of the present invention clearer, the technical solution of the embodiment of the present invention will be clearly and completely described in conjunction with the drawings of specific embodiments of the present invention. The same figure marks in the drawings represent the same parts. It should be noted that the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the described embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0080] Unless otherwise defined, the technical terms or scientific terms used herein shall have the usual meanings understood by persons of ordinary skill in the field to which the present invention belongs. The words "first", "second" and similar words used in the patent application specification and claims of the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. Similarly, words such as "one" or "one" do not necessarily indicate a quantity limitation. Words such as "include" or "comprise" mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. Words such as "connect" or "connected" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0081] According to the first aspect of the present invention, a coal rock mass fracture field inversion method capable of interpreting a selected state space framework is provided. Figure 1 As shown, the following steps are included:

[0082] S10: arranging CT detection sensors and acoustic emission monitoring sensors on the coal-rock mass, and placing the coal-rock mass in a testing machine for a load fracture test, and collecting CT detection data at staged moments and full-time and full-space acoustic emission monitoring data during the process of coal-rock mass load fracture through the CT detection sensors and acoustic emission monitoring sensors; wherein the CT detection data is full-space fracture parameter data at staged moments, and the acoustic emission monitoring data includes waveform data and scalar parameter data; specifically, Figure 2As 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 box 170; wherein the loading device 110, the acoustic emission sensor 120, the coupler 130, the preamplifier 140, the X-ray detector 150 and the X-ray source 160 are all installed in the box 170, and the coal rock body 200 is arranged in the loading device 110, and the acoustic emission sensors 120 are installed on both sides of the transverse direction X of the coal rock body 200, wherein the acoustic emission sensor 120 is electrically connected to the coupler 130 and to the preamplifier 140, and the X-ray detector 150 and the X-ray source 160 are respectively arranged on both sides of the longitudinal direction Y of the coal rock body 200, 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 rock mass 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 fracture process of the coal rock mass 200. At the same time, the acoustic emission sensor 120 obtains full-time and space-time acoustic emission monitoring data.

[0083] S20: Corresponding the CT detection data at the stage in the process of coal rock mass load fracture and the full-time and space acoustic emission monitoring data according to the time, obtaining a fracture parameter data set, and dividing the fracture parameter data set into a training set, a verification set and a test set;

[0084] S30: A neural network based on a selective state space architecture is used to extract features of the full-space fracture parameter data, waveform data at the corresponding time, and scalar parameter data at the corresponding time of the training set and the verification set in the fracture parameter data set to obtain a real-time quantitative inversion model of fracture parameters; the real-time quantitative inversion model of fracture parameters obtains the fracture parameters in the full-time and space fracture field during the loading process by inverting the full-time and space acoustic emission monitoring data, that is, the evolution process of the full-time and space fracture field parameters of the coal rock mass under loading is obtained;

[0085] S40: The full-time and space-time acoustic emission scalar parameter data of the test set in the fracture parameter data set and the full-time and space-time fracture parameter inversion results obtained by the fracture parameter real-time quantitative inversion model are input into the SHAP interpreter to obtain the influencing factors between the acoustic emission scalar parameters and the fracture parameter inversion results, and then obtain the fracture parameter real-time quantitative inversion mechanism model based on the acoustic emission scalar parameters.

[0086] This coal-rock fracture field inversion method focuses on the neural network method to invert the full-time and space fracture field of the loaded coal-rock mass through the detection-monitoring information in the coal-rock mass load fracture experiment, combining the neural network based on the selective state space architecture and the SHAP interpreter to achieve full-time and space reproduction of the loaded coal-rock fracture field and reveal the real-time quantitative inversion mechanism model of fracture parameters. Compared with traditional numerical simulation and neural network methods, this coal-rock spatiotemporal fracture field inversion method based on interpretable deep learning has higher computational efficiency and accuracy, and can effectively make up for the limitations of traditional methods.

[0087] The real-time quantitative inversion model of fracture parameters in the coal-rock fracture field inversion method is obtained by extracting the spatiotemporal features of the data in the fracture parameter data set through a neural network based on a selective state space architecture, and a feature extraction method for fracture parameter data and acoustic emission data during the coal-rock loading process is proposed. Based on the proposed spatiotemporal feature extraction method, the full-time and spatiotemporal fracture parameters can be inverted through the full-time and spatiotemporal acoustic emission data obtained in the laboratory during the coal-rock loading process.

[0088] The model inversion mechanism in the coal-rock mass fracture field inversion method extracts features from the acoustic emission scalar parameter data and the full-time and space fracture parameter inversion results through an interpreter to obtain the influencing factors in the mechanism model. The interpreter can obtain the real-time quantitative inversion mechanism model of the fracture parameters of the coal-rock mass during the loading process through the acoustic emission data and the prediction results of the model, revealing the black box process of fracture parameter inversion of the coal-rock mass during the loading process.

[0089] In one example of the present invention, Figure 3 As shown, in step S30, the neural network based on the selective state space architecture performs feature extraction on the stage-by-stage full-space crack parameter data, the corresponding moment waveform data and the corresponding moment scalar parameter data of the training set and the verification set in the crack parameter data set, including the following steps:

[0090] S31: The acoustic emission scalar parameters are embedded into the waveform data through the data coding layer to obtain the parameter-enhanced acoustic emission data, and the parameter-enhanced acoustic emission features AEfs are extracted. se ;

[0091] S32: The encoded data is concatenated to the feature extraction layer B through residual connection to obtain the feature AE b middle;

[0092] S33: Parameter enhancement of acoustic emission features AEfs through the first Normalization function se Regularize to get feature AE n ;

[0093] S34: Regularized feature AE nThe feature extraction is performed through the residual connection composed of the second linear feature extraction layer A and the second activation function layer (σ2) to obtain AE a+s Feature AE concatenated to the output of the selective state-space architecture sssm Get AE SSSM ;

[0094] S35: Feature AE output by the first linear feature extraction layer A a Feature extraction is performed through the convolution layer and the first activation function layer (σ1) to obtain a higher-dimensional feature AE conv+s ;

[0095] S36: The high-dimensional features are gated by the selective state space architecture, and the features extracted by the residual connection formed by the second linear feature extraction layer A and the second activation function layer are concatenated to obtain the feature AE SSSM ;

[0096] S37: The linear feature extraction layer B extracts the concatenated features and concatenates the extracted features with the features of the encoded data by the residual connection layer to obtain the feature AEf b+se ;

[0097] S38: Regularize the concatenated features through the second Normalization function to obtain feature AEf n ;

[0098] S39: The features extracted by the fully connected layer are calculated to obtain the crack parameter F N .

[0099] In an example of the present invention, in step S36, the selective state space architecture is specifically implemented as follows:

[0100] First, a state space framework is constructed by continuous state space equations, wherein the expression of the continuous state space equations is as follows:

[0101] h t =Ah t-1 +Bx t

[0102] y t =Ch t

[0103] Among them, h t-1 Represents the system status; h t represents the updated system state; matrix A represents the state transition matrix; B represents the input gating matrix; C represents the output mapping matrix; x t Data representing the current state of the input schema.

[0104] Secondly, the state space architecture is adapted to discrete data by discrete state space equations; the equation expression of the state transfer matrix A is as follows:

[0105]

[0106] Where n is the row index associated with the orthogonal basis function in the polynomial space, and k is the column index associated with the orthogonal basis function in the polynomial space.

[0107] In one example of the present invention, in order to simplify the calculation process, it also includes: randomly initializing the state transfer matrix A, and using the low-rank matrix after matrix decomposition, the low-rank matrix equation formula is as follows:

[0108]

[0109] Where P and Q are both vectors of length N; V represents the eigenvector matrix; V * represents the transposed matrix; Λ represents the characteristic matrix; * represents the projection operation.

[0110] In one example of the present invention, the expression of the discrete state space equation is as follows:

[0111]

[0112] In the formula, h k and l They represent the memory control unit and output control unit of the current state in the state space architecture; h l-1 represents the memory control unit of the previous state in the state space architecture; h0 represents the initialization state of the memory control unit; x k Represents the data of the current state input architecture; x0 represents the data of the initial state input architecture; They are all matrices after discrete adaptation of A, B, and C.

[0113] They are all matrices after discrete adaptation of A, B, and C.

[0114] In one example of the present invention, the matrix It is derived from the following formula:

[0115]

[0116] Where Δ represents the discretized nonlinear function, which discretizes the continuous state; Ι represents the unit matrix.

[0117] In an example of the present invention, in step S40, the expression formula of the SHAP interpreter is as follows:

[0118]

[0119] in, represents the impact factor; M represents the number of input acoustic emission scalar parameters; represents the set of all permutations and combinations of M features, represents the set of all feature combinations before feature k in the permutation R, f x (S) = E(f(X)|X s =x s ) represents the conditional expectation when the feature subset S is known; i represents the influencing factor number.

[0120] In one example of the present invention, the influencing factors, acoustic emission scalar parameters and crack parameters obtained by the SHAP interpreter constitute a mechanism model for real-time quantitative inversion of crack parameters, and the expression equation is as follows:

[0121]

[0122] Among them, MM represents the mechanism model for real-time quantitative inversion of fracture parameters; F N Represents crack parameters; AE si Represents the acoustic emission scalar parameter.

[0123] According to the second aspect of the present invention, a coal rock mass fracture field inversion system capable of interpreting a selected state space architecture comprises:

[0124] The information acquisition module is configured to arrange CT detection sensors and acoustic emission monitoring sensors on the coal rock mass, and place the coal rock mass in a testing machine for a load fracture test, and acquire CT detection data at staged moments and full-time and space acoustic emission monitoring data during the process of coal rock mass load fracture through the CT detection sensors and acoustic emission monitoring sensors; wherein the CT detection data is full-space fracture parameter data at staged moments, and the acoustic emission monitoring data includes waveform data and scalar parameter data;

[0125] A data division module is configured to correspond the stage-by-stage CT detection data and the full-time and space acoustic emission monitoring data in the process of coal rock mass load fracture according to time, obtain a fracture parameter data set, and divide the fracture parameter data set into a training set, a verification set and a test set;

[0126] The full-time and space-time fracture parameter quantitative inversion module is configured to extract features of the full-time and space-time fracture parameter data, the waveform data at the corresponding time, and the scalar parameter data at the corresponding time of the training set and the verification set in the fracture parameter data set based on the neural network of the selective state space architecture, so as to obtain a real-time and quantitative inversion model of fracture parameters; the real-time and quantitative inversion model of fracture parameters obtains the fracture parameters of the full-time and space-time fracture field during the loading process by inverting the full-time and space-time acoustic emission monitoring data, that is, obtains the full-time and space-time fracture field parameter evolution process of the coal rock mass under loading;

[0127] The inversion mechanism model module is configured to input the full-time and space-time acoustic emission scalar parameter data of the test set in the fracture parameter data set and the full-time and space-time fracture parameter inversion results obtained by the real-time quantitative inversion model of the fracture parameters into the SHAP interpreter, obtain the influencing factors between the acoustic emission scalar parameters and the fracture parameter inversion results, and then obtain the real-time quantitative inversion mechanism model of the fracture parameters based on the acoustic emission scalar parameters.

[0128] The coal-rock fracture field inversion system focuses on the neural network method to invert the full-time and space fracture field of the loaded coal-rock mass through the detection and monitoring information in the coal-rock mass load fracture experiment, combining the neural network based on the selective state space architecture and the SHAP interpreter to achieve the full-time and space reproduction of the loaded coal-rock fracture field and reveal the real-time quantitative inversion mechanism model of fracture parameters. Compared with traditional numerical simulation and neural network methods, this coal-rock spatiotemporal fracture field inversion method based on interpretable deep learning has higher computational efficiency and accuracy, and can effectively make up for the limitations of traditional methods.

[0129] The real-time quantitative inversion model of fracture parameters in the coal-rock fracture field inversion system is obtained by extracting the spatiotemporal features of the data in the fracture parameter data set through a neural network based on a selective state space architecture, and a feature extraction method for fracture parameter data and acoustic emission data during the coal-rock loading process is proposed. Based on the proposed spatiotemporal feature extraction method, the full-time and spatiotemporal fracture parameters can be inverted through the full-time and spatiotemporal acoustic emission data obtained in the laboratory during the coal-rock loading process.

[0130] The model inversion mechanism in the coal-rock mass fracture field inversion system extracts features from the acoustic emission scalar parameter data and the full-time and space fracture parameter inversion results through the interpreter to obtain the influencing factors in the mechanism model. The interpreter can obtain the real-time quantitative inversion mechanism model of the fracture parameters of the coal-rock mass during the loading process through the acoustic emission data and the prediction results of the model, revealing the black box process of fracture parameter inversion of the coal-rock mass during the loading process.

[0131] In one example of the present invention, Figure 3As shown, the full-time and space crack parameter quantitative inversion module includes: a data encoding layer, a first Normalization function layer, a first linear feature extraction layer A, a convolution layer, a first activation function layer, a selective state space architecture, a linear feature extraction layer B, a second Normalization function layer and a fully connected layer connected in series; wherein the output end of the data encoding layer and the output end of the linear feature extraction layer B are connected by residual connection to splice the encoded data to the feature AE obtained by the feature extraction layer B b In the example, a residual connection consisting of a second linear feature extraction layer A and a second activation function layer is connected in series between the output end of the first Normalization function layer and the output end of the selective state space architecture; for example, the first activation function layer and the second activation function layer are both SiLU activation functions;

[0132] The data encoding layer is configured to embed the acoustic emission scalar parameter into the waveform data to obtain the parameter-enhanced acoustic emission data, and extract the parameter-enhanced acoustic emission feature AEf se ;

[0133] The first Normalization function layer is configured to enhance the acoustic emission feature AEf se Regularize to get feature AE n ;

[0134] The residual connection configuration composed of the second linear feature extraction layer A and the second activation function is used to extract the feature AE a+s Feature AE concatenated to the output of the selective state-space architecture sssm Get AE SSSM ;

[0135] The convolution layer and the first activation function layer are configured to extract the feature AE output by the first linear feature extraction layer A a Perform feature extraction to obtain higher-dimensional feature AE conv+s ;

[0136] The selective state space architecture is configured to perform gated calculations on high-dimensional features and concatenate the features extracted by the residual connection formed by the second linear feature extraction layer A and the second activation function layer to obtain the feature AE SSSM ;

[0137] The linear feature extraction layer B is configured to extract features from the concatenated features, and concatenate the extracted features with the features of the data encoded by the residual connection layer to obtain the feature AEf b+se ;

[0138] The second Normalization function layer is configured to regularize the concatenated features to obtain the feature AEf n ;

[0139] The fully connected layer is configured to calculate the extracted features to obtain the crack parameter F N .

[0140] That is to say, the neural network based on the selective state space architecture performs feature extraction on the full-space crack parameter data at the stage moment of the training set and the verification set in the crack parameter data set, the waveform data at the corresponding moment, and the scalar parameter data at the corresponding moment, including the following steps:

[0141] The acoustic emission scalar parameters are embedded into the waveform data through the data coding layer to obtain the parameter-enhanced acoustic emission data, and the parameter-enhanced acoustic emission features AEf are extracted. se ;

[0142] The encoded data is concatenated to the feature extraction layer B through the residual connection layer to obtain the feature AE b middle;

[0143] The acoustic emission characteristics AEf are enhanced by the first Normalization function. se Regularize to get feature AE n ;

[0144] Regularized feature AE n The feature extraction is performed through the residual connection composed of the second linear feature extraction layer A and the second activation function layer to obtain AE a+s Feature AE concatenated to the output of the selective state-space architecture sssm Get AE SSSM ;

[0145] The feature AE output by the linear feature extraction layer A a Feature extraction is performed through the convolution layer and the first activation function layer to obtain a higher-dimensional feature AE conv+s ;

[0146] The high-dimensional features are gated through the selective state space architecture and concatenated with the features extracted by the convolution layer and the first activation function layer to obtain the feature AE. SSSM ;

[0147] The linear feature extraction layer B extracts the concatenated features and concatenates the extracted features with the features of the encoded data by the residual connection layer to obtain the feature AEf b+se ;

[0148] The concatenated features are regularized by the second Normalization function to obtain the feature AEf n ;

[0149] The features extracted by the fully connected layer are calculated to obtain the crack parameter F N .

[0150] The exemplary implementation scheme of the coal rock fracture field inversion method and system with an interpretable selected state space architecture proposed in the present invention is described in detail above with reference to the preferred embodiments. However, those skilled in the art will appreciate that, without departing from the concept of the present invention, various modifications and variations can be made to the above-mentioned specific embodiments, and various technical features and structures proposed in the present invention can be combined in various ways without exceeding the protection scope of the present invention, which is determined by the attached claims.

Claims

1. A coal-rock mass fracture field inversion method capable of interpreting a selected state space framework, characterized in that: The steps include: S10: arranging CT detection sensors and acoustic emission monitoring sensors on the coal-rock mass, and placing the coal-rock mass in a testing machine for a load fracture test, and collecting CT detection data at staged moments and full-time and full-space acoustic emission monitoring data during the process of the coal-rock mass being loaded and fractured by the CT detection sensors and the acoustic emission monitoring sensors; wherein the CT detection data is full-space fracture parameter data at staged moments, and the acoustic emission monitoring data includes waveform data and scalar parameter data; S20: Corresponding the CT detection data at the stage in the process of coal rock mass load fracture and the full-time and space acoustic emission monitoring data according to the time, obtaining a fracture parameter data set, and dividing the fracture parameter data set into a training set, a verification set and a test set; S30: A neural network based on a selective state space architecture is used to extract features of the full-space fracture parameter data, waveform data at the corresponding time, and scalar parameter data at the corresponding time of the training set and the verification set in the fracture parameter data set to obtain a real-time quantitative inversion model of fracture parameters; the real-time quantitative inversion model of fracture parameters obtains the fracture parameters in the full-time and space fracture field during the loading process by inverting the full-time and space acoustic emission monitoring data, that is, the evolution process of the full-time and space fracture field parameters of the coal rock mass under loading is obtained; S40: The full-time and space-time acoustic emission scalar parameter data of the test set in the fracture parameter data set and the full-time and space-time fracture parameter inversion results obtained by the fracture parameter real-time quantitative inversion model are input into the SHAP interpreter to obtain the influencing factors between the acoustic emission scalar parameters and the fracture parameter inversion results, and then obtain the fracture parameter real-time quantitative inversion mechanism model based on the acoustic emission scalar parameters.

2. The coal-rock mass fracture field inversion method with interpretable selective state space framework according to claim 1 is characterized in that: In step S30, the neural network based on the selective state space architecture performs feature extraction on the stage-by-stage full-space crack parameter data, the corresponding moment waveform data and the corresponding moment scalar parameter data of the training set and the validation set in the crack parameter data set, including the following steps: S31: The acoustic emission scalar parameters are embedded into the waveform data through the data coding layer to obtain the parameter-enhanced acoustic emission data, and the parameter-enhanced acoustic emission features AEf are extracted. se ; S32: The encoded data is concatenated to the feature extraction layer B through residual connection to obtain the feature AE b middle; S33: Enhance the acoustic emission characteristics AEf through the first Normalization function se Regularize to get feature AE n ; S34: Regularized feature AE n The feature extraction is performed through the residual connection composed of the second linear feature extraction layer A and the second activation function layer to obtain AE a+s Feature AE concatenated to the output of the selective state-space architecture sssm Get AE SSSM ; S35: Feature AE output by the first linear feature extraction layer A a Feature extraction is performed through the convolution layer and the first activation function layer to obtain a higher-dimensional feature AE conv+s ; S36: The high-dimensional features are gated by the selective state space architecture, and the features extracted by the residual connection formed by the second linear feature extraction layer A and the second activation function layer are concatenated to obtain the feature AE SSSM ; S37: The linear feature extraction layer B extracts the concatenated features and concatenates the extracted features with the features of the encoded data by the residual connection layer to obtain the feature AEf b+se ; S38: Regularize the concatenated features through the second Normalization function to obtain feature AEf n ; S39: The features extracted by the fully connected layer are calculated to obtain the crack parameter F N .

3. The coal-rock mass fracture field inversion method capable of interpreting a selected state space framework according to claim 2 is characterized in that: In step S36, the selective state space architecture is specifically implemented as follows: First, a state space framework is constructed by continuous state space equations, wherein the expression of the continuous state space equations is as follows: h t =Ah t-1 +Bx t the t =Ch t Among them, h t-1 Represents the system status; h t represents the updated system state; matrix A represents the state transition matrix; B represents the input gating matrix; C represents the output mapping matrix; x t Data representing the current state of the input architecture; Secondly, the state space architecture is adapted to discrete data by discrete state space equations; the equation expression of the state transfer matrix A is as follows: Where n is the row index associated with the orthogonal basis function in the polynomial space, and k is the column index associated with the orthogonal basis function in the polynomial space.

4. The coal-rock mass fracture field inversion method capable of interpreting a selected state space framework according to claim 3 is characterized in that: It also includes: randomly initializing the state transfer matrix A and using the low-rank matrix after matrix decomposition. The low-rank matrix equation formula is as follows: Where P and Q are both vectors of length N; V represents the eigenvector matrix; V * represents the transposed matrix; Λ represents the characteristic matrix; * represents the projection operation.

5. The coal-rock mass fracture field inversion method capable of interpreting a selected state space framework according to claim 3 is characterized in that: The expression of the discrete state space equation is as follows: In the formula, h k and k They represent the memory control unit and output control unit of the current state in the state space architecture; h k-1 represents the memory control unit of the previous state in the state space architecture; h0 represents the initialization state of the memory control unit; x k Represents the data of the current state input architecture; x0 represents the data of the initial state input architecture; They are all matrices after discrete adaptation of A, B, and C.

6. The coal-rock mass fracture field inversion method capable of interpreting a selected state space framework according to claim 5 is characterized in that: matrix It is derived from the following formula: Where Δ represents the discretized nonlinear function, which discretizes the continuous state; Ι represents the unit matrix.

7. The coal-rock mass fracture field inversion method capable of interpreting a selected state space framework according to claim 1, characterized in that: In step S40, the expression formula of the SHAP interpreter is as follows: in, represents the impact factor; M represents the number of input acoustic emission scalar parameters; represents the set of all permutations and combinations of M features, represents the set of all feature combinations before feature k in the permutation R, f x (S) = E(f(X)|X s =x S ) represents the conditional expectation when the feature subset S is known; i represents the influencing factor number.

8. The coal-rock mass fracture field inversion method capable of interpreting a selected state space framework according to claim 7 is characterized in that: The influencing factors, acoustic emission scalar parameters and crack parameters obtained by the SHAP interpreter constitute the mechanism model for real-time quantitative inversion of crack parameters, and its expression equation is as follows: Among them, MM represents the mechanism model for real-time quantitative inversion of fracture parameters; F N Represents crack parameters; AE si Represents the acoustic emission scalar parameter.

9. A coal-rock mass fracture field inversion system capable of interpreting a selected state space architecture, characterized in that: include: The information acquisition module is configured to arrange CT detection sensors and acoustic emission monitoring sensors on the coal rock mass, and place the coal rock mass in a testing machine for a load fracture test, and acquire CT detection data at staged moments and full-time and space acoustic emission monitoring data during the process of coal rock mass load fracture through the CT detection sensors and acoustic emission monitoring sensors; wherein the CT detection data is full-space fracture parameter data at staged moments, and the acoustic emission monitoring data includes waveform data and scalar parameter data; A data division module is configured to correspond the stage-by-stage CT detection data and the full-time and space acoustic emission monitoring data in the process of coal rock mass load fracture according to time, obtain a fracture parameter data set, and divide the fracture parameter data set into a training set, a verification set and a test set; The full-time and space-time fracture parameter quantitative inversion module is configured to extract features of the full-time and space-time fracture parameter data, the waveform data at the corresponding time, and the scalar parameter data at the corresponding time of the training set and the verification set in the fracture parameter data set based on the neural network of the selective state space architecture, so as to obtain a real-time and quantitative inversion model of fracture parameters; the real-time and quantitative inversion model of fracture parameters obtains the fracture parameters of the full-time and space-time fracture field during the loading process by inverting the full-time and space-time acoustic emission monitoring data, that is, obtains the full-time and space-time fracture field parameter evolution process of the coal rock mass under loading; The inversion mechanism model module is configured to input the full-time and space-time acoustic emission scalar parameter data of the test set in the fracture parameter data set and the full-time and space-time fracture parameter inversion results obtained by the real-time quantitative inversion model of the fracture parameters into the SHAP interpreter, obtain the influencing factors between the acoustic emission scalar parameters and the fracture parameter inversion results, and then obtain the real-time quantitative inversion mechanism model of the fracture parameters based on the acoustic emission scalar parameters.

10. The coal-rock mass fracture field inversion system capable of interpreting a selected state space framework according to claim 9, characterized in that: The full-time and space crack parameter quantitative inversion module includes: a data encoding layer, a first Normalization function layer, a first linear feature extraction layer A, a convolution layer, a first activation function layer, a selective state space architecture, a linear feature extraction layer B, a second Normalization function layer and a fully connected layer connected in series in sequence; wherein the output end of the data encoding layer and the output end of the linear feature extraction layer B are connected by residual connection to splice the encoded data to the feature AE obtained by the feature extraction layer B b In the example, a residual connection consisting of a second linear feature extraction layer A and a second activation function layer is connected in series between the output end of the first Normalization function layer and the output end of the selective state space architecture; The data encoding layer is configured to embed the acoustic emission scalar parameter into the waveform data to obtain the parameter-enhanced acoustic emission data, and extract the parameter-enhanced acoustic emission feature AEf se ; The first Normalization function layer is configured to enhance the acoustic emission feature AEf se Regularize to get feature AE n ; The residual connection configuration composed of the second linear feature extraction layer A and the second activation function is used to extract the feature AE a+s Feature AE concatenated to the output of the selective state-space architecture sssm Get AE SSSM ; The convolution layer and the first activation function layer are configured to extract the feature AE output by the first linear feature extraction layer A a Perform feature extraction to obtain higher-dimensional feature AE conv+s ; The selective state space architecture is configured to perform gated calculations on high-dimensional features and concatenate the features extracted by the residual connection formed by the second linear feature extraction layer A and the second activation function layer to obtain the feature AE SSSM ; The linear feature extraction layer B is configured to extract features from the concatenated features, and concatenate the extracted features with the features of the data encoded by the residual connection layer to obtain the feature AEf b+se ; The second Normalization function layer is configured to regularize the concatenated features to obtain the feature AEf n ; The fully connected layer is configured to calculate the extracted features to obtain the crack parameter F N .

Images