Coal seam three-dimensional stress prediction and visualization method, system, device and medium

Abstract

The application relates to the technical field of coal seam stress prediction, in particular to a coal seam three-dimensional stress prediction and visualization method, system, device and medium, which comprises the following steps: setting an artificial seismic source and a detection point, exciting a seismic source signal and collecting a seismic wave response signal and a stress value; performing wave field forward modeling through a time fractional order viscoelastic wave equation to obtain a forward simulation wave field, performing full waveform inversion to output three-dimensional wave velocity data; performing reverse time migration imaging on the seismic wave response signal to obtain a three-dimensional coal seam structure mask graph; combining the three-dimensional wave velocity data and the three-dimensional coal seam structure mask graph into an input tensor after slicing along the same spatial dimension, inputting the input tensor into a convolutional neural network, and outputting a two-dimensional stress prediction slice; splicing the two-dimensional stress prediction slices to construct a three-dimensional stress body, generating three-dimensional stress point cloud data, and further generating a three-dimensional stress grid model and a structure-stress integrated visual three-dimensional cloud graph. The application can realize high-precision three-dimensional stress prediction and real-time early warning of the coal seam.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of coal seam stress prediction, and in particular to a coal seam three-dimensional stress prediction and visualization method, system, device and medium. BACKGROUND

[0002] With the continuous increase of coal mining depth and intensity, the geological structure in the mine is increasingly complex, and abnormal structures such as faults, fracture zones, and weak interlayers are ubiquitous. These geological anomalies are prone to induce dynamic disasters such as rock burst, coal and gas outburst, and roof accidents, which seriously threaten the safety of miners and the stability of coal production. Therefore, developing high-precision and intelligent coal seam geological detection technology to realize real-time identification and early warning of disaster sources has become a core requirement for ensuring coal mine safety.

[0003] The existing technology mainly uses seismic wave reflection / transmission method for geological imaging, combines anchor stress monitoring point data to back-propagate regional stress field, and uses Kriging and other spatial interpolation methods to realize the prediction of coal seam stress distribution. For example, seismic wave inversion based on elastic wave equation is used to obtain wave velocity data, and stress prediction map is generated by artificially calibrating the linear or power function relationship between wave velocity and stress. At the same time, some existing technologies also use reverse time migration imaging technology to identify geological boundaries, assist in building a static risk area display system to guide mining operations.

[0004] However, the existing coal seam stress distribution prediction technology has the following significant shortcomings: first, the existing wave velocity and stress mapping method relies on a simple empirical model and cannot fully capture the nonlinear coupling characteristics of coal and rock media, resulting in low stress prediction accuracy; second, the existing seismic wave simulation method is based on the assumption of non-damping elasticity, ignoring the coal and rock viscoelastic dissipation and dispersion effect, causing high-frequency attenuation and phase delay not to be modeled, and wave velocity inversion distortion is serious; third, the stress prediction process does not integrate geological boundary information, resulting in the phenomenon of "misinsertion across faults", which destroys the continuity of the stress field in complex structure areas and affects the detection reliability. SUMMARY

[0005] In view of the technical problems that the existing coal seam stress distribution prediction technology relies on a simple empirical model of wave velocity-stress, ignores coal and rock viscoelastic dissipation effect, and does not integrate geological boundary information, resulting in serious stress prediction distortion and cross-fault misinsertion, the present application provides a coal seam three-dimensional stress prediction and visualization method, system, device and medium, which realizes high-precision three-dimensional stress prediction and real-time early warning through viscoelastic wave field forward modeling, deep learning mapping constrained by structure mask, and three-dimensional visualization full-process closed loop, and solves the problem of detection distortion.

[0006] In a first aspect, the present application provides a coal seam three-dimensional stress prediction and visualization method, comprising the following steps:

[0007] S1. setting an artificial seismic source and a plurality of detection points in a to-be-detected region of a coal seam, exciting a seismic source signal at the artificial seismic source, and synchronously collecting seismic wave response signals and stress values of each detection point;

[0008] S2. based on the seismic source signal, performing wave field forward modeling by a time fractional order viscoelastic wave equation to obtain a forward simulated wave field wherein a short memory fast algorithm is used to solve a fractional derivative term of the time fractional order viscoelastic wave equation;

[0009] based on the forward simulated wave field and the seismic wave response signal, performing full waveform inversion to output three-dimensional wave velocity data of the to-be-detected region ;

[0010] wherein, is a spatial position vector;

[0011] is a time variable;

[0012] S3. based on the three-dimensional wave velocity data, performing reverse time migration imaging on the seismic wave response signal to obtain a three-dimensional coal seam structure mask map, including:

[0013] time-reversing the seismic wave response signal of each detection point to generate a virtual seismic source signal;

[0014] taking the virtual seismic source signal as an excitation source, performing wave field backward propagation based on the three-dimensional wave velocity data to generate a backward propagation wave field;

[0015] performing time-space domain cross-correlation calculation on the forward simulated wave field and the backward propagation wave field to obtain a three-dimensional imaging value;

[0016] performing gradient enhancement and threshold segmentation on the three-dimensional imaging value to generate a three-dimensional coal seam structure mask map identifying a geological boundary;

[0017] S4. combining the three-dimensional wave velocity data and the three-dimensional coal seam structure mask map after being sliced along the same spatial dimension into an input tensor, inputting the input tensor into a pre-trained convolutional neural network, and outputting a two-dimensional stress prediction slice;

[0018] It should be noted that "prediction" here refers to inferring the stress spatial distribution of an unmeasured region based on wave velocity data and a structure mask through a convolutional neural network (CNN), which is data reconstruction in the spatial dimension, rather than future state prediction in the time dimension;

[0019] S5. splicing all two-dimensional stress prediction slices to construct a three-dimensional stress body, and generating three-dimensional stress point cloud data based on the three-dimensional stress body;

[0020] converting the three-dimensional stress point cloud data into a three-dimensional stress grid model through a three-dimensional reconstruction algorithm;

[0021] Critical areas were delineated in a three-dimensional stress mesh model using mechanical analysis methods.

[0022] By combining the 3D stress mesh model and the 3D coal seam structure mask, stress value-color mapping rendering is performed to generate a visualized 3D cloud map integrating structure and stress, and dangerous areas are marked in the visualized 3D cloud map.

[0023] It should be further explained that in step S1, a channel wave detector and a stress monitoring anchor are configured at each detection point. The channel wave detector is used to collect seismic wave response signals, and the stress monitoring anchor is used to collect stress values.

[0024] It should be further noted that in step S2, the time-fractional viscoelastic wave equation is:

[0025]

[0026] in, Density of the coal seam medium;

[0027] It is a displacement vector field;

[0028] For the displacement vector field in directional components, ;

[0029] Let be the stress tensor components, where The normal to the plane where stress acts. Represents the direction of stress application. ;

[0030] for Spatial coordinate components of direction;

[0031] The focal term represents The volume force in the direction satisfies:

[0032]

[0033] Coordinates of the artificial earthquake source;

[0034] It is the spatial Dirac function;

[0035] The time function of the source signal;

[0036] The constant is the viscoelastic constant, calibrated through laboratory dynamic mechanical testing.

[0037] For Kronek functions, hour =1, otherwise =0;

[0038] For fractional order, ∈[0.3,0.8];

[0039] For displacement divergence;

[0040] For strain tensor, ;

[0041] The formula for solving fractional derivative terms using the short memory fast algorithm is:

[0042]

[0043] in, This is the current time step;

[0044] The step size is one time step;

[0045] L is the finite length of the memory window;

[0046] fractional order The determined weighting coefficients;

[0047] The forward wave modeling simulation uses the finite difference method to discretize the wave equations and calculates the forward simulated wave field over the entire spatial domain. .

[0048] It should be further noted that step S3 includes:

[0049] S301. Based on seismic wave response signals The virtual seismic source signal is generated by the following expression:

[0050]

[0051] in, For the first Each testing point at time The seismic wave response signal;

[0052] This represents the total recording duration of the seismic wave response signal;

[0053] For the first Each testing point at time The virtual seismic source model of the monitoring point;

[0054] Will As a virtual seismic source applied to the corresponding detection point, the wavefield is backpropagated using the same time-fractional viscoelastic wave equation based on the three-dimensional wave velocity data to obtain the reverse wavefield. ;

[0055] S302. Calculate three-dimensional imaging values :

[0056] ;

[0057] S303. After performing gradient enhancement and high-pass filtering on the 3D imaging values, set a threshold. According to threshold The expression for generating a structure mask image is:

[0058] .

[0059] It should be further noted that in step S3, the geological boundary includes geological faults and interlayer boundaries.

[0060] It should be further noted that step S4 includes:

[0061] S401. Transfer three-dimensional wave velocity data With 3D coal seam structure mask image Slicing along the Y-axis with a fixed step size yields a sequence of wave velocity data slices. Coal seam structure mask image slice sequence ;

[0062] in, The total number of slices, Indicates the first Zhang slices;

[0063] S402. Combine the wave velocity data slices with the corresponding coal seam structure mask image slices into an input tensor:

[0064] ;

[0065] S403. Input the input tensor into a pre-trained convolutional neural network, and output a two-dimensional stress prediction slice. .

[0066] It should be further noted that the convolutional neural network uses a U-Net encoder-decoder structure, including an encoder and a decoder, with skip connections between the encoder and decoder.

[0067] The encoder is composed of a plurality of sequentially connected encoding blocks, each encoding block containing two 3x3 convolutional layers, each convolutional layer followed by a batch normalization layer and a ReLU activation function, and a 2x2 max-pooling layer with a step size of 2 connected at the end of each encoding block to perform spatial down-sampling;

[0068] The decoder is composed of a plurality of sequentially connected decoding blocks, each decoding block containing a 2x2 transposed convolutional layer with a step size of 2 for performing spatial up-sampling, followed by a feature concatenation layer that concatenates the feature map output by the transposed convolutional layer with the encoder feature map at the corresponding level in the channel dimension, and two 3x3 convolutional layers connected after the feature concatenation layer, each convolutional layer followed by a batch normalization layer and a ReLU activation function;

[0069] The skip connection connects the encoder feature map output by each encoding block in the encoder to the feature concatenation layer at the corresponding level in the decoder.

[0070] Further, in the skip connection process, a structural boundary attention mechanism is used. After obtaining the encoder feature map, the mask feature map obtained by down-sampling the structural mask map is fused with the encoder feature map to obtain a fused feature map, and then the fused feature map is input into the feature concatenation layer. The specific fusion method of the structural boundary attention mechanism is:

[0071] For each encoder depth level , the following calculation is performed:

[0072]

[0073] represents the feature map output by the layer encoder;

[0074] is the mask feature map obtained by down-sampling the structural mask map times;

[0075] is a trainable guide factor, with a value range of [0.5, 2.0];

[0076] represents an element-wise multiplication operation;

[0077] represents the fused feature map.

[0078] Further, the training samples used by the convolutional neural network are derived from three-dimensional wave velocity data obtained by inverting seismic wave response signals using a time fractional order viscoelastic wave equation, and a three-dimensional coal seam structure mask map extracted based on reverse time migration imaging technology.

[0079] The convolutional neural network adopts the measured stress interpolation map as a label for semi-supervised training, and is jointly trained on multiple geological samples, wherein the measured stress interpolation map is a coarse resolution stress distribution map generated by interpolation method based on the measured stress value of each detection point;

[0080] The loss function of the convolutional neural network is:

[0081]

[0082] wherein, is the measured stress interpolation map composed of stress values of all detection points with the same height on the Y axis;

[0083] , , is a weight coefficient,

[0084] MAE is the mean absolute error;

[0085] SSIM is the structural similarity index;

[0086] Edgeloss is the boundary alignment loss, that is, the difference between the predicted stress gradient and the measured stress gradient at the geological boundary identified by the three-dimensional coal seam structure mask map.

[0087] Further, in step S5, the step of generating three-dimensional stress point cloud data includes:

[0088] S501. Splice all two-dimensional prediction slices back to three-dimensional coordinate axes in the original spatial order to form a complete three-dimensional stress body:

[0089] ;

[0090] S502. Voxel sampling is performed on the three-dimensional stress body to extract voxel coordinates and stress values, and three-dimensional stress point cloud data is generated:

[0091]

[0092] wherein, is the spatial coordinate of the i-th sampling point in the three-dimensional stress body;

[0093] is the predicted stress value of the i-th sampling point in the three-dimensional stress body; is the stress display lower limit threshold.

[0094]

[0095] ​​​It should be further explained that in step S5, the mechanical analysis method includes finite element method (FEM), discrete element method (DEM) and boundary element method (BEM).

[0096] It should be further explained that in step S5, in the process of stress value-color mapping rendering, the corresponding relationship between the stress value and the color is as follows:

[0097] .

[0098] In a second aspect, the application provides a coal seam three-dimensional stress prediction and visualization system for implementing the coal seam three-dimensional stress prediction and visualization method described above, comprising:

[0099] A data acquisition module is configured to set an artificial seismic source and a plurality of detection points in a to-be-detected region of a coal seam, excite a seismic source signal at the artificial seismic source, and synchronously acquire seismic wave response signals and stress values at the detection points;

[0100] A wave field forward simulation module is configured to perform wave field forward simulation based on the seismic source signal through a time fractional order viscoelastic wave equation to obtain a forward simulation wave field;

[0101] A full waveform inversion module is configured to perform full waveform inversion based on the forward simulation wave field and the seismic wave response signals to output three-dimensional wave velocity data of the to-be-detected region;

[0102] A reverse-time migration imaging module is configured to perform reverse-time migration imaging on the seismic wave response signals based on the three-dimensional wave velocity data to obtain a three-dimensional coal seam structure mask image;

[0103] A two-dimensional stress prediction module is configured to combine the three-dimensional wave velocity data and the three-dimensional coal seam structure mask image after being sliced along the same spatial dimension to form an input tensor, input the pre-trained convolutional neural network, and output a two-dimensional stress prediction slice;

[0104] A three-dimensional visualization module is configured to:

[0105] splice and construct a three-dimensional stress body from all the two-dimensional stress prediction slices, and generate three-dimensional stress point cloud data based on the three-dimensional stress body;

[0106] convert the three-dimensional stress point cloud data into a three-dimensional stress grid model through a three-dimensional reconstruction algorithm;

[0107] delimit a dangerous region in the three-dimensional stress grid model using a mechanical analysis method;

[0108] perform stress value-color mapping rendering after combining the three-dimensional stress grid model and the three-dimensional coal seam structure mask image to generate a structure-stress integrated visual three-dimensional cloud image, and mark the dangerous region in the visual three-dimensional cloud image.

[0109] In a third aspect, the present application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor is configured to implement the steps of the coal seam three-dimensional stress prediction and visualization method when executing the computer program.

[0110] In a fourth aspect, the present application provides a storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the coal seam three-dimensional stress prediction and visualization method.

[0111] From the above technical solutions, the present application has the following advantages:

[0112] 1. The present application sets artificial seismic sources and multiple detection points in the to-be-detected area of the coal seam, synchronously collects seismic wave response signals and stress values, and performs wave field forward modeling based on a time fractional order viscoelastic wave equation, and solves the fractional derivative term using a short memory fast algorithm, thereby solving the wave speed distortion problem caused by the neglect of coal rock viscoelastic characteristics in the prior art, achieving more accurate three-dimensional wave speed data inversion, and effectively simulating energy dissipation and phase delay of seismic waves in the coal seam.

[0113] 2. The present application performs reverse time migration imaging based on three-dimensional wave speed data, including time reversal to generate a virtual source signal, wave field backward propagation, and cross-correlation calculation, and generates a three-dimensional coal seam structure mask graph through gradient enhancement and threshold segmentation, thereby solving the cross-fault misinsertion problem caused by the lack of structure information in the prior stress prediction technology, achieving fine identification of geological faults, and ensuring the spatial continuity of stress field modeling.

[0114] 3. The present application combines three-dimensional wave speed data and three-dimensional coal seam structure mask graph slicing into an input tensor, inputs a convolutional neural network, and supervises training with stress values, thereby solving the low precision problem of the wave speed-stress simple mapping model in the prior art, achieving end-to-end nonlinear stress prediction, and improving the accuracy of two-dimensional stress slice output.

[0115] 4. The present application constructs a three-dimensional stress body by splicing all two-dimensional stress prediction slices, generates three-dimensional stress point cloud data, converts the three-dimensional stress point cloud data into a three-dimensional stress grid model, and performs stress value-color mapping rendering in combination with the three-dimensional coal seam structure mask graph, generates a structure-stress integrated visual three-dimensional cloud map, marks a dangerous area in the visual three-dimensional cloud map, solves the poor interactivity and update delay problem of the existing visualization method, and achieves real-time three-dimensional cloud map display of structure-stress integration, which can support second-level early warning of high-risk areas in coal mines.

[0116] 5. The present application solves the problems of error accumulation and poor real-time performance caused by step-by-step independent processing in the prior art through integrated processing of the whole process from viscoelastic wave field forward modeling, full waveform inversion, reverse time migration imaging to deep learning stress prediction, and realizes high-precision and high-efficiency closed loop of disaster source detection and early warning, which can dynamically update the three-dimensional stress cloud map of the coal mining face and improve the disaster response speed. BRIEF DESCRIPTION OF DRAWINGS

[0117] In order to more clearly illustrate the technical solutions of the present application, the following will briefly introduce the drawings needed to be used in the description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0118] Figure 1 is a flow chart of a coal seam three-dimensional stress prediction and visualization method in an embodiment of the present application.

[0119] Figure 2 is a schematic block diagram of a coal seam three-dimensional stress prediction and visualization system in an embodiment of the present application.

[0120] Figure 3 is a schematic diagram of the hardware structure of an electronic device in an embodiment of the present application. DETAILED DESCRIPTION

[0121] In order to make the application purposes, features and advantages of the present application more obvious and easy to understand, the following will use specific embodiments and drawings to clearly and completely describe the technical solutions protected by the present application. Obviously, the following described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present patent, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present patent.

[0122] The coal seam three-dimensional stress prediction and visualization method of the present application will be described in detail below. In order to illustrate but not to limit, specific details such as specific system structures and technologies are proposed to thoroughly understand the embodiments of the present application. However, those skilled in the art should understand that the present application can also be implemented in other embodiments without these specific details.

[0123] In the coal seam three-dimensional stress prediction and visualization method of the present application, the term "comprising" indicates the existence of the described features, whole, steps, operations, elements and / or components, but does not exclude the existence or addition of one or more other features, whole, steps, operations, elements, components and / or sets thereof. The terms "comprising", "including", "having" and their variants mean "including but not limited to", unless otherwise specifically emphasized.

[0124] In order to clearly describe the technical solutions of the present application, the terms "first", "second", etc. are used to distinguish the same or similar items or functions and roles. Those skilled in the art can understand that the terms "first", "second", etc. do not limit the quantity and execution order, and the terms "first", "second", etc. also do not necessarily mean different.

[0125] The phrases "one embodiment" or "some embodiments" appearing in the present application mean that the specific features, structures or characteristics described in the embodiment are included in one or more embodiments of the present application. Therefore, the phrases "in one embodiment", "in some embodiments", "in other some embodiments", "in further some embodiments" appearing in the present application do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized.

[0126] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application.

[0127] The coal seam three-dimensional stress prediction and visualization method provided by the embodiments of the present application is executed by a computer device, and accordingly, the coal seam three-dimensional stress prediction and visualization system runs in the computer device.

[0128] Figure 1 is a flowchart of the coal seam three-dimensional stress prediction and visualization method of one embodiment of the present application. In the flowchart, Figure 1 The execution subject can be a coal seam three-dimensional stress prediction and visualization system. According to different needs, the order of steps in the flowchart can be changed, and some steps can be omitted.

[0129] As shown in Figure 1 The coal seam three-dimensional stress prediction and visualization method includes:

[0130] Step S1, setting an artificial seismic source and a plurality of detection points in a to-be-detected area of a coal seam, exciting a seismic source signal at the artificial seismic source, and synchronously collecting seismic wave response signals and stress values of each detection point.

[0131] By setting the artificial seismic source and the detection points, exciting the seismic source signal, and synchronously collecting the seismic wave response signals and the stress values of each detection point, the synchronous acquisition of the original seismic wave and stress data is realized, the time alignment and spatial coverage of the data can be ensured, the basic input for subsequent wave field forward simulation and full waveform inversion is provided, and the error accumulation caused by different synchronization of data acquisition is avoided

[0132] In some specific embodiments, a channel wave detector and a stress monitoring anchor are configured in each detection point, the channel wave detector is used to collect the seismic wave response signal, and the stress monitoring anchor is used to collect the stress value.

[0133] By configuring the channel wave detector in each detection point to collect the seismic wave response signal and using the stress monitoring anchor to collect the stress value, synchronous high-precision collection of the seismic wave signal and the stress value is realized, the consistency and reliability of the data can be ensured, the data deviation caused by the mismatch of equipment is avoided, and accurate input basis is provided for subsequent wave field forward simulation and full waveform inversion.

[0134] Step S2, based on the source signal, wave field forward simulation is performed through a time fractional order viscoelastic wave equation to obtain a forward simulation wave field Wherein, a short memory fast algorithm is used to solve the fractional derivative term of the time fractional order viscoelastic wave equation;

[0135] Full waveform inversion is performed based on the forward simulation wave field and the seismic wave response signal to output three-dimensional wave velocity data of the region to be detected ;

[0136] Wherein, is a spatial position vector;

[0137] is a time variable.

[0138] By using the time fractional order viscoelastic wave equation to perform wave field forward simulation, combining the forward simulation wave field with the seismic wave response signal to perform full waveform inversion and outputting three-dimensional wave velocity data, high-precision inversion of the wave velocity characteristics of the coal seam is realized, the medium viscoelastic behavior can be accurately captured, and reliable three-dimensional wave velocity data basis is provided for reverse time migration imaging.

[0139] In some specific embodiments, the time fractional order viscoelastic wave equation is:

[0140]

[0141] Wherein, is the medium density of the coal seam;

[0142] is a displacement vector field;

[0143] is the component of the displacement vector field in the direction of ;

[0144] is a stress tensor component, wherein represents the normal of the stress action plane, represents the stress action direction, ;​

[0145] For the spatial coordinate component of the direction;

[0146] For the source term, representing the volume force in the direction, satisfying:

[0147]

[0148] For the coordinates of the artificial seismic source;

[0149] For the spatial Dirac function;

[0150] For the time function of the source signal;

[0151] For the viscoelastic constant, calibrated by laboratory dynamic mechanical testing;

[0152] For the Kronicker function, When =1, otherwise =0;

[0153] For the fractional order, ∈[0.3,0.8];

[0154] For the divergence of displacement;

[0155] For the strain tensor, ;

[0156] The formula for solving the fractional derivative term by the short memory fast algorithm is:

[0157]

[0158] Where, is the current time step;

[0159] is the step length of 1 time step;

[0160] L is the finite memory window length;

[0161] is the fractional order determined by the weight coefficient;

[0162] The wave field forward modeling adopts finite difference method to discretize the wave equation, and calculates the full range of forward simulation wave field in spatial domain .

[0163] By employing the time-fractional viscoelastic wave equation and utilizing a short-memory fast algorithm to solve the fractional derivative terms (the specific formula is an approximate summation of weighting coefficients and a finite memory window length), combined with the finite difference method to discrete the wave equation, forward modeling of the wave field is performed, thus achieving accurate simulation of the viscoelastic properties of coal seam media. This method can efficiently calculate the forward simulated wave field across the entire spatial domain, improving the accuracy of three-dimensional wave velocity data from full waveform inversion.

[0164] Step S3: Perform reverse time migration imaging on the seismic wave response signal based on the three-dimensional wave velocity data to obtain a three-dimensional coal seam structure mask image, including:

[0165] Time-reversal is performed on the seismic wave response signals at each detection point to generate virtual source signals;

[0166] Using a virtual seismic source signal as the excitation source, the wave field is back propagated based on three-dimensional wave velocity data to generate a back propagation wave field.

[0167] Spatiotemporal cross-correlation calculations were performed on the forward simulated wavefield and the backward propagating wavefield to obtain the three-dimensional imaging values;

[0168] Gradient enhancement and threshold segmentation are performed on the 3D imaging values ​​to generate a 3D coal seam structure mask map that identifies geological boundaries.

[0169] By performing reverse time migration imaging on seismic wave response signals based on three-dimensional wave velocity data, clear extraction of geological boundaries was achieved, which can enhance the boundary recognition capability of coal seam structure mask images and provide structural constraint information for convolutional neural network input.

[0170] In some specific embodiments, step S3 includes:

[0171] S301. Based on seismic wave response signals The virtual seismic source signal is generated by the following expression:

[0172]

[0173] in, For the first Each testing point at time The seismic wave response signal;

[0174] This represents the total recording duration of the seismic wave response signal;

[0175] For the first Each testing point at time The virtual seismic source model of the monitoring point;

[0176] Will As the position of the virtual source applied to the corresponding detection point, the wave field is back propagated based on the three-dimensional wave velocity data using the same time fractional order viscoelastic wave equation to obtain the back wave field

[0177] S302. Calculate the three-dimensional imaging value

[0178]

[0179] S303. After gradient enhancement and high-pass filtering of the three-dimensional imaging value, set a threshold value , generate a structure mask map according to the threshold value , the expression is:

[0180]

[0181] By generating a virtual source signal based on the seismic wave response signal (the expression is a time reversal function), generating a back wave field based on three-dimensional wave velocity data using the virtual source signal as an excitation source, calculating the three-dimensional imaging value by calculating the space-time domain cross-correlation integral of the forward simulation wave field and the back propagation wave field, and setting a threshold value after gradient enhancement and high-pass filtering of the three-dimensional imaging value to generate a structure mask map, the clear identification of the coal seam geological boundary is realized, which can enhance the boundary recognition ability of reverse time migration imaging and reduce noise interference

[0182] In some specific embodiments, the geological boundary includes a geological fault and a interlayer boundary.

[0183] By defining the geological boundary as including the geological fault and the interlayer boundary, the targeted processing of the structure mask map in reverse time migration imaging is realized, which can focus on the detection of key geological structures and improve the practicality and geological interpretation accuracy of the coal seam structure mask map.

[0184] Step S4, combine the three-dimensional wave velocity data and the three-dimensional coal seam structure mask map after slicing along the same spatial dimension to form an input tensor, input the pre-trained convolutional neural network, and output a two-dimensional stress prediction slice.

[0185] By using the convolutional neural network to output a two-dimensional stress prediction slice, effective fusion of wave velocity and structure characteristics and deep learning modeling are realized, which can improve the local accuracy and generalization of stress prediction and provide high-quality slice data for three-dimensional stress body construction

[0186] In some specific embodiments, step S4 includes:

[0187] S401. Slice the three-dimensional wave velocity data and the three-dimensional coal seam structure mask map ​​​​slicing the wave velocity data along the Y axis with a fixed step size to obtain a wave velocity data slice sequence and a coal seam structure mask map slice sequence ;

[0188] wherein, is the total number of slices, represents the th slice;

[0189] S402. Combine the wave velocity data slice and the corresponding coal seam structure mask map slice into an input tensor:

[0190] ;

[0191] S403. Input the input tensor into the pre-trained convolutional neural network to output a two-dimensional stress prediction slice .

[0192] By slicing the three-dimensional wave velocity data and the three-dimensional coal seam structure mask map along the Y axis with a fixed step size to obtain a wave velocity data slice sequence and a coal seam structure mask map slice sequence, and combining the wave velocity data slice and the corresponding coal seam structure mask map slice into an input tensor inputting into a convolutional neural network to output a two-dimensional stress prediction slice, efficient preprocessing and feature fusion of data are realized, the input structure of the convolutional neural network can be optimized, and the local precision and calculation efficiency of stress prediction can be improved.

[0193] In some specific embodiments, the convolutional neural network is a U-Net encoding-decoding structure, including an encoder and a decoder, and the encoder and the decoder are connected by a skip connection, wherein:

[0194] The encoder is composed of a plurality of sequentially connected encoding blocks, each encoding block contains two 3x3 convolutional layers, each convolutional layer is followed by a batch normalization layer and a ReLU activation function, and a 2x2 max-pooling layer with a step size of 2 is connected at the end of each encoding block to perform spatial downsampling;

[0195] The decoder is composed of a plurality of sequentially connected decoding blocks, each decoding block contains a 2x2 transposed convolutional layer with a step size of 2 for performing spatial upsampling, and a feature concatenation layer is connected after the transposed convolutional layer, which concatenates the feature map output by the transposed convolutional layer with the encoder feature map at the corresponding level in the channel dimension, and two 3x3 convolutional layers are connected after the concatenation, each convolutional layer is followed by a batch normalization layer and a ReLU activation function;

[0196] The skip connection connects the encoder feature map output by each encoding block in the encoder to the feature concatenation layer at the corresponding level in the decoder.

[0197] By adopting the U-Net encoding-decoding structure, multi-scale extraction and fusion of feature maps are realized, which can enhance the modeling ability of the convolutional neural network for the relationship between the coal seam structure and stress, and improve the accuracy of the two-dimensional stress prediction slice.

[0198] In some specific embodiments, a structural boundary attention mechanism is adopted in the skip connection process. After obtaining the encoder feature map, the mask feature map obtained by down-sampling the structural mask map is fused with the encoder feature map to obtain a fused feature map, and then the fused feature map is input into a feature splicing layer. The specific fusion method of the structural boundary attention mechanism is as follows:

[0199] For each encoder depth level , the calculation is performed as follows:

[0200]

[0201] denotes the feature map output by the i-th encoder;

[0202] is the mask feature map obtained by down-sampling the structural mask map for times;

[0203] is a trainable guide factor, and the value range is [0.5, 2.0];

[0204] denotes an element-wise multiplication operation;

[0205] denotes the fused feature map.

[0206] By adopting the structural boundary attention mechanism in the skip connection process, dynamic weighted fusion of the encoder feature map and the structural mask information is realized, which can strengthen the feature expression of the geological boundary area and improve the stress prediction accuracy of the convolutional neural network at the boundary.

[0207] In some specific embodiments, the training samples used by the convolutional neural network are derived from three-dimensional wave velocity data obtained by inverting the seismic wave response signal through the time fractional order viscoelastic wave equation, and three-dimensional coal seam structure mask maps extracted based on the reverse time migration imaging technology;

[0208] The convolutional neural network adopts a measured stress interpolation map as a label for semi-supervised training, and is jointly trained on multiple geological samples. The measured stress interpolation map is a coarse resolution stress distribution map generated by an interpolation method based on the measured stress values of each detection point.

[0209] The loss function of the convolutional neural network is as follows:

[0210] ​​

[0211] wherein, is a weight coefficient, a measured stress interpolation map composed of stress values of all detection points with the same height on the Y axis;

[0212] , , is a weight coefficient,

[0213] MAE is the mean absolute error;

[0214] SSIM is the structural similarity index;

[0215] Edgeloss is the edge alignment loss, that is, the difference between the predicted stress gradient and the measured stress gradient at the geological boundary identified by the three-dimensional coal seam structure mask.

[0216] By setting the loss function, multi-objective optimization of convolutional neural network training is realized, which can balance the overall stress prediction accuracy, structural similarity and geological boundary alignment, and improve the robustness and generalization ability of the two-dimensional stress prediction slice

[0217] Step S5, all two-dimensional stress prediction slices are spliced to construct a three-dimensional stress body, and three-dimensional stress point cloud data is generated based on the three-dimensional stress body;

[0218] The three-dimensional stress point cloud data is converted into a three-dimensional stress grid model by a three-dimensional reconstruction algorithm;

[0219] A dangerous area is demarcated in the three-dimensional stress grid model using a mechanical analysis method;

[0220] After combining the three-dimensional stress grid model and the three-dimensional coal seam structure mask, stress value-color mapping rendering is performed to generate a structure-stress integrated visual three-dimensional cloud chart, and the dangerous area is marked in the visual three-dimensional cloud chart.

[0221] By splicing two-dimensional stress prediction slices to construct a three-dimensional stress body, generating three-dimensional stress point cloud data, converting it into a three-dimensional stress grid model and demarcating a dangerous area, and then combining it with a three-dimensional coal seam structure mask for stress value-color mapping rendering, a structure-stress integrated visual three-dimensional cloud chart is further generated, and the dangerous area is marked, realizing three-dimensional intuitive display of stress distribution, supporting visual analysis of the coupling relationship between geological structure and stress, and facilitating coal mine safety assessment and decision-making.

[0222] In some specific embodiments, the step of generating three-dimensional stress point cloud data includes:

[0223] S501. All two-dimensional prediction slices Splice back to three-dimensional coordinate axis in original spatial order, form a complete three-dimensional stress body:

[0224] ;

[0225] S502. Voxel sampling is performed on the three-dimensional stress body to extract voxel coordinates and stress values, and three-dimensional stress point cloud data is generated:

[0226]

[0227] wherein, is the spatial coordinate of the i-th sampling point in the three-dimensional stress body;

[0228] is the predicted stress value of the i-th sampling point in the three-dimensional stress body;

[0229] is the lower threshold value of stress display.

[0230] In some embodiments, the mechanical analysis method includes finite element method (FEM), discrete element method (DEM), and boundary element method (BEM).

[0231] In some embodiments, the execution steps of the finite element method include:

[0232] Model discretization: divide the three-dimensional stress body grid model into tetrahedral or hexahedral element grids;

[0233] Constitutive model loading: assign an elastoplastic constitutive relationship (such as Mohr-Coulomb criterion or Drucker-Prager model) to the coal rock mass;

[0234] Boundary condition setting: apply displacement constraints (such as bottom fixation, lateral pressure) according to the mining working conditions;

[0235] Stress solving: solve the balance equation to output the stress tensor of each element grid;

[0236] Dangerous determination: if the stress long-span burst of the element grid meets the uniaxial compressive strength calibrated in the laboratory, the element grid is marked as a dangerous area.

[0237] In some embodiments, the execution steps of the discrete element method include:

[0238] Particle model construction: discretize the coal rock mass into a set of rigid particles, and the stress is transmitted between the particles through contact force chains;

[0239] Joint network import: embed the geometric parameters of faults / fractures based on the three-dimensional stress body grid model;

[0240] ​​Dynamic analysis: simulate the particle motion under mining disturbance, calculate the contact force distribution;

[0241] Dangerous judgment: count the average contact force of the local area, if it exceeds the preset threshold, it is determined that the area is a dangerous area of instability.

[0242] In some embodiments, the execution steps of the boundary element method include:

[0243] Boundary discretization: only the surface of the three-dimensional stress body grid model is divided into elements;

[0244] Integral equation solution: calculate the boundary stress

[0245] Fault activation analysis: calculate the stress intensity factor K on the fault surface, if K>K1 (preset fracture toughness), mark as a potential water / gas outburst dangerous area.

[0246] In some embodiments, in the process of stress value-color mapping rendering, the corresponding relationship between stress value and color is:

[0247] .

[0248] By setting the corresponding relationship between stress value and color, the hierarchical color expression of the visualized three-dimensional cloud chart is realized, which can intuitively distinguish low, medium and high stress areas, and facilitate quick identification of potential dangerous geological structures.

[0249] In one embodiment, the steps of the coal seam three-dimensional stress prediction and visualization method include:

[0250] Step S1, setting an artificial seismic source and a plurality of detection points in the to-be-detected area of the coal seam, exciting a seismic source signal at the artificial seismic source, configuring a channel wave detector and a stress monitoring anchor rod in each detection point, using the channel wave detector to synchronously collect seismic wave response signals of each detection point, and using the stress monitoring anchor rod to synchronously collect stress values of each detection point.

[0251] Step S2, based on the seismic source signal, performing wave field forward modeling by time fractional order viscoelastic wave equation to obtain forward modeling wave field , wherein the fractional derivative term of the time fractional order viscoelastic wave equation is solved by using a short memory fast algorithm;

[0252] Based on the forward modeling wave field and the seismic wave response signal, full waveform inversion is performed to output three-dimensional wave velocity data of the to-be-detected area ;

[0253] , wherein, is a spatial position vector;

[0254] is a time variable;

[0255] The time fractional viscoelastic wave equation is:

[0256]

[0257] where, is the density of coalbed medium;

[0258] is the displacement vector field;

[0259] is the component of the displacement vector field in direction, ;

[0260] is the stress tensor component, where represents the normal of the stress action plane, represents the stress action direction, ;

[0261] is the spatial coordinate component in direction;

[0262] is the source term, representing the volume force in direction, satisfying:

[0263]

[0264] is the artificial seismic source coordinate;

[0265] is the spatial Dirac function;

[0266] is the time function of the source signal;

[0267] is the viscoelastic constant, calibrated by laboratory dynamic mechanical testing;

[0268] is the Kronicker function, when = 1, otherwise = 0;

[0269] is the fractional order, ∈ [0.3, 0.8];

[0270] is the divergence of displacement;

[0271] is the strain tensor, ;

[0272] The formula for solving the fractional derivative term by the short memory fast algorithm is:

[0273]

[0274] wherein, is the current time step;

[0275] is the step length of 1 time step;

[0276] L is the length of the limited memory window;

[0277] is the fractional order determined weight coefficient;

[0278] The wave field forward simulation adopts the finite difference method to discretize the wave equation, and the full-range forward simulation wave field in the spatial domain is calculated .

[0279] Step S3, based on the three-dimensional wave velocity data, the reverse time migration imaging of the seismic wave response signal is performed, and a three-dimensional coal seam structure mask map is obtained, including:

[0280] S301. Based on the seismic wave response signal generate a virtual source signal, the expression is:

[0281]

[0282] wherein, is the seismic wave response signal of the mth detection point at time t; is the total recording duration of the seismic wave response signal;

[0283] is the virtual source model of the mth detection point at time t;

[0284]

[0285] is applied to the corresponding detection point position as a virtual source, and the same time fractional order viscoelastic wave equation is used based on the three-dimensional wave velocity data to perform wave field backward propagation, and the backward wave field is obtained;

[0286] S302. Calculate the three-dimensional imaging value :

[0287] ; ​​​​​

[0288] S303. After gradient enhancement and high-pass filtering of the three-dimensional imaging values, a threshold is set , and a three-dimensional coal seam structure mask map identifying geological boundaries is generated according to the threshold , expressed as:

[0289] ;

[0290] The geological boundaries include geological faults and interlayer boundaries.

[0291] Step S4, after slicing the three-dimensional wave velocity data and the three-dimensional coal seam structure mask map along the same spatial dimension to combine into an input tensor, inputting the pre-trained convolutional neural network, outputting a two-dimensional stress prediction, specifically including:

[0292] S401. Slice the three-dimensional wave velocity data and the three-dimensional coal seam structure mask map along the Y axis according to a fixed step size to obtain a wave velocity data slice sequence and a coal seam structure mask map slice sequence ;

[0293] wherein, is the total number of slices, represents the th slice;

[0294] S402. Combine the wave velocity data slice and the corresponding coal seam structure mask slice into an input tensor:

[0295] ;

[0296] S403. Input the input tensor into the pre-trained convolutional neural network to output a two-dimensional stress prediction slice ;

[0297] wherein, the convolutional neural network is a U-Net encoding-decoding structure, including an encoder and a decoder, and the encoder and the decoder are connected by a skip connection, wherein:

[0298] The encoder is composed of multiple sequentially connected encoding blocks, each encoding block contains two 3x3 convolution layers, each convolution layer is followed by a batch normalization layer and a ReLU activation function, and a 2x2 maximum pooling layer with a step size of 2 is connected at the end of each encoding block to perform spatial down-sampling;

[0299] The parameter settings of the encoder are shown in Table 1:

[0300] Table 1 Encoder parameter setting table

[0301]

[0302] The decoder is composed of a plurality of sequentially connected decoding blocks, each decoding block containing a 2x2 transposed convolution layer with a step size of 2 for performing spatial upsampling, a feature concatenation layer connected after the transposed convolution layer, the feature concatenation layer concatenating the feature map output by the transposed convolution layer and the encoder feature map of the corresponding level in the channel dimension, and two 3x3 convolution layers connected after the concatenation, each convolution layer followed by a batch normalization layer and a ReLU activation function;

[0303] The parameter settings of the decoder are shown in Table 2:

[0304] Table 2 Decoder parameter setting table

[0305]

[0306] The skip connection connects the encoder feature map output by each encoding block in the encoder to the feature concatenation layer of the corresponding level in the decoder;

[0307] In the skip connection process, a structural boundary attention mechanism is used. After obtaining the encoder feature map, the mask feature map obtained by downsampling the structural mask map is fused with the encoder feature map to obtain a fused feature map, and then the fused feature map is input into the feature concatenation layer. The specific fusion method of the structural boundary attention mechanism is:

[0308] For each encoder depth level , the following calculation is performed:

[0309]

[0310] represents the feature map output by the layer encoder;

[0311] is the mask feature map obtained by downsampling the structural mask map times;

[0312] is a trainable guide factor, and the value range is [0.5, 2.0];

[0313] represents an element-wise multiplication operation;

[0314] represents the fused feature map;

[0315] The training samples used by the convolutional neural network are derived from three-dimensional wave velocity data obtained by inverting seismic wave response signals through a time fractional order viscoelastic wave equation and a three-dimensional coal seam structure mask map extracted based on reverse time migration imaging technology;

[0316] The convolutional neural network is semi-supervised trained by using the measured stress interpolation map as a label, and is jointly trained on multiple geological samples, wherein the measured stress interpolation map is a coarse resolution stress distribution map generated by interpolation method based on the measured stress value of each detection point;

[0317] The loss function of the convolutional neural network is:

[0318]

[0319] wherein, is the stress value of the detection point corresponding to the coordinate point, the measured stress interpolation map composed of stress values of all detection points with the same height on the Y axis;

[0320] , , is a weight coefficient,

[0321] MAE is the mean absolute error;

[0322] SSIM is the structural similarity index;

[0323] Edgeloss is the boundary alignment loss, that is, the difference between the predicted stress gradient and the measured stress gradient at the geological boundary identified by the three-dimensional coal seam structure mask.

[0324] Step S5, all two-dimensional stress prediction slices are spliced to construct a three-dimensional stress body, and three-dimensional stress point cloud data is generated based on the three-dimensional stress body;

[0325] The three-dimensional stress point cloud data is converted into a three-dimensional stress grid model by a three-dimensional reconstruction algorithm;

[0326] The finite element method is used to delimit a dangerous area in the three-dimensional stress grid model;

[0327] After combining the three-dimensional stress grid model and the three-dimensional coal seam structure mask, stress value-color mapping rendering is performed to generate a structure-stress integrated visual three-dimensional cloud image, and the dangerous area is marked in the visual three-dimensional cloud image, and the step of generating three-dimensional stress point cloud data comprises:

[0328] S501. All two-dimensional prediction slices are spliced back to the three-dimensional coordinate axis in the original spatial order to form a complete three-dimensional stress body:

[0329] ;

[0330] S502. Voxel sampling is performed on the three-dimensional stress body to extract voxel coordinates and stress values, and three-dimensional stress point cloud data is generated:

[0331]

[0332] wherein, is the spatial coordinate of the i-th sampling point in the three-dimensional stress body;

[0333] is the predicted stress value of the i-th sampling point in the three-dimensional stress body;

[0334] is the stress display lower threshold value;

[0335] The steps of using the finite element method to delineate the dangerous area in the three-dimensional stress grid model include:

[0336] Model discretization: divide the three-dimensional stress body grid model into tetrahedral or hexahedral element grid;

[0337] Constitutive model loading: give the coal and rock mass a elastoplastic constitutive relationship;

[0338] Boundary condition setting: apply displacement constraints according to the mining working condition;

[0339] Stress solution: solve the balance equation and output the stress tensor of each element grid;

[0340] Dangerous determination: if the stress long-span burst of the element grid meets the uniaxial compressive strength calibrated in the laboratory, mark the element grid as a dangerous area;

[0341] In the process of stress value-color mapping rendering, the corresponding relationship between stress value and color is:

[0342]

[0343] The optional color scale scheme includes Jet, Viridis, Thermal or Turbo;

[0344] The visualization technology selection includes:

[0345] Use OpenGL / WebGL / Potree / Three.js or VTK to realize 3D point cloud rendering;

[0346] The rendering parameters include point size, color transparency, and response speed optimization.

[0347] The following is an embodiment of the coal seam three-dimensional stress prediction and visualization system provided by the present application, which belongs to the same inventive concept as the coal seam three-dimensional stress prediction and visualization method of each embodiment described above. Details not described in the embodiment of the coal seam three-dimensional stress prediction and visualization system can be referred to the embodiment of the coal seam three-dimensional stress prediction and visualization method described above.​​

[0348] As Figure 2 shown, the coal seam three-dimensional stress prediction and visualization system comprises:

[0349] a data acquisition module, configured to set an artificial seismic source and a plurality of detection points in a to-be-detected region of a coal seam, excite a seismic source signal at the artificial seismic source, and synchronously acquire seismic wave response signals and stress values of the detection points;

[0350] a wave field forward simulation module, configured to perform wave field forward simulation on the seismic source signal based on a time fractional order viscoelastic wave equation to obtain a forward simulation wave field;

[0351] a full waveform inversion module, configured to perform full waveform inversion on the forward simulation wave field and the seismic wave response signals to output three-dimensional wave velocity data of the to-be-detected region;

[0352] a reverse-time migration imaging module, configured to perform reverse-time migration imaging on the seismic wave response signals based on the three-dimensional wave velocity data to obtain a three-dimensional coal seam structure mask image;

[0353] a two-dimensional stress prediction module, configured to combine three-dimensional wave velocity data and a three-dimensional coal seam structure mask image after being sliced along the same spatial dimension to form an input tensor, input a pre-trained convolutional neural network, and output a two-dimensional stress prediction slice;

[0354] a three-dimensional visualization module, configured to:

[0355] splice all two-dimensional stress prediction slices to construct a three-dimensional stress body, and generate three-dimensional stress point cloud data based on the three-dimensional stress body;

[0356] convert the three-dimensional stress point cloud data into a three-dimensional stress grid model through a three-dimensional reconstruction algorithm;

[0357] delineate a dangerous region in the three-dimensional stress grid model using a mechanical analysis method;

[0358] combine the three-dimensional stress grid model and the three-dimensional coal seam structure mask image, perform stress value-color mapping rendering, generate a structure-stress integrated visual three-dimensional cloud image, and mark the dangerous region in the visual three-dimensional cloud image.

[0359] The coal seam three-dimensional stress prediction and visualization system of the embodiment is used to implement the coal seam three-dimensional stress prediction and visualization method.

[0360] The application also provides an electronic device for implementing various embodiments of the application, Figure 3 a hardware structure schematic diagram of an electronic device for implementing various embodiments of the application, as Figure 3 shown, the electronic device comprises a memory, a processor, and a computer program stored on the memory and executable on the processor.

[0361] Those skilled in the art can understand that the electronic device structure involved in the embodiments of the present application does not constitute a limitation on the electronic device, and the electronic device can include more or fewer components than the illustration, or combine certain components, or different component arrangements.

[0362] In the embodiments of the present application, the electronic device includes but is not limited to a laptop computer, a desktop computer, a workstation, a personal digital assistant, a server, a blade server, a mainframe computer, and other suitable computers. The electronic device can also represent various forms of mobile devices and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be examples only, and are not intended to limit the implementations of the embodiments of the present application described and / or claimed herein.

[0363] In the embodiments of the present application, the processor can be implemented by using at least one of an application specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a processor, a controller, a microcontroller, a microprocessor, an electronic unit designed to perform the functions described herein, and in some cases, such an implementation can be implemented in a controller. For software implementation, the implementation of such as a process or a function can be implemented with a separate software module allowing at least one function or operation to be performed. The software code can be implemented by a software application (or program) written in any appropriate programming language, which can be stored in a memory and executed by a controller.

[0364] In addition, the electronic device includes some functional modules that are not shown and will not be described here.

[0365] Those skilled in the art can understand that various aspects of the present application provide an electronic device can be implemented as a system, a method or a program product. Therefore, various aspects of the present application can be embodied as a complete hardware implementation, a complete software implementation (including firmware, microcode, etc.), or a combination of hardware and software aspects, which can be collectively referred to as "circuitry", "module" or "system" here.

[0366] The application also provides a storage medium, in which a program product capable of realizing the coal seam three-dimensional stress prediction and visualization method is stored. In some possible embodiments, various aspects of the application can also be implemented in the form of a program product, which includes program codes for causing a terminal device to perform the steps according to various exemplary embodiments of the application described in the above “Exemplary Method” section of the specification when the program product is run on the terminal device.

[0367] The storage medium can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium may, for example, be but is not limited to an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or apparatus, or any combination thereof. More specific examples (non-exhaustive list) of the readable storage medium include an electrical connection having one or more wires, a portable disc, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0368] The above description of disclosed embodiments enables one of ordinary skill in the art to make or use the application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein can be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for predicting and visualizing three-dimensional stress in coal seams, characterized in that, include: S1. Set up artificial seismic sources and multiple detection points in the area of ​​the coal seam to be detected. Excite source signals at the artificial seismic sources and simultaneously collect seismic wave response signals and measured stress values ​​at each detection point. S2. Based on the source signal, a forward modeling simulation of the wave field is performed using the time-fractional viscoelastic wave equation to obtain the forward simulated wave field. The short-memory fast algorithm is used to solve the fractional derivative terms of the time fractional viscoelastic wave equation; Full waveform inversion is performed based on the forward simulated wavefield and seismic wave response signal to output three-dimensional wave velocity data of the area to be investigated. ; in, It is a spatial position vector; It is a time variable; S3. Based on the three-dimensional wave velocity data, reverse time migration imaging is performed on the seismic wave response signal to obtain a three-dimensional coal seam structure mask image, including: Time-reversal is performed on the seismic wave response signals at each detection point to generate virtual source signals; Using a virtual seismic source signal as the excitation source, the wave field is back propagated based on three-dimensional wave velocity data to generate a back propagation wave field. Spatiotemporal cross-correlation calculations were performed on the forward simulated wavefield and the backward propagating wavefield to obtain the three-dimensional imaging values; Gradient enhancement and threshold segmentation are performed on the 3D imaging values ​​to generate a 3D coal seam structure mask map that identifies geological boundaries; S4. After slicing the three-dimensional wave velocity data and the three-dimensional coal seam structure mask along the same spatial dimension, combine them into an input tensor, input it into a pre-trained convolutional neural network, and output a two-dimensional stress prediction slice. S5. Construct a three-dimensional stress volume by stitching together all the two-dimensional stress prediction slices, and generate three-dimensional stress point cloud data based on the three-dimensional stress volume; The three-dimensional stress point cloud data is transformed into a three-dimensional stress mesh model through a three-dimensional reconstruction algorithm; Critical areas were delineated in a three-dimensional stress mesh model using mechanical analysis methods. By combining the 3D stress mesh model and the 3D coal seam structure mask, stress value-color mapping rendering is performed to generate a visualized 3D cloud map integrating structure and stress, and dangerous areas are marked in the visualized 3D cloud map.

2. The method for predicting and visualizing three-dimensional stress in coal seams as described in claim 1, characterized in that, In step S2, the time-fractional viscoelastic wave equation is: in, Density of the coal seam medium; It is a displacement vector field; For the displacement vector field in directional components, ; Let be the stress tensor components, where The normal to the plane where stress acts. Represents the direction of stress application. ; for Spatial coordinate components of direction; The focal term represents... The volume force in the direction satisfies: Coordinates of the artificial earthquake source; It is the spatial Dirac function; The time function of the source signal; The first viscoelastic constant is determined through laboratory dynamic mechanical testing. The second viscoelastic constant is determined through laboratory dynamic mechanical testing. For Kronek functions, hour =1, otherwise =0; For fractional order, ∈[0.3,0.8]; For displacement divergence; For strain tensor, ; The formula for solving fractional derivative terms using the short memory fast algorithm is: in, This is the current time step; The step size is one time step; L is the finite length of the memory window; fractional order The determined weighting coefficients; The forward wave modeling simulation uses the finite difference method to discretize the wave equations and calculates the forward simulated wave field over the entire spatial domain. .

3. The method for predicting and visualizing three-dimensional stress in coal seams as described in claim 1, characterized in that, Step S3 includes: S301. Based on seismic wave response signals The virtual seismic source signal is generated by the following expression: in, For the first Each detection point in time variable The seismic wave response signal; This represents the total recording time of the seismic wave response signal; For the first Each detection point in time variable The virtual seismic source signal; Will As a virtual seismic source applied to the corresponding detection point, the wavefield is backpropagated using the same time-fractional viscoelastic wave equation based on the three-dimensional wave velocity data, resulting in the backpropagated wavefield. ; S302. Calculate three-dimensional imaging values : ； S303. After performing gradient enhancement and high-pass filtering on the 3D imaging values, set a threshold. According to threshold The expression for generating a structure mask image is: 。 4. The method for predicting and visualizing three-dimensional stress in coal seams as described in claim 1, characterized in that, Step S4 includes: S401. Transfer three-dimensional wave velocity data With 3D coal seam structure mask image Slicing along the Y-axis with a fixed step size yields a sequence of wave velocity data slices. Coal seam structure mask image slice sequence ; in, The total number of slices, Indicates the first Zhang slices; S402. Combine the wave velocity data slices with the corresponding coal seam structure mask image slices into an input tensor: ； S403. Input the input tensor into a pre-trained convolutional neural network, and output a two-dimensional stress prediction slice. .

5. The method for predicting and visualizing three-dimensional stress in coal seams as described in claim 1, characterized in that, In step S4, the convolutional neural network is a U-Net encoder-decoder structure, including an encoder and a decoder. Skip connections are used between the encoder and decoder, wherein: The encoder consists of multiple sequentially connected coding blocks. Each coding block contains two 3×3 convolutional layers. Each convolutional layer is followed by a batch normalization layer and a ReLU activation function. At the end of each coding block, a 2×2 max pooling layer with a stride of 2 is connected to perform spatial downsampling. The decoder consists of multiple sequentially connected decoding blocks. Each decoding block contains a 2×2 transposed convolutional layer with a stride of 2, which is used to perform spatial upsampling. After the transposed convolutional layer, a feature concatenation layer is connected. The feature concatenation layer concatenates the feature map output by the transposed convolutional layer with the encoder feature map of the corresponding layer in the channel dimension. After concatenation, two 3×3 convolutional layers are connected. Each convolutional layer is followed by a batch normalization layer and a ReLU activation function. Skip connections connect the encoder feature map output from each coding block in the encoder to the corresponding feature concatenation layer in the decoder.

6. The method for predicting and visualizing three-dimensional stress in coal seams as described in claim 5, characterized in that, The training samples used by the convolutional neural network are derived from three-dimensional wave velocity data obtained by inverting seismic wave response signals through the time fractional viscoelastic wave equation and three-dimensional coal seam structure mask images extracted based on inverse time migration imaging technology. The convolutional neural network uses measured stress interpolation maps as labels for semi-supervised training and is jointly trained on multiple geological samples. The measured stress interpolation maps are coarse-resolution stress distribution maps generated by interpolation methods based on the measured stress values ​​of each detection point. The loss function of a convolutional neural network is: in, Two-dimensional stress prediction slices; To and A measured stress interpolation graph composed of the measured stress values ​​of all detection points at the same height on the Y-axis; , , These are the weighting coefficients. MAE stands for Mean Absolute Error; SSIM is the structural similarity index; Edgeloss is the boundary alignment loss, which is the difference between the predicted stress gradient and the measured stress gradient at the geological boundary marked by the 3D coal seam structure mask.

7. The method for predicting and visualizing three-dimensional stress in coal seams as described in claim 1, characterized in that, Step S5, the steps for generating three-dimensional stress point cloud data, include: S501. Slice all two-dimensional stress predictions The parts are then pieced back together according to the original spatial order to form a complete three-dimensional stress body. ； in, This represents the total number of slices. Indicates the first Zhang slices; S502. Perform voxel sampling on the 3D stress volume, extract voxel coordinates and predicted stress values, and generate 3D stress point cloud data: in, For the third stress body in three-dimensional stress body Spatial coordinates of each sampling point; For the third stress body in three-dimensional stress body Predicted stress values ​​at each sampling point; This is the lower limit threshold for stress display.

8. A three-dimensional stress prediction and visualization system for coal seams, characterized in that, A method for predicting and visualizing three-dimensional stress in coal seams as described in any one of claims 1-7, comprising: The data acquisition module is used to set up artificial seismic sources and multiple detection points in the area to be explored in the coal seam, to excite source signals at the artificial seismic sources, and to simultaneously acquire seismic wave response signals and measured stress values ​​at each detection point; The wavefield forward modeling module is used to perform forward modeling of the wavefield based on the source signal and the time-fractional viscoelastic wave equation to obtain the forward simulated wavefield. The full waveform inversion module is used to perform full waveform inversion based on the forward simulated wavefield and seismic wave response signal, and output three-dimensional wave velocity data of the area to be detected. The reverse time migration imaging module is used to perform reverse time migration imaging on the seismic wave response signal based on three-dimensional wave velocity data to obtain a three-dimensional coal seam structure mask. The two-dimensional stress prediction module is used to combine three-dimensional wave velocity data and three-dimensional coal seam structure mask image into an input tensor after slicing along the same spatial dimension. The input tensor is then fed into a pre-trained convolutional neural network, and the output is a two-dimensional stress prediction slice. The 3D visualization module is used for: All two-dimensional stress prediction slices are stitched together to construct a three-dimensional stress volume, and three-dimensional stress point cloud data is generated based on the three-dimensional stress volume. The three-dimensional stress point cloud data is transformed into a three-dimensional stress mesh model through a three-dimensional reconstruction algorithm; Critical areas were delineated in a three-dimensional stress mesh model using mechanical analysis methods. By combining the 3D stress mesh model and the 3D coal seam structure mask, stress value-color mapping rendering is performed to generate a visualized 3D cloud map integrating structure and stress, and dangerous areas are marked in the visualized 3D cloud map.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes a computer program, it implements the steps of the coal seam three-dimensional stress prediction and visualization method as described in any one of claims 1-7.

10. A storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the coal seam three-dimensional stress prediction and visualization method as described in any one of claims 1-7.

Images