A method and system for processing CT scan data of underground caverns with filled structural surfaces

Abstract

This invention relates to the field of underground cavern engineering inspection technology, and discloses a method and system for processing CT scan data of underground caverns containing infilled structural surfaces. The method acquires three-dimensional CT images of the infilled structural surfaces, and collects shear displacement data and environmental parameters in real time. A U-Net neural network is used to segment the infilled structural surfaces in the three-dimensional CT images to obtain their features. Digital image correlation and sensor-acquired data are used to calculate the shear displacement field data of the infilled structural surfaces. Based on the extracted infilled structural surface features and shear displacement field data, a damage evolution model of the structural surfaces is constructed. The finite element method is used to simulate the plastic zone expansion path of the infilled structural surfaces, reconstructing the damage evolution process. The reconstructed damage evolution process of the infilled structural surfaces is then visualized in three dimensions, presenting the damage state and shear displacement of the infilled structural surfaces. This invention improves the accuracy and efficiency of inspection.

Description

Technical Field

[0001] This invention relates to the field of underground cavern engineering inspection technology, specifically to a method and system for processing CT scan data of underground caverns containing filled structural surfaces. Background Technology

[0002] During the construction and operation of underground caverns, the stability of the infill structure has a crucial impact on the safety of the cavern. The damage evolution and shear displacement of the infill structure are key factors in assessing its stability. Currently, CT scanning technology has been applied to the detection of underground cavern structural surfaces, which can obtain detailed image information of the structural surfaces. However, existing CT scan data processing methods have the following problems when processing data containing infill structures: First, it is difficult to accurately extract the features of the infill structure from CT scan data, resulting in insufficient accuracy in the identification and positioning of the infill structure; second, it is impossible to monitor the shear displacement of the fault in real time, and cannot reflect the dynamic changes of the fault in a timely manner; third, in the process of reconstructing the damage evolution of the structural surface, there is a lack of effective algorithms to integrate multi-source data, making the reconstruction results of damage evolution inaccurate and incomplete. Summary of the Invention

[0003] The purpose of this invention is to solve the above-mentioned problems by designing a method and system for processing CT scan data of underground caverns with infilled structural surfaces.

[0004] The first aspect of this invention provides a method for processing CT scan data of underground caverns containing filled structural surfaces, the method comprising the following steps: Acquire 3D CT images of the filled structure surface, and collect shear displacement data and environmental parameters in real time through sensors deployed on the filled structure surface; The U-Net neural network was used to segment the filling structure surface of the three-dimensional CT image, and the contour and pore distribution features of the structure surface were extracted to obtain the filling structure surface features; The shear displacement field data of the infill structure surface is calculated using digital image correlation and data collected by sensors. Based on the extracted features of the filling structure and shear displacement field data, a damage evolution model of the structure is constructed. Combining the gray-level gradient information of the three-dimensional CT image, the finite element method is used to simulate the expansion path of the plastic zone of the filling structure and reconstruct the damage evolution process of the filling structure. The damage evolution process of the reconstructed filling structure surface is displayed in three dimensions, showing the damage state and shear displacement of the filling structure surface.

[0005] Optionally, in a first implementation of the first aspect of the present invention, the U-Net neural network includes an encoder and a decoder, wherein the encoder consists of multiple 3D convolutional layers with a kernel size of 3×3×3 and a stride of 1, and max pooling layers with a kernel size of 2×2×2 and a stride of 2. The decoder consists of deconvolutional and convolutional layers with a kernel size of 2×2×2 and a stride of 2; The intermediate connection layer uses skip connections to combine shallow features from the encoder with deep features from the decoder.

[0006] Optionally, in a second implementation of the first aspect of the present invention, the step of using a U-Net neural network to segment the filling structure surface of the three-dimensional CT image, extracting the structure surface contour and pore distribution features, and obtaining the filling structure surface features includes: The original 3D CT image is preprocessed and then input into the U-Net neural network. The encoder gradually reduces the size of the feature map and captures features at different scales, including shallow features and deep semantic features. By directly connecting feature maps of the same scale in the encoder to the corresponding layer of the decoder, the feature maps are upsampled and fused, and the decoder outputs a probability map with the same size as the original 3D CT image. Image segmentation and post-processing are performed on the probability map output by the decoder to obtain the filling structure surface features.

[0007] Optionally, in a third implementation of the first aspect of the present invention, the step of using digital image correlation and sensor-acquired data to calculate the shear displacement field data of the filled structure surface includes: Preprocessing of 3D CT images at different times includes denoising, grayscale normalization, and image registration. The three-dimensional Harris corner detection algorithm is used to extract feature points from the infill structure surface and surrounding rock area in the preprocessed image; Centered on each feature point, a 3D sub-image of size 15×15×15 pixels is selected as a template. The corresponding region is searched in the target image, the cross-correlation coefficient is calculated, and the region with a cross-correlation coefficient greater than 0.8 is taken as the matching point. The displacement components of each point in the X, Y, and Z directions are calculated based on the coordinates of the matching feature points. The discrete displacement points are interpolated using the cubic spline interpolation method to obtain continuous shear displacement field data. By combining the shear displacement data collected by the sensor, the interpolated shear displacement field data is calibrated to finally obtain the shear displacement field data of the filled structure surface.

[0008] Optionally, in a fourth implementation of the first aspect of the present invention, the step of constructing a structural surface damage evolution model based on the extracted features of the filling structure surface and shear displacement field data includes: A three-dimensional geometric model of the filling structure is constructed based on the extracted features of the filling structure, and initial material parameters and damage variables are imported. Shear displacement field data is obtained according to preset displacement increments, stress distribution in each region of the structure is calculated, and elements with stress exceeding yield strength are identified. For elements exceeding the yield strength, the damage variable increment is calculated based on pore characteristics and current shear displacement data, and the element stiffness and bearing capacity are updated accordingly. Regions with damage variables greater than or equal to 0.5 are designated as potential crack propagation zones, and the structural surface damage evolution model is obtained through iterative optimization.

[0009] Optionally, in a fifth implementation of the first aspect of the present invention, the step of combining the grayscale gradient information of three-dimensional CT images and using the finite element method to simulate the plastic zone expansion path of the filled structure surface and reconstruct the damage evolution process of the filled structure surface includes: The grayscale gradient information of the 3D CT image is converted into the initial state of material damage. An implicit finite element solver is used to perform nonlinear static analysis. Shear displacement loads are applied step by step to calculate the stress, strain and damage distribution of the structural surface and obtain the equivalent plasticity. When the equivalent plastic strain exceeds the critical value of 0.01, it is determined that the plastic state has been entered, and the expansion path of the plastic region is recorded.

[0010] Optionally, in a sixth implementation of the first aspect of the present invention, the region with a large gray-scale gradient is used as the region with a higher initial damage value, and the initial damage value is determined by a mapping function.

[0011] A second aspect of the present invention provides a CT scan data processing system for underground caverns containing filled structural surfaces, the system comprising: The acquisition module is used to acquire three-dimensional CT images of the filled structure surface and to collect shear displacement data and environmental parameters in real time through sensors arranged on the filled structure surface. The extraction module is used to segment the filling structure surface of the three-dimensional CT image using the U-Net neural network, extract the structure surface contour and pore distribution features, and obtain the filling structure surface features. The calculation module is used to calculate the shear displacement field data of the infill structure surface using digital image correlation and data collected by sensors; The evolution module is used to construct a damage evolution model of the filling structure based on the extracted features of the filling structure and shear displacement field data. It combines the gray-level gradient information of the three-dimensional CT image and uses the finite element method to simulate the expansion path of the plastic zone of the filling structure and reconstruct the damage evolution process of the filling structure. The visualization module is used to display the damage evolution process of the reconstructed filling structure surface in three dimensions, presenting the damage state and shear displacement of the filling structure surface.

[0012] A third aspect of the present invention provides a CT scan data processing device for underground caverns with infilled structural surfaces, the CT scan data processing device for underground caverns with infilled structural surfaces including a memory and at least one processor, the memory storing instructions; the at least one processor calling the instructions in the memory to cause the CT scan data processing device for underground caverns with infilled structural surfaces to perform the various steps of the CT scan data processing method for underground caverns with infilled structural surfaces as described in any of the preceding claims.

[0013] A fourth aspect of the present invention provides a computer-readable storage medium storing instructions that, when executed by a processor, implement the steps of the CT scan data processing method for underground caverns with infilled structural surfaces as described in any of the preceding claims.

[0014] The technical solution provided by this invention involves acquiring a three-dimensional CT image of the filled structure surface, and using sensors deployed on the filled structure surface to collect shear displacement data and environmental parameters in real time. A U-Net neural network is used to segment the filled structure surface from the three-dimensional CT image, extracting the surface contour and pore distribution features to obtain the features of the filled structure surface. Digital image correlation (DIC) and the sensor-acquired data are used to calculate the shear displacement field data of the filled structure surface. Based on the extracted features and shear displacement field data, a damage evolution model of the filled structure surface is constructed. Combining the gray-level gradient information of the three-dimensional CT image, the finite element method is used to simulate the plastic zone expansion path of the filled structure surface, reconstructing the damage evolution process of the filled structure surface. The reconstructed damage evolution process of the filled structure surface is then visualized in three dimensions, presenting the damage state and shear displacement of the filled structure surface. This invention uses a U-Net neural network for filled structure surface segmentation, automatically extracts complex structural features, improves processing efficiency and accuracy, reduces manual intervention, integrates gray-level gradient information to construct a damage evolution model, and combines the finite element method to simulate the plastic zone expansion, realistically reflecting the damage process of the filled structure surface in the actual environment. This effectively solves the problems of existing CT... This paper addresses the challenges of using scanning data processing methods in the detection of underground caverns with infilled structural surfaces, aiming to improve detection accuracy and efficiency and provide strong technical support for the safety assessment and maintenance of underground caverns. Attached Figure Description

[0015] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention.

[0016] Figure 1 A flowchart illustrating a method for processing CT scan data of underground caverns containing filled structural surfaces, provided in an embodiment of the present invention; Figure 2This is a schematic diagram of the structure of the CT scan data processing system for underground caverns with filled structural surfaces provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of a CT scan data processing device for underground caverns with filled structural surfaces, provided in an embodiment of the present invention. Detailed Implementation

[0017] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” or “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, apparatus, product, or device that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0018] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 A flowchart of a method for processing CT scan data of underground caverns containing filled structural surfaces provided in this embodiment of the invention. The method specifically includes the following steps: Step 101: Obtain a three-dimensional CT image of the filled structure surface and collect shear displacement data and environmental parameters in real time through sensors placed on the filled structure surface. In this embodiment, the area to be scanned in the underground cavern is cleaned to remove surface debris, water, and other interfering materials. Positioning markers are placed near the filling structure surface. The CT scanning equipment is aligned with the filling structure surface, and the scanning center position is determined by the laser positioning system. A spiral scanning mode is used to continuously scan from the top to the bottom of the cavern to ensure coverage of the entire filling structure surface and the surrounding rock within a certain range. During the scanning process, the three-dimensional CT image data of the filling structure surface is acquired in real time. At the same time, the pixel density matrix in the initial state is recorded. The matrix format is a three-dimensional array, where X, Y, and Z are the number of pixels in the horizontal, vertical, and depth directions of the image, respectively, and each pixel stores the corresponding density value. High-precision displacement sensors are symmetrically arranged in the surrounding rock on both sides of the infill structure. Distributed fiber Bragg grating displacement sensors are used, with a sensor spacing of 0.5-1m, evenly distributed along the fault strike and dip direction, with 3-5 rows on each side and 2-4 sensors in each row, ensuring coverage of the main shear deformation area of ​​the infill structure. Multi-parameter environmental sensors are also deployed, including temperature sensors, humidity sensors, air pressure sensors, and geostress sensors. The environmental sensors are placed in different locations in the cavern, such as the surrounding rock surface near the infill structure, the cavern top, and the sidewalls, with a spacing of 5-10m, to obtain comprehensive environmental parameters. A local coordinate system for the 3D CT image is established, with the geometric center of the filling structure as the origin, the X-axis along the fault direction, the Y-axis perpendicular to the fault direction pointing to the outside of the cave, and the Z-axis vertically upward. The coordinate system of the sensor adopts the cave engineering coordinate system. By setting global control points in the cave, the transformation relationship between the local coordinate system and the engineering coordinate system is established. The transformation parameters include translation vector and rotation matrix. For each sensor, its coordinates in the engineering coordinate system are recorded. The coordinates are then converted to the coordinates in the local coordinate system of the CT image. In the 3D CT image, a certain range of areas is selected as the mapping space of the sensor, with the sensor coordinates as the center. The shear displacement data and environmental parameters collected by the sensor are then associated with the CT image data of the corresponding area.

[0019] Step 102: Use U-Net neural network to segment the filling structure surface of the 3D CT image, extract the contour of the structure surface and the pore distribution features, and obtain the filling structure surface features; In this embodiment, the U-Net neural network includes an encoder and a decoder. The encoder consists of multiple 3D convolutional layers with a kernel size of 3×3×3 and a stride of 1, and max pooling layers with a kernel size of 2×2×2 and a stride of 2. The decoder consists of deconvolutional layers with a kernel size of 2×2×2 and a stride of 2, and convolutional layers. The intermediate connection layer adopts skip connections to combine the shallow features of the encoder with the deep features of the decoder. In this embodiment, the original 3D CT image is preprocessed and then input into the U-Net neural network. The encoder gradually reduces the size of the feature map to capture features at different scales, including shallow features and deep semantic features. Feature maps of the same scale in the encoder are directly connected to the corresponding layer of the decoder. The feature maps are upsampled and fused. The decoder outputs a probability map with the same size as the original 3D CT image. The probability map output by the decoder is then subjected to image segmentation and post-processing to obtain the filling structure surface features.

[0020] In this embodiment, typical regions containing different types of filling structures are selected from the acquired 3D CT images. The contours of the filling structures and the distribution of pores are manually labeled. The labeled data includes the binary mask of the filling structure, the location and size of the pores, and a total of 200-500 sets of training samples are prepared, of which 70% are used for training, 20% for validation, and 10% for testing. The Adam optimizer is used, with the learning rate initialized to 1e-4 and decaying by 0.5 every 50 epochs. The loss function combines Dice loss and cross-entropy loss, with a Dice coefficient weight of 0.6 and a cross-entropy weight of 0.4. During training, data augmentation techniques are used to improve the model's generalization ability. 100 batches are trained per epoch, with a batch size of 8, for a total of 200-300 training epochs. In this embodiment, the decoder outputs a probability map of the same size as the input image. Each voxel value represents the probability that it belongs to the filled structure surface. The probabilities are then adjusted using an adaptive threshold or a fixed threshold. Figure 2 The initial segmentation result is obtained by value-based processing. Morphological operations such as dilation, erosion, and opening / closing are applied to remove small noise regions, fill small voids inside the pores, smooth the boundaries of the structural surfaces, identify and separate different connected regions, filter out mis-segmented small regions based on features such as region area and shape, and retain the true filling structural surface regions. Within the segmented structural surface regions, the pore distribution is further identified, and parameters such as the number, size, location, and shape of the pores are calculated to generate a pore distribution map. Calculate the geometric parameters of the filling structure surface, such as perimeter, area, aspect ratio, and circularity. Use the boundary tracking algorithm to extract the contour curve and use Fourier descriptors to characterize the contour shape features. Statistically analyze parameters such as the number, size, distribution density, and connectivity of pores, and draw a histogram of pore size distribution and a spatial distribution cloud map.

[0021] Step 103: Calculate the shear displacement field data of the filled structure surface using digital image correlation method and data collected by sensors; In this embodiment, the 3D CT images at different times are preprocessed, including denoising, grayscale normalization, and image registration. A 3D Harris corner detection algorithm is used to extract feature points from the filled structure surface and surrounding rock area in the preprocessed image. A 15×15×15 pixel 3D sub-image is selected as a template, centered on each feature point. The corresponding region is searched in the target image, and the cross-correlation coefficient is calculated. Regions with a cross-correlation coefficient greater than 0.8 are selected as matching points. The displacement components in the X, Y, and Z directions of each point are calculated based on the coordinates of the matching feature points. The discrete displacement points are interpolated using cubic spline interpolation to obtain continuous shear displacement field data. The interpolated shear displacement field data is calibrated by combining the shear displacement data collected by the sensor, ultimately yielding the shear displacement field data of the filled structure surface.

[0022] In this embodiment, the reference image is a 3D CT image acquired in the initial state, used to provide the original geometric features and grayscale distribution information of the infilled structural surface. The target image is a 3D CT image of the same scene acquired at subsequent monitoring times, forming a time series comparison with the reference image to capture the deformation differences of the infilled structural surface. By comparing the coordinate changes of feature points at the same location in the reference image and the target image, the shear displacement field of the infilled structural surface is calculated. The target image is a key data carrier reflecting the state of the structural surface after deformation, and together with the reference image, it constitutes the core input data of the digital image correlation method.

[0023] Step 104: Based on the extracted features of the filling structure surface and shear displacement field data, construct a damage evolution model of the structure surface. Combine the gray-level gradient information of the three-dimensional CT image and use the finite element method to simulate the expansion path of the plastic zone of the filling structure surface and reconstruct the damage evolution process of the filling structure surface. In this embodiment, a three-dimensional geometric model of the filled structural surface is constructed based on the extracted features of the filled structural surface, and initial material parameters and damage variables are imported; shear displacement field data are obtained according to preset displacement increments, stress distribution in each region of the structural surface is calculated, and elements with stress exceeding the yield strength are identified; for elements exceeding the yield strength, the damage variable increment is calculated based on pore characteristics and current shear displacement data, and the element stiffness and bearing capacity are updated; regions with damage variables greater than or equal to 0.5 are marked as potential crack propagation zones, and the structural surface damage evolution model is obtained through iterative optimization.

[0024] In this embodiment, the grayscale gradient information of the three-dimensional CT image is converted into the initial state of material damage. An implicit finite element solver is used to perform nonlinear static analysis, and shear displacement loads are applied step by step to calculate the stress, strain and damage distribution of the structural surface to obtain the equivalent plasticity. When the equivalent plastic strain exceeds the critical value of 0.01, it is determined that the plastic state has been entered, and the expansion path of the plastic zone is recorded.

[0025] In this embodiment, regions with large gray-scale gradients are designated as regions with higher initial damage values, and the initial damage value is determined through a mapping function. In this embodiment, the degradation process of the mechanical properties of the structural surface under shear displacement is described by quantifying the degree of damage to the structural surface, providing a basis for simulating the expansion of the plastic zone; based on the pore distribution characteristics, the closing / expansion behavior of a single pore under shear stress is simulated; combined with the profile undulation and surface roughness of the structural surface, the crack initiation path along the structural surface is analyzed; through the global distribution of the shear displacement field, the stress concentration area and the dominant direction of damage evolution of the entire structural surface are determined. Pore ​​damage degree is obtained based on the pore area change rate, reflecting the influence of pore structure on bearing capacity; contour damage degree is obtained based on the integrity of the structural surface contour, describing the degree of damage to the geometric morphology of the structural surface; comprehensive damage degree is formed by weighted fusion of pore and contour damage degree to form a unified damage variable. The extracted porosity, pore connectivity, profile undulation, and other parameters are used as the initial conditions for calculating damage variables. The high strain regions in the shear displacement field are automatically marked as hot spots for damage evolution, which preferentially trigger the initiation of microcracks. Assuming the structural surface material is an elastic-damage-plastic medium, the elastic stage follows Hooke's law, the damage stage introduces stiffness degradation, and the plastic stage uses the Mohr-Coulomb criterion to describe yielding behavior. Environmental parameters are incorporated to correct material parameters; for example, increased humidity leads to increased pore water pressure, reducing the effective cohesion of the structural surface; temperature changes affect the material's thermal expansion coefficient, altering the initial stress state. Shear displacement field data is used as the dynamic load input, and loading is performed step-by-step according to a time series to simulate the structural surface's response under the coupled action of static initial stress and dynamic shear displacement. For the three-dimensional displacement field, shear displacement along the structural surface direction, diagonal slip displacement, and vertical compression / tension displacement are distinguished, with a focus on the evolution of shear damage in the XY plane. The model parameters were adjusted through trial and error to make the damage distribution predicted by the model match the actual location of microcracks in the CT images; the softening parameters in the material constitutive model were corrected by comparing the displacement-stress curves of the structural surface measured by the sensor with the curves calculated by the model. The model-predicted damage area is spatially superimposed with the gray-scale abnormal area in the CT image, and the overlap rate is calculated. The change in structural surface permeability when the shear displacement reaches a specific value is statistically analyzed to verify the rationality of the correlation between porosity damage degree and permeability in the model. For areas with large model prediction deviations, the CT image feature extraction is traced to see if there are any omissions. Combined with the simulation results of multiple sets of different working conditions, a parameter-result mapping table for damage evolution is established.

[0026] In this embodiment, based on three-dimensional CT images, a geometric model of an underground cavern containing the filling structure surface is constructed using threshold segmentation and three-dimensional reconstruction techniques. The model scope covers an area 5-10 times the diameter of the cavity surrounding the infill structure to eliminate boundary effects. Tetrahedral meshes are used to divide the model, with mesh refinement applied to the infill structure and its surrounding area. An elastoplastic constitutive model is used for the surrounding rock material, while a damaged elastoplastic constitutive model is used for the infill structure. The mechanical behavior is described by combining damage evolution equations. Environmental parameters are used as boundary condition inputs. The initial stress field is applied based on sensor data, and temperature and humidity are corrected using the temperature-humidity correlation of material parameters. The grayscale gradient information of the 3D CT image is converted into the initial state of material damage. Areas with large grayscale gradients are considered as areas with higher initial damage values. The initial damage value of the element is determined by a mapping function and incorporated into the initial conditions of the finite element model. An implicit finite element solver is used for nonlinear static analysis, gradually applying shear displacement loads to calculate the stress, strain, and damage distribution of the structure. When the equivalent plastic strain of an element exceeds a critical value, the element is considered to have entered a plastic state, and the plastic zone expansion path is recorded. Through iterative calculations, the damage evolution process and plastic zone expansion of the infill structure under shear displacement are simulated.

[0027] Step 105: Visualize the damage evolution process of the reconstructed filling structure surface in three dimensions, presenting the damage state and shear displacement of the filling structure surface.

[0028] In this embodiment, professional 3D visualization software, such as ParaView, VTK, or Unity, is used in conjunction with Python programming language for secondary development to achieve dynamic display of the damage evolution process. The geometric model of the filling structure surface, plastic zone distribution, and shear displacement field data obtained from finite element simulation are converted into a visualization format, and mesh optimization and surface smoothing are performed. Physical quantities such as damage variables, shear displacement components, stress, and strain are mapped to visualization attributes such as color, transparency, and texture. For example, the degree of damage is represented by a red gradient, and the magnitude of shear displacement is represented by arrow length and color. The dynamic process of damage evolution of the filling structure surface is played in the order of scanning time, showing the entire process from the initiation to the expansion of the plastic zone. User interactive operations are supported, such as rotating, scaling, and translating the 3D scene, selecting specific areas to view detailed damage parameters and displacement data, and adding annotations and comments to explain key evolution stages. At the same time, a 3D overall view, a cross-sectional view of the filling structure surface, and data curves are displayed, providing a multi-dimensional analysis perspective.

[0029] In this embodiment, an innovative layered composite casting technology is used to accurately simulate the filling structure surface: 3D printed templates are used to control the fault dip angle, C40 concrete simulates the complete rock strata, gypsum / sand / rubber particle mixture simulates the fault fracture zone, and C20 concrete simulates the influence zone. A 0.5mm serrated PE film is pre-embedded to enhance the friction characteristics. The cavern forming adopts a soluble salt 3D printed mold core followed by water injection and dissolution process. The supporting three-dimensional loading system contains 6 sets of servo hydraulic actuators, integrating a laser displacement sensor, acoustic emission probe, and FBG fiber array to achieve real-time monitoring of fault shear displacement. CT scanning technology is used to reconstruct the structural surface damage evolution. The experimental method simulates in-situ stress loading through step-by-step dissolution excavation and DIC full-field strain analysis, innovatively solving the problem that traditional methods cannot simulate the progressive failure of faults.

[0030] Please see Figure 2 A schematic diagram of the structure of a CT scan data processing system for underground caverns with filled structural surfaces provided in an embodiment of the present invention. The system includes: The acquisition module is used to acquire three-dimensional CT images of the filled structure surface and to collect shear displacement data and environmental parameters in real time through sensors arranged on the filled structure surface. The extraction module is used to segment the filling structure surface of the three-dimensional CT image using the U-Net neural network, extract the structure surface contour and pore distribution features, and obtain the filling structure surface features. The calculation module is used to calculate the shear displacement field data of the infill structure surface using digital image correlation and data collected by sensors; The evolution module is used to construct a damage evolution model of the filling structure based on the extracted features of the filling structure and shear displacement field data. It combines the gray-level gradient information of the three-dimensional CT image and uses the finite element method to simulate the expansion path of the plastic zone of the filling structure and reconstruct the damage evolution process of the filling structure. The visualization module is used to display the damage evolution process of the reconstructed filling structure surface in three dimensions, presenting the damage state and shear displacement of the filling structure surface.

[0031] Figure 3This is a schematic diagram of a CT scan data processing device for underground caverns with infilled structural surfaces, provided by an embodiment of the present invention. The CT scan data processing device 300 for underground caverns with infilled structural surfaces can vary significantly due to different configurations or performance characteristics. It may include one or more central processing units (CPUs) 310 (e.g., one or more processors) and a memory 320, and one or more storage media 330 (e.g., one or more mass storage devices) storing application programs 333 or data 332. The memory 320 and storage media 330 can be temporary or persistent storage. The program stored in the storage media 330 may include one or more modules (not shown in the diagram), each module including a series of instruction operations on the CT scan data processing device 300 for underground caverns with infilled structural surfaces. Furthermore, the processor 310 may be configured to communicate with the storage media 330 and execute the series of instruction operations in the storage media 330 on the CT scan data processing device 300 for underground caverns with infilled structural surfaces to implement the method provided in the above embodiment.

[0032] The CT scan data processing equipment 300 for underground caverns with filled structural surfaces may also include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input / output interfaces 360, and / or one or more operating devices 331, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will understand that... Figure 3 The structure of the CT scan data processing equipment for underground caverns with filled structural surfaces shown does not constitute a limitation on the computer equipment provided by the present invention. It may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0033] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when the instructions are executed on a computer, cause the computer to perform the various steps of the CT scan data processing method for underground caverns with filled structural surfaces provided in the above embodiments.

[0034] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the above-described equipment or apparatus / unit can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0035] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0036] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for processing CT scan data of underground caverns containing infilled structural surfaces, characterized in that, The method includes the following steps: Acquire 3D CT images of the filled structure surface, and collect shear displacement data and environmental parameters in real time through sensors deployed on the filled structure surface; The U-Net neural network was used to segment the filling structure surface of the three-dimensional CT image, and the contour and pore distribution features of the structure surface were extracted to obtain the filling structure surface features; The shear displacement field data of the infill structure surface is calculated using digital image correlation and data collected by sensors. Based on the extracted features of the filling structure and shear displacement field data, a damage evolution model of the structure is constructed. Combining the gray-level gradient information of the three-dimensional CT image, the finite element method is used to simulate the expansion path of the plastic zone of the filling structure and reconstruct the damage evolution process of the filling structure. The damage evolution process of the reconstructed filling structure surface is displayed in three dimensions, showing the damage state and shear displacement of the filling structure surface. The calculation of shear displacement field data of the filled structure surface using digital image correlation and sensor-acquired data includes: Preprocessing of 3D CT images at different times includes denoising, grayscale normalization, and image registration. The three-dimensional Harris corner detection algorithm is used to extract feature points from the infill structure surface and surrounding rock area in the preprocessed image; Centered on each feature point, a 3D sub-image of size 15×15×15 pixels is selected as a template. The corresponding region is searched in the target image, the cross-correlation coefficient is calculated, and the region with a cross-correlation coefficient greater than 0.8 is taken as the matching point. The displacement components of each point in the X, Y, and Z directions are calculated based on the coordinates of the matching feature points. The discrete displacement points are interpolated using the cubic spline interpolation method to obtain continuous shear displacement field data. By combining the shear displacement data collected by the sensor, the interpolated shear displacement field data is calibrated to finally obtain the shear displacement field data of the filled structure surface. The structural damage evolution model, based on the extracted features of the infill structure and shear displacement field data, includes: A three-dimensional geometric model of the filling structure is constructed based on the extracted features of the filling structure, and initial material parameters and damage variables are imported. Shear displacement field data is obtained according to preset displacement increments, stress distribution in each region of the structure is calculated, and elements with stress exceeding yield strength are identified. For elements exceeding the yield strength, the damage variable increment is calculated based on pore characteristics and current shear displacement data, and the element stiffness and bearing capacity are updated accordingly. Regions with damage variables greater than or equal to 0.5 are designated as potential crack propagation zones, and the structural surface damage evolution model is obtained through iterative optimization. The method of combining grayscale gradient information from 3D CT images and using the finite element method to simulate the expansion path of the plastic zone of the filled structure surface and reconstructing the damage evolution process of the filled structure surface includes: The grayscale gradient information of the 3D CT image is converted into the initial state of material damage. An implicit finite element solver is used to perform nonlinear static analysis. Shear displacement loads are applied step by step to calculate the stress, strain and damage distribution of the structural surface and obtain the equivalent plasticity. When the equivalent plastic strain exceeds the critical value of 0.01, it is determined that the plastic state has been entered, and the expansion path of the plastic region is recorded.

2. The method for processing CT scan data of underground caverns with filled structural surfaces as described in claim 1, characterized in that, The U-Net neural network includes an encoder and a decoder. The encoder consists of multiple 3D convolutional layers with a kernel size of 3×3×3 and a stride of 1, and max pooling layers with a kernel size of 2×2×2 and a stride of 2. The decoder consists of deconvolutional and convolutional layers with a kernel size of 2×2×2 and a stride of 2; The intermediate connection layer uses skip connections to combine shallow features from the encoder with deep features from the decoder.

3. The method for processing CT scan data of underground caverns with filled structural surfaces as described in claim 2, characterized in that, The method employs a U-Net neural network to segment the filling structure surface of a 3D CT image, extracting the surface contour and pore distribution features to obtain the filling structure surface features, including: The original 3D CT image is preprocessed and then input into the U-Net neural network. The encoder gradually reduces the size of the feature map and captures features at different scales, including shallow features and deep semantic features. By directly connecting feature maps of the same scale in the encoder to the corresponding layer of the decoder, the feature maps are upsampled and fused, and the decoder outputs a probability map with the same size as the original 3D CT image. Image segmentation and post-processing are performed on the probability map output by the decoder to obtain the filling structure surface features.

4. The method for processing CT scan data of underground caverns with filled structural surfaces as described in claim 1, characterized in that, Regions with large gray-level gradients are considered regions with higher initial damage values, and the initial damage value is determined through a mapping function.

5. A CT scan data processing system for underground caverns containing infilled structural surfaces, characterized in that, The system includes: The acquisition module is used to acquire three-dimensional CT images of the filled structure surface and to collect shear displacement data and environmental parameters in real time through sensors arranged on the filled structure surface. The extraction module is used to segment the filling structure surface of the three-dimensional CT image using the U-Net neural network, extract the structure surface contour and pore distribution features, and obtain the filling structure surface features. The calculation module is used to calculate the shear displacement field data of the infill structure surface using digital image correlation and sensor-acquired data. It preprocesses 3D CT images at different times, including denoising, grayscale normalization, and image registration. A 3D Harris corner detection algorithm is used to extract feature points from the infill structure surface and surrounding rock area in the preprocessed image. A 15×15×15 pixel 3D sub-image is selected as a template centered on each feature point. The corresponding region is searched in the target image, and the cross-correlation coefficient is calculated. Regions with a cross-correlation coefficient greater than 0.8 are selected as matching points. The displacement components in the X, Y, and Z directions of each point are calculated based on the coordinates of the matching feature points. The discrete displacement points are interpolated using cubic spline interpolation to obtain continuous shear displacement field data. Finally, the interpolated shear displacement field data is calibrated using the sensor-acquired shear displacement data to obtain the shear displacement field data of the infill structure surface. The evolution module is used to construct a damage evolution model of the infill structure based on extracted features and shear displacement field data. Combining grayscale gradient information from 3D CT images, it uses the finite element method to simulate the expansion path of the plastic zone of the infill structure, reconstructing the damage evolution process: It constructs a 3D geometric model of the infill structure based on extracted features, imports initial material parameters and damage variables; acquires shear displacement field data according to preset displacement increments, calculates the stress distribution in each region of the structure, and identifies elements whose stress exceeds the yield strength; for elements exceeding the yield strength, it... Based on pore characteristics and current shear displacement data, the damage variable increment is calculated, and the element stiffness and bearing capacity are updated. Regions with damage variables greater than or equal to 0.5 are marked as potential crack propagation zones, and the structural surface damage evolution model is obtained through iterative optimization. The gray-level gradient information of the 3D CT image is converted into the initial state of material damage. An implicit finite element solver is used for nonlinear static analysis. Shear displacement loads are applied step by step, and the stress, strain, and damage distribution of the structural surface are calculated to obtain the equivalent plasticity. When the equivalent plastic strain exceeds the critical value of 0.01, it is determined that the plastic state has been entered, and the plastic zone propagation path is recorded. The visualization module is used to display the damage evolution process of the reconstructed filling structure surface in three dimensions, presenting the damage state and shear displacement of the filling structure surface.

6. A CT scan data processing device for underground caverns containing filled structural surfaces, characterized in that, The CT scan data processing device for underground caverns with filled structural surfaces includes a memory and at least one processor. The memory stores instructions. The at least one processor invokes the instructions in the memory to cause the CT scan data processing device for underground caverns with filled structural surfaces to perform the various steps of the CT scan data processing method for underground caverns with filled structural surfaces as described in any one of claims 1-4.

7. A computer-readable storage medium storing instructions thereon, characterized in that, When the instructions are executed by the processor, they implement the various steps of the CT scan data processing method for underground caverns with filled structural surfaces as described in any one of claims 1-4.

Images