A mechanical optimization analysis and design system for concrete pole connection

Abstract

The present application relates to the field of digital design and computational mechanics topology optimization of power engineering foundation, specifically to a concrete pole connection mechanical optimization analysis and design system, comprising: a micro image acquisition module: constructing microstructure image dataset; a heterogeneous space mapping module: constructing anisotropic permeability tensor field, thereby mapping the mechanical transmission path of solid materials to a digitally defined virtual fluid permeation field; a parallel flow field solving module: calculating to obtain a stress flux distribution thermodynamic map representing stress transmission strength; a topology inversion optimization module: reconstructing the geometric reinforcement parameters and material distribution of the connector based on the identification results; a design verification closed loop module: triggering a new round of simulation calculation until the preset convergence condition is met, outputting the final optimized design drawing; the present application solves the contradiction between microstructure complexity and engineering design timeliness, supports rapid verification and multiple optimization of design schemes.

Description

Technical Field

[0001] This invention relates to the field of digital design and computational mechanics topology optimization technology for power engineering infrastructure, specifically a mechanical optimization analysis and design system for concrete pole connections. Background Technology

[0002] In the field of power infrastructure design and mechanical analysis, refined modeling and structural optimization of the connection parts of concrete poles are the core links to ensure the reliability of the poles throughout their entire life cycle. At present, mechanical analysis in this field mainly relies on the traditional finite element analysis method based on mesh generation.

[0003] While traditional methods are widely used in macroscopic structural mechanics assessments, they face significant bottlenecks when dealing with typical heterogeneous microstructures like concrete. Converting industrial CT image sequences containing aggregates, mortar, and complex pores into high-fidelity finite element meshes is extremely lengthy and prone to misjudging stress concentration points due to mesh distortion or geometric model simplification. The geometric complexity at the microscale leads to a surge in computational degrees of freedom, triggering the impossible trinity problem of computational explosion, making large-scale 3D voxel data mechanical simulations difficult to achieve under the time-sensitive requirements of engineering design. Furthermore, the lack of effective visualization and mapping methods for microscopic stress transmission paths makes it difficult for designers to accurately identify efficient force transmission areas and redundant material areas when reconstructing the topology of key connectors such as flanges and steel rings, resulting in a lack of physical-level fine-grained support for lightweight layout schemes. Therefore, how to overcome the computational efficiency bottleneck caused by traditional mesh dependence while preserving the authenticity of the concrete microscopic topology, and achieve intuitive mapping and automated topology optimization of mechanical transmission paths, has become a pressing technical problem in the current field of image processing and mechanical simulation integration. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a mechanical optimization analysis and design system for concrete pole connections. Specifically, the technical solution of this invention includes:

[0005] Microscopic image acquisition module: configured to acquire industrial CT slice image sequences of the connection parts of concrete poles and perform data cleaning to construct a microstructure image dataset;

[0006] Heterogeneous space mapping module: configured to extract the grayscale features and texture gradient features of voxels in the microstructure image dataset, and construct an anisotropic permeability tensor field through a voxel grayscale to permeability mapping algorithm, thereby mapping the mechanical transmission path of the physical material into a digitally defined virtual fluid permeation field;

[0007] Parallel flow field solution module: configured as a GPU-parallel lattice Boltzmann method, it simulates the flow evolution of virtual fluid driven by pressure difference in the virtual fluid infiltration field and calculates the stress flux distribution thermogram characterizing the stress transmission intensity.

[0008] Topology inversion optimization module: configured to identify efficient force transmission regions and inefficient redundant regions in the stress flux distribution heat map, and reconstruct the geometric reinforcement parameters and material distribution of the connector based on the identification results;

[0009] Design verification closed-loop module: configured to feed back the reconstructed geometric reinforcement parameters to the model definition corresponding to the microstructure image dataset, trigger a new round of simulation calculations until the preset convergence conditions are met, and output the final optimized design drawings.

[0010] Optionally, modules can be integrated using the following methods:

[0011] S1. Obtain an industrial CT slice image sequence of the connection part of the concrete pole through the microscopic image acquisition module, perform noise reduction and binarization preprocessing on the image sequence, and construct a microscopic structure image dataset.

[0012] S2. Input the microstructure image dataset into the heterogeneous space mapping module, extract the gray value and gradient direction of each voxel point, and generate an anisotropic permeability tensor field that is mapped point-to-point with the image resolution through the anisotropic voxel permeability mapping function.

[0013] S3. Input the anisotropic permeability tensor field into the parallel flow field solving module, set virtual high-pressure boundary and virtual low-pressure boundary, use the lattice Boltzmann algorithm to calculate the flow state of the virtual fluid in the voxel grid in parallel, and output the stress flux distribution heat map containing the velocity scalar and the flow direction vector.

[0014] S4. Input the stress flux distribution heat map into the topology inversion optimization module, extract the high-velocity connected region as the main stress transmission channel, extract the low-velocity region as the material redundancy region, and adjust the topology parameters of the connector accordingly.

[0015] S5. Calculate the adjusted flow field energy dissipation index through the design verification closed-loop module, and determine whether the index meets the preset convergence condition. If it does, output the final optimized design drawing.

[0016] Optionally, S1 specifically includes:

[0017] S11. Use high-precision industrial CT scanning equipment to perform tomographic scanning on the connection interface of concrete poles to obtain raw DICOM format data;

[0018] S12. Perform artifact removal and noise reduction on the original DICOM format data to generate a high-resolution grayscale slice image;

[0019] S13. Perform 3D reconstruction and voxel alignment on the high-resolution grayscale slice image, remove background noise, and extract the region of interest containing aggregate, mortar and pore features.

[0020] S14. Serialize and store the image data of the region of interest according to spatial coordinates to construct a standardized microstructure image dataset.

[0021] Optionally, S2 specifically includes:

[0022] S21. Traverse each voxel in the microstructure image dataset and obtain the gray intensity value and gray gradient vector of that voxel.

[0023] S22. Construct an anisotropic voxel permeability mapping function, which is configured as follows: establish a positive correlation mapping between gray values ​​and virtual permeability, wherein high gray intensity values ​​representing high-density aggregates are mapped to high permeability values ​​to simulate the characteristics of easy stress transmission; and low gray intensity values ​​representing pores or cracks are mapped to permeability values ​​approaching zero to simulate stress blocking characteristics.

[0024] S23. Based on the direction of the gray-level gradient vector, construct a local rotation matrix to determine the principal axis direction of the local permeability tensor of the voxel point, so that the flow dominance direction of the virtual fluid is consistent with the material texture direction.

[0025] S24. Combine the local permeability tensors of all voxel points to generate a global permeability tensor field that can characterize the anisotropy of stress transmission within heterogeneous materials.

[0026] Optionally, S3 specifically includes:

[0027] S31. Load the anisotropic permeability tensor field data into the GPU's video memory to construct a virtual flow field calculation grid that corresponds one-to-one with the voxel grid.

[0028] S32. Set a constant high-pressure boundary condition at one end of the connection of the virtual flow field mesh and a constant low-pressure boundary condition at the other end of the connection to form a virtual pressure difference that drives the flow of virtual fluid.

[0029] S33. The collision step and migration step calculations of the lattice Boltzmann method are executed in parallel using the CUDA core. During the calculation, a matrix resistance term based on Darcy's law is introduced. The distribution function of fluid particles is corrected according to the local permeability tensor of each voxel point to simulate the obstruction effect of different material media on the flow.

[0030] S34. Perform iterative calculations until the relative error of the macroscopic fluid quantities across the entire field is less than the preset steady-state threshold. Extract the fluid velocity vector and flow density of each grid point to generate a visual stress flux distribution heat map.

[0031] Optionally, S4 specifically includes:

[0032] S41. Threshold segmentation is performed on the velocity scalar value of the stress flux distribution heatmap, and a high flux threshold and a low flux threshold are set, wherein the high flux threshold is strictly greater than the low flux threshold.

[0033] S42. Identify voxel connected regions with flow rates greater than or equal to the high throughput threshold, mark them as stress concentration main channels, and extract the centerline of the channel as the skeleton topology chain.

[0034] S43. Identify voxel regions with flow rates less than the low throughput threshold and mark them as invalid material regions;

[0035] S44. Calculate the local vorticity value at each point in the flow field, detect areas where the vorticity value exceeds the preset turbulence threshold, and mark them as potential fatigue fracture risk points.

[0036] S45. Execute parameter reconstruction logic: Increase reinforcement density in the coordinate region where the skeleton topology chain is located; perform material reduction operation in the coordinate region corresponding to the invalid material region to reduce the thickness of the connector; perform chamfer curvature smoothing processing on the potential fatigue fracture risk point to generate optimized geometric reinforcement parameters.

[0037] Optionally, S5 specifically includes:

[0038] S51. Obtain the optimized geometric reinforcement parameters and update the material distribution definition in the calculation model constructed based on the microstructure image dataset;

[0039] S52. Calculate the updated total energy dissipation rate of the flow field. If the change in the total energy dissipation rate is less than the preset convergence threshold, the optimization is considered complete.

[0040] S53. If the change in total energy dissipation rate is greater than or equal to the preset convergence threshold, then return to execute steps S2 to S4 for the next iteration.

[0041] S54. After the optimization is completed, convert the final geometric reinforcement parameters into CAD engineering drawing format and output the final optimized design drawings.

[0042] Optionally, the parallel flow field solution module, when performing calculations:

[0043] The distribution function of fluid particles is stored in double-precision floating-point format, and the shared memory block of the GPU is used for the exchange and synchronization of local mesh data.

[0044] For image data whose resolution exceeds the preset video memory limit, a multi-GPU spatial decomposition strategy is adopted to divide the anisotropic permeability tensor field into multiple sub-regions, which are then computed in parallel on different GPU nodes, and the computation results are merged through a boundary exchange protocol.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] 1. This invention directly maps voxels of sliced ​​images to computational units, eliminating the cumbersome geometric model reconstruction and mesh generation process in traditional finite element analysis. This processing method effectively eliminates numerical calculation deviations caused by mesh distortion, accurately preserves the true topological structure of aggregates, mortar and micropores inside concrete, and can more sensitively capture the micro-stress concentration at the pole connection interface, significantly improving the physical fidelity of the analysis results.

[0047] 2. This invention utilizes the mathematical isomorphism between fluid motion and stress transmission, combined with a large-scale parallel computing strategy, to transform the complex solution of heterogeneous solid mechanics into an efficient flow field evolution simulation. This innovation significantly reduces the computational complexity when processing massive amounts of microscopic data, shortening the analysis process that originally took several hours to minutes or even seconds. It successfully resolves the contradiction between the complexity of microstructures and the timeliness of engineering design, and supports the rapid verification and multiple optimization of design schemes.

[0048] 3. By constructing a virtual fluid infiltration field and generating a stress flux heat map, the system transforms the abstract and invisible mechanical transmission path inside concrete into an intuitive flow channel distribution. Designers can clearly identify the efficient force transmission channels and dead water redundancy zones at the connection points, realizing the transparency of the mechanical distribution. This intuitive physical feedback provides a scientific quantitative basis for the layout of flanges and reinforcing bars, avoiding the blindness caused by relying on experience in traditional design.

[0049] 4. Based on topological inversion logic, this invention achieves refined material layout and lightweight design while ensuring connection strength. At the same time, it innovatively introduces a vortex detection mechanism to identify potential fatigue risk areas and eliminates stress singularities through geometric smoothing. This not only significantly reduces the consumption of materials such as steel, but also effectively reduces the probability of fatigue failure at pole connections under long-term alternating stresses such as wind loads, thus greatly extending the service life of power infrastructure. Attached Figure Description

[0050] The present invention will be further explained below with reference to the accompanying drawings and embodiments:

[0051] Figure 1 This is a structural diagram of the system of the present invention. Detailed Implementation

[0052] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0053] Example 1:

[0054] Please see Figure 1 A mechanical optimization analysis and design system for concrete pole connections, comprising:

[0055] Microscopic image acquisition module: configured to acquire industrial CT slice image sequences of the connection parts of concrete poles and perform data cleaning to construct a microstructure image dataset;

[0056] Heterogeneous space mapping module: configured to extract the grayscale features and texture gradient features of voxels in the microstructure image dataset, and construct an anisotropic permeability tensor field through a voxel grayscale to permeability mapping algorithm, thereby mapping the mechanical transmission path of the physical material into a digitally defined virtual fluid permeation field;

[0057] Parallel flow field solution module: configured as a GPU-parallel lattice Boltzmann method, it simulates the flow evolution of virtual fluid driven by pressure difference in the virtual fluid infiltration field and calculates the stress flux distribution thermogram characterizing the stress transmission intensity.

[0058] Topology inversion optimization module: configured to identify efficient force transmission regions and inefficient redundant regions in the stress flux distribution heat map, and reconstruct the geometric reinforcement parameters and material distribution of the connector based on the identification results;

[0059] Design verification closed-loop module: configured to feed back the reconstructed geometric reinforcement parameters to the model definition corresponding to the microstructure image dataset, trigger a new round of simulation calculations until the preset convergence conditions are met, and output the final optimized design drawings.

[0060] This embodiment details the architecture logic of the mechanical optimization analysis and design system for concrete pole connections. This system aims to solve the difficulties in mesh generation and the impossible trinity problem of computational explosion faced by traditional finite element analysis (FEA) when dealing with heterogeneous microstructures of concrete. The microscopic image acquisition module, as the sensing front end of the system, is configured to communicate with industrial CT equipment to acquire CT slice image sequences of the concrete pole connection parts and perform data cleaning operations, including artifact removal, noise reduction, and enhancement, to construct a microstructure image dataset.

[0061] The heterogeneous space mapping module realizes the dimensionality reduction transformation from physical entity space to virtual computing space, extracts the gray-level features of voxels in the microstructure image dataset, which represent material density, and the texture gradient features, which represent material directionality. Through a preset voxel gray-level to permeability mapping algorithm, an anisotropic permeability tensor field is constructed. This step essentially maps the complex mechanical transmission path of physical materials mathematically isomorphically into a digitally defined virtual fluid infiltration field in a porous medium. The parallel flow field solving module uses fluid dynamics operators to solve solid mechanics problems. Based on the parallel lattice Boltzmann method (LBM) of the GPU graphics processor, it simulates the flow evolution process of virtual fluid driven by pressure difference in the virtual fluid infiltration field. By calculating the fluid velocity and flux, a stress flux distribution thermogram representing the stress transmission intensity inside the connection part is obtained.

[0062] In this mapping, high-velocity regions correspond to high-stress transmission regions, while low-velocity or stagnant regions correspond to material redundancy regions. Based on this, the topology inversion optimization module reconstructs the structure based on the visualized mechanical distribution, identifies the efficient force transmission regions (main channels) and inefficient redundant regions (dead water regions) in the stress flux distribution heatmap, and reconstructs the topology of the geometric reinforcement parameters and material distribution of connectors such as steel plate rings and flanges based on the identification results. The design verification closed-loop module ensures the convergence and reliability of the design, feeds back the reconstructed geometric reinforcement parameters to the model definition corresponding to the microstructure image dataset, updates the permeability field, triggers a new round of simulation calculations, until the energy dissipation index calculated by the system meets the preset convergence conditions, and finally outputs the optimized design drawings.

[0063] This embodiment establishes a stress-fluid isomorphic mapping mechanism, abandoning the traditional mesh-based finite element analysis path in the specific scenario of high-voltage power transmission support for concrete poles. It directly uses voxel points from microscopic images as computational units, avoiding the complex mesh generation process. Combined with the GPU-parallel LBM algorithm, the computational efficiency for heterogeneous microstructures is reduced from hours to seconds. The system realizes visualized topology optimization of the invisible stress paths inside concrete, effectively solving the core contradiction between the complexity of microstructures and the timeliness of engineering design, making it possible to carry out refined material layout in high-stress concentration areas such as pole flange connections.

[0064] The stress-fluid mapping mechanism used in this embodiment is based on the principle of minimum power consumption in mechanical systems. That is, the steady-state flow path of the virtual fluid in the anisotropic permeation field is consistent with the principal stress transmission path of the elastic body under load in terms of mathematical topology, thereby ensuring the physical rationality of inverting the stress distribution through the fluid momentum distribution.

[0065] Example 2:

[0066] In this embodiment, the modules are connected through the following method:

[0067] S1. Obtain an industrial CT slice image sequence of the connection part of the concrete pole through the microscopic image acquisition module, perform noise reduction and binarization preprocessing on the image sequence, and construct a microscopic structure image dataset.

[0068] S2. Input the microstructure image dataset into the heterogeneous space mapping module, extract the gray value and gradient direction of each voxel point, and generate an anisotropic permeability tensor field that is mapped point-to-point with the image resolution through the anisotropic voxel permeability mapping function.

[0069] S3. Input the anisotropic permeability tensor field into the parallel flow field solving module, set virtual high-pressure boundary and virtual low-pressure boundary, use the lattice Boltzmann algorithm to calculate the flow state of the virtual fluid in the voxel grid in parallel, and output the stress flux distribution heat map containing the velocity scalar and the flow direction vector.

[0070] S4. Input the stress flux distribution heat map into the topology inversion optimization module, extract the high-velocity connected region as the main stress transmission channel, extract the low-velocity region as the material redundancy region, and adjust the topology parameters of the connector accordingly; S5. Calculate the adjusted flow field energy dissipation index through the design verification closed-loop module, and determine whether the index meets the preset convergence condition. If it does, output the final optimized design drawing.

[0071] This embodiment details the specific implementation timing of the collaborative work between modules; the system executes step S1, constructing a micro-dataset, and acquires an industrial CT slice image sequence of the connection part of the concrete pole through the micro-image acquisition module; the system performs noise reduction preprocessing on the image sequence, such as Gaussian filtering and binarization, specifically using Otsu's method to automatically calculate the global threshold of the image. grayscale values The regions are labeled as solid materials, the environmental background is removed, and only the regions containing aggregates, mortar and steel bars are retained to construct a standardized microstructure image dataset;

[0072] The S2 tensor field generation step is executed, and the microstructure image dataset is input into the heterogeneous space mapping module; the module extracts the gray value of each voxel. With gradient direction Using the anisotropic voxel permeability mapping function, each voxel point is converted into a second-order tensor, generating an anisotropic permeability tensor field that is mapped point-to-point to the image resolution. In the S3 parallel flow field solution stage, the anisotropic permeability tensor field is input into the parallel flow field solution module. The system sets virtual high-pressure boundaries at both ends of the connection interface, such as the upper and lower faces. With virtual low-pressure boundary The Lattice Boltzmann (LBM) algorithm is used to perform parallel computation of the distribution function evolution of virtual fluid in a voxel grid, and outputs a stress flux distribution thermogram containing velocity scalars and flow direction vectors.

[0073] Based on this, the S4 topology adjustment is performed, inputting the stress flux distribution heat map into the topology inversion optimization module; high-velocity connected regions are extracted as main stress transmission channels, i.e., regions to be retained or reinforced, and low-velocity regions are extracted as material redundancy areas, i.e., regions to be drilled or thinned, and the topology parameters of the connectors are automatically adjusted accordingly; the S5 closed-loop verification is performed, and the energy dissipation index of the adjusted flow field is calculated through the design verification closed-loop module; in response to the rate of change of this index meeting the preset convergence condition, such as the change in dissipation rate being less than 0.1% in two consecutive iterations, the system outputs the final optimized design drawings;

[0074] This embodiment constructs a complete closed-loop workflow of perception-simulation-optimization-verification, which is particularly suitable for the refined design of concrete pole connectors. Through tensor field generation in S2, the system can accurately capture the subtle influence of concrete material anisotropy on force transmission. Through flow velocity-based topology adjustment in S4, a biomimetic trabecular optimal force transmission structure can be generated. This method significantly reduces the amount of steel used while ensuring connection strength, achieving the dual goals of lightweight and high strength in pole connections.

[0075] Example 3:

[0076] S1 specifically includes:

[0077] S11. Use high-precision industrial CT scanning equipment to perform tomographic scanning on the connection interface of concrete poles to obtain raw DICOM format data;

[0078] S12. Perform artifact removal and noise reduction on the original DICOM format data to generate a high-resolution grayscale slice image;

[0079] S13. Perform 3D reconstruction and voxel alignment on the high-resolution grayscale slice image, remove background noise, and extract the region of interest containing aggregate, mortar and pore features.

[0080] S14. Serialize and store the image data of the region of interest according to spatial coordinates to construct a standardized microstructure image dataset.

[0081] This embodiment specifies step S1, particularly clarifying how to acquire high-quality microscopic image data. Step S11, DICOM data acquisition, utilizes a high-precision industrial CT scanning device with an accuracy of 50μm to perform a tomographic scan of the concrete pole connection interface, acquiring raw DICOM format data. Step S12, artifact removal, removes hardening artifacts and reduces noise from the raw DICOM format data, generating a high-resolution grayscale slice image. Noise reduction employs nonlocal mean filtering to preserve the edge features of fine cracks. Step S13, voxel alignment, performs 3D reconstruction and voxel alignment on the high-resolution grayscale slice image. Voxel alignment refers to forcibly orthogonally aligning the 3D voxel coordinate axes of the image with the Euler grid coordinate axes used for subsequent fluid calculations, removing background noise and extracting the region of interest (ROI) containing aggregate, mortar, and pore features. Step S14, serialization storage, stores the image data of the ROI according to spatial coordinates. Serialization storage constructs a standardized dataset of microstructure images, ensuring data continuity in memory and preparing for GPU read optimization;

[0082] In this embodiment, the voxel alignment step S13 is of crucial significance in the microscopic analysis of concrete poles. It eliminates the intermediate errors in the geometric model to mesh conversion process in traditional methods, and directly establishes the data foundation of voxels, i.e., the mesh. This processing method ensures the geometric fidelity of the tiny pores inside the concrete and the aggregate interface, and prevents misjudgment of stress concentration caused by mesh distortion, thus laying a solid geometric data foundation for subsequent high-precision flow field simulation.

[0083] Example 4:

[0084] S2 specifically includes:

[0085] S21. Traverse each voxel in the microstructure image dataset and obtain the gray intensity value and gray gradient vector of that voxel.

[0086] S22. Construct an anisotropic voxel permeability mapping function, which is configured as follows: establish a positive correlation mapping between gray values ​​and virtual permeability, wherein high gray intensity values ​​representing high-density aggregates are mapped to high permeability values ​​to simulate the characteristics of easy stress transmission; and low gray intensity values ​​representing pores or cracks are mapped to permeability values ​​approaching zero to simulate stress blocking characteristics.

[0087] S23. Based on the direction of the gray-level gradient vector, construct a local rotation matrix to determine the principal axis direction of the local permeability tensor of the voxel point, so that the flow dominance direction of the virtual fluid is consistent with the material texture direction.

[0088] S24. Combine the local permeability tensors of all voxel points to generate a global permeability tensor field that can characterize the anisotropy of stress transmission within heterogeneous materials.

[0089] This embodiment elaborates on step S2, detailing how to construct the anisotropic permeability tensor field; during this period, feature extraction is performed in S21, traversing each voxel in the microstructure image dataset, which is a voxel in the three-dimensional dataset. The gray intensity value of the voxel can be obtained using the Sobel or Scharr operator. With gray gradient vector ; Execute the S22 mapping function construction to construct the anisotropic voxel permeability mapping function. ;

[0090] The function is configured to: establish a positive correlation between grayscale values ​​and virtual permeability, where high grayscale intensity values ​​representing high-density aggregates are mapped to high permeability values ​​to simulate the characteristic of easy stress transmission; and low grayscale intensity values ​​representing pores or cracks are mapped to minimal positive real permeability values ​​that are close to but not zero. For example, take as This serves as a numerical regularization constraint to prevent numerical overflow during the subsequent calculation of the resistance term matrix inversion and to simulate stress-blocking characteristics; the formula is... ,in The preset maximum virtual penetration rate is set within a certain range. Between, preferably ; This is a non-linear adjustment coefficient, and its value range is set within... Between, preferably , usually take To enhance the conductivity advantage of high-density areas, and These are the minimum and maximum gray values ​​in the image dataset, respectively.

[0091] S23 local tensor construction is performed based on grayscale gradient vectors. Construct a local rotation matrix in the direction of rotation. Determine the principal axis direction of the local permeability tensor at the voxel point; specifically, calculate the unit vector of the gradient direction. Construct rotation matrix Rotate the global coordinate system to the local principal axis coordinate system, so that one of the principal axes of the local coordinate system is aligned with... Parallel; define a diagonal matrix The component along the texture direction, i.e., in the plane perpendicular to the gradient direction, is set as... The component along the gradient direction is set as , Here are the anisotropy coefficients, with a range of values ​​of [value range missing]. Preferred ; through tensor transformation formula Calculate the local permeability tensor of this voxel point, where, voxel point The gray-level gradient vector, For local rotation matrices, The matrix is ​​diagonal, ensuring that the flow dominance direction of the virtual fluid aligns with the material texture direction; S24 full-field generation is performed, combining the local permeability tensors of all voxel points. Generate a global penetration tensor field;

[0092] This embodiment constructs anisotropic tensors by introducing gradient directions, achieving accurate simulation of texture guidance in the material analysis of concrete pole connectors. It not only considers the hardness of the material, i.e., grayscale, but also accurately simulates the guiding effect of aggregate arrangement direction on force transmission. This approach allows virtual fluid to flow along the direction of aggregate, more realistically restoring the transmission path of stress flow in heterogeneous concrete in the physical world, and significantly improving the physical realism of mechanical analysis.

[0093] Example 5:

[0094] S3 specifically includes:

[0095] S31. Load the anisotropic permeability tensor field data into the GPU's video memory to construct a virtual flow field calculation grid that corresponds one-to-one with the voxel grid.

[0096] S32. Set a constant high-pressure boundary condition at one end of the connection of the virtual flow field mesh and a constant low-pressure boundary condition at the other end of the connection to form a virtual pressure difference that drives the flow of virtual fluid.

[0097] S33. The collision step and migration step calculations of the lattice Boltzmann method are executed in parallel using the CUDA core. During the calculation, a matrix resistance term based on Darcy's law is introduced. The distribution function of fluid particles is corrected according to the local permeability tensor of each voxel point to simulate the obstruction effect of different material media on the flow.

[0098] S34. Perform iterative calculations until the relative error of the macroscopic fluid quantities across the entire field is less than the preset steady-state threshold. Extract the fluid velocity vector and flow density of each grid point to generate a visual stress flux distribution heat map.

[0099] This embodiment elaborates on step S3, detailing the GPU-parallel flow field solution process; S31, memory loading, loads the anisotropic permeability tensor field data into the GPU's memory, using TextureMemory to improve access speed, directly constructing a voxel mesh, which in the 3D dataset is a virtual flow field calculation mesh corresponding to each voxel mesh; S32, boundary setting, sets a constant high-pressure boundary condition at one end of the virtual flow field mesh, as shown on the surface above. A constant low-pressure boundary condition is set at the other end. This creates a virtual pressure difference that drives the flow of the virtual fluid. ;

[0100] Parallel computation of S33LBM is performed, utilizing the CUDA core to execute Lattice Boltzmann Method (LBM) computation in parallel; specifically, the LBGK model is adopted, and the evolution equation is as follows:

[0101]

[0102] in, for Time is located Position and along The distribution function of virtual fluid particles in directional motion. It is a discrete velocity vector. The equilibrium distribution function in the corresponding direction; relaxation time. Based on virtual fluid viscosity Confirmed, the formula is: This is used to control the diffusion rate of virtual fluid in micropores; where, The preset virtual computation time step is used; a matrix drag term based on Darcy's law is introduced as a source term during the computation process. This resistance term originates from the coupling between the local permeability tensor and the macroscopic flow velocity. Its physical meaning is the resistance force exerted by the porous medium on the fluid. The calculation formula is as follows:

[0103]

[0104] in For virtual hydrodynamic viscosity scalars. This is the inverse matrix of the permeability tensor of the voxel point after regularization. The preset minimum disturbance factor, The identity matrix is ​​used; this treatment ensures that in pore regions with extremely low permeability, the resistance term tends to a maximum value rather than infinity, thus achieving effective truncation of virtual fluid flow at the numerical calculation level. This represents the macroscopic velocity vector. During the collision step, the distribution function of fluid particles is corrected based on this resistance term to simulate the different hindering effects of different material media, such as high-density aggregate and loose mortar, on stress flow. S34 steady-state extraction is performed, and iterative calculations are executed until the relative error of the macroscopic fluid quantity (i.e., the average velocity) is less than a preset steady-state threshold. Extract the fluid velocity vector at each grid point Combined with flow density, a visual heatmap of stress flux distribution is generated;

[0105] This embodiment successfully transforms the heterogeneity of concrete microstructure into porous media resistance in fluid mechanics by introducing a Darcy drag term coupled to the local permeability tensor into the LBM; this innovation enables large-scale parallel computing on GPUs, for example... The large-scale three-dimensional voxel mesh for pole connections can be solved in minutes; this efficient computing power allows engineers to iterate quickly during the design phase, visually see how stress flows between complex concrete aggregates, and thus accurately locate weak points in the connections.

[0106] Example 6:

[0107] S4 specifically includes:

[0108] S41. Threshold segmentation is performed on the velocity scalar value of the stress flux distribution heatmap, and a high flux threshold and a low flux threshold are set, wherein the high flux threshold is strictly greater than the low flux threshold.

[0109] S42. Identify voxel connected regions with flow rates greater than or equal to the high throughput threshold, mark them as stress concentration main channels, and extract the centerline of the channel as the skeleton topology chain.

[0110] S43. Identify voxel regions with flow rates less than the low throughput threshold and mark them as invalid material regions;

[0111] S44. Calculate the local vorticity value at each point in the flow field, detect areas where the vorticity value exceeds the preset turbulence threshold, and mark them as potential fatigue fracture risk points.

[0112] S45. Execute parameter reconstruction logic: Increase reinforcement density in the coordinate region where the skeleton topology chain is located; perform material reduction operation in the coordinate region corresponding to the invalid material region to reduce the thickness of the connector; perform chamfer curvature smoothing processing on the potential fatigue fracture risk point to generate optimized geometric reinforcement parameters.

[0113] This embodiment elaborates on the topology inversion optimization module in step S4, detailing how to transform flow field data into design parameters; and performs threshold segmentation in S41 to process the velocity scalar values ​​of the stress flux distribution heatmap. Perform analysis and set a high-throughput threshold. With low throughput threshold ,For example 70% of the maximum flow rate The maximum flow rate is 10%; S42 main channel extraction is performed to identify the flow rate value. The voxel connected regions are marked as the main channels of stress concentration; the centerline of the channel is extracted as the skeleton topology chain using the skeletonization algorithm.

[0114] Simultaneously, S43 invalid region identification is performed to identify flow rate values. The voxel regions are marked as invalid material areas; the focus is on S44 fatigue risk point detection, and the local vorticity value at each point in the flow field is calculated. Among them, operators The solution is discretized using a spatial center-difference scheme based on the voxel grid. Local rotational features of the flow field are extracted by calculating the velocity vector difference between adjacent voxels. Regions where the vorticity value exceeds a preset turbulence threshold are detected. For the coordinate region where the skeleton topological chain is located, the voxel points of the topological chain are used as the center. Construct an expansion region for the radius, and increase the diameter within this expansion region by a Boolean union operation. The reinforcement steel unit increases the reinforcement ratio parameter. ,like For the coordinate region corresponding to the invalid material area, identify the boundary contour of this region in the CAD model, and remove redundant material voxels within the contour through Boolean subtraction operations to achieve material reduction operations, such as opening holes or reducing weight to reduce the thickness of the connector; for potential fatigue fracture risk points, perform chamfer curvature smoothing processing to reduce the local fillet radius. Increase to the original value This is multiplied by 10 to eliminate stress singularities, thereby generating optimized geometric reinforcement parameters;

[0115] This embodiment innovatively introduces vorticity to predict fatigue risk, which has unique value in the design of concrete pole connectors. In traditional static FEA, it is difficult to directly find fatigue points, but in the fluid mapping of this system, the high stress change zone is manifested as high vorticity turbulence. This intuitive mapping makes the identification of potential fracture points exceptionally sensitive. By chamfering and smoothing these high vorticity regions, the probability of fatigue failure of the pole under long-term wind load can be significantly reduced.

[0116] Example 7:

[0117] S5 specifically includes:

[0118] S51. Obtain the optimized geometric reinforcement parameters and update the material distribution definition in the calculation model constructed based on the microstructure image dataset;

[0119] S52. Calculate the updated total energy dissipation rate of the flow field. If the change in the total energy dissipation rate is less than the preset convergence threshold, the optimization is considered complete.

[0120] S53. If the change in total energy dissipation rate is greater than or equal to the preset convergence threshold, then return to execute steps S2 to S4 for the next iteration.

[0121] S54. After the optimization is completed, convert the final geometric reinforcement parameters into CAD engineering drawing format and output the final optimized design drawings.

[0122] This embodiment specifies the design verification closed-loop module in step S5, detailing the logic of convergence judgment; executes model update in S51, obtains the optimized geometric reinforcement parameters, and updates the computational model constructed based on the microstructure image dataset; the specific operation is as follows: the optimized reinforcement area is mapped to an extremely high permeability voxel, and the area with reduced material is mapped to the aforementioned regularized minimum permeability voxel, thereby completing the dynamic update from geometric topology change to physical parameter field;

[0123] Perform S52 energy dissipation calculation to calculate the updated total energy dissipation rate of the flow field. The formula for calculating this indicator is: ,in, For pressure gradient, For flow rate, The computational domain is the entire field, which physically represents the volume occupied by all effective voxels. Physically, a lower energy dissipation rate indicates higher stress transfer efficiency and a smoother structure. The S53 convergence check is performed in response to the current iteration. Compared to the previous round Change value Less than a preset convergence threshold, such as 0.1%, where, Indicates the current iteration round number. For the first If the total energy dissipation rate is calculated in each iteration, the optimization is considered complete; otherwise, return to steps S2 to S4; execute step S54 to output the drawings. After the optimization is considered complete, use the DXF / DWG conversion interface to convert the final geometric reinforcement model into CAD engineering drawing format and output the final optimized design drawings.

[0124] This embodiment uses the total energy dissipation rate as the objective function, enabling the system to automatically find a structural form that conforms to the principle of minimum work. In the design of concrete pole connections, this means that the optimized connectors not only have high strength, but also minimize internal friction and loss during stress transmission. This design orientation ensures that the internal damage accumulation caused by stress concentration in the connectors is minimized during long-term service, thereby effectively extending the service life of the pole.

[0125] Example 8:

[0126] When performing calculations, the parallel flow field solving module uses double-precision floating-point format to store the distribution function of fluid particles and utilizes the shared memory blocks of the GPU to exchange and synchronize local grid data. For image data with resolution exceeding the preset video memory limit, a multi-GPU spatial decomposition strategy is adopted to divide the anisotropic permeability tensor field into multiple sub-regions, which are then computed in parallel on different GPU nodes. The calculation results are then merged through a boundary exchange protocol.

[0127] This embodiment details the hardware optimization strategy for the parallel flow field solution module, aiming to overcome the bottleneck of single-machine GPU memory. To address numerical accuracy issues, the distribution function of fluid particles is stored in double-precision floating-point FP64 format to prevent the accumulation of truncation errors over tens of thousands of iterations. During this process, local meshing is performed using the GPU's shared memory blocks, such as... Data exchange and synchronization within blocks reduce latency in accessing global video memory; for industrial-grade ultra-high resolution, such as... For voxel-scale data where a single GPU's video memory is insufficient, the system employs a multi-GPU spatial decomposition strategy. This strategy divides the anisotropic permeability tensor field into multiple sub-domains and assigns each sub-domain to different GPU nodes for parallel computation. At the end of each iteration step, the system uses the HaloExchange boundary exchange protocol to exchange one or more layers of mesh data at the edges of adjacent sub-domains via NVLink or PCIe bus, and merges the computation results.

[0128] The hardware acceleration strategy in this embodiment breaks through the computing power ceiling of large-scale microscopic analysis, making it possible to perform micron-level panoramic scanning data analysis on full-size concrete pole connection sections. Through a multi-GPU strategy, the system achieves a leap from sampling analysis to holographic analysis, ensuring that the timeliness required for engineering design can still be maintained when processing massive amounts of microscopic data, and providing strong computing power support for the digital design of large-scale infrastructure.

[0129] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A mechanical optimization analysis and design system for concrete utility pole connections, characterized in that, include: Microscopic image acquisition module: configured to acquire industrial CT slice image sequences of the connection parts of concrete poles and perform data cleaning to construct a microstructure image dataset; Heterogeneous space mapping module: configured to extract the grayscale features and texture gradient features of voxels in the microstructure image dataset, and construct an anisotropic permeability tensor field through a voxel grayscale to permeability mapping algorithm, thereby mapping the mechanical transmission path of the physical material into a digitally defined virtual fluid permeation field; Parallel flow field solution module: configured as a GPU-parallel lattice Boltzmann method, it simulates the flow evolution of virtual fluid driven by pressure difference in the virtual fluid infiltration field and calculates the stress flux distribution thermogram characterizing the stress transmission intensity. The topology inversion optimization module is configured to identify efficient force transmission regions and inefficient redundant regions in the stress flux distribution heatmap, and reconstruct the geometric reinforcement parameters and material distribution of the connectors based on the identification results; perform threshold segmentation on the velocity scalar values ​​of the stress flux distribution heatmap, setting high-flux thresholds and low-flux thresholds, wherein the high-flux threshold is strictly greater than the low-flux threshold; identify voxel connected regions with velocity values ​​greater than or equal to the high-flux threshold, mark them as main stress concentration channels, and extract the centerline of the channel as the skeleton topology chain; identify voxel regions with velocity values ​​less than the low-flux threshold, and mark them as invalid material regions; calculate the local vorticity value at each point in the flow field, detect regions with vorticity values ​​exceeding a preset turbulence threshold, and mark them as potential fatigue fracture risk points; execute parameter reconstruction logic: increase reinforcement density for the coordinate region where the skeleton topology chain is located; perform material reduction operation for the coordinate region corresponding to the invalid material region to reduce the thickness of the connector; perform chamfer curvature smoothing processing for the potential fatigue fracture risk points to generate optimized geometric reinforcement parameters; Design verification closed-loop module: configured to feed back the reconstructed geometric reinforcement parameters to the model definition corresponding to the microstructure image dataset, trigger a new round of simulation calculations until the preset convergence conditions are met, and output the final optimized design drawings.

2. A method for optimizing the mechanical analysis and design of concrete pole connections, implemented using the mechanical optimization analysis and design system for concrete pole connections as described in claim 1, characterized in that... Includes the following steps: S1. Obtain an industrial CT slice image sequence of the connection part of the concrete pole through the microscopic image acquisition module, perform noise reduction and binarization preprocessing on the image sequence, and construct a microscopic structure image dataset. S2. Input the microstructure image dataset into the heterogeneous space mapping module, extract the gray value and gradient direction of each voxel point, and generate an anisotropic permeability tensor field that is mapped point-to-point with the image resolution through the anisotropic voxel permeability mapping function. S3. Input the anisotropic permeability tensor field into the parallel flow field solving module, set virtual high-pressure boundary and virtual low-pressure boundary, use the lattice Boltzmann algorithm to calculate the flow state of the virtual fluid in the voxel grid in parallel, and output the stress flux distribution heat map containing the velocity scalar and the flow direction vector. S4. Input the stress flux distribution heat map into the topology inversion optimization module, extract the high-velocity connected region as the main stress transmission channel, extract the low-velocity region as the material redundancy region, and adjust the topology parameters of the connector accordingly. S5. Calculate the adjusted flow field energy dissipation index through the design verification closed-loop module, and determine whether the index meets the preset convergence condition. If it does, output the final optimized design drawing.

3. The method for mechanical optimization analysis and design of concrete pole connections according to claim 2, characterized in that, S1 specifically includes: S11. Use high-precision industrial CT scanning equipment to perform tomographic scanning on the connection interface of concrete poles to obtain raw DICOM format data; S12. Perform artifact removal and noise reduction on the original DICOM format data to generate a high-resolution grayscale slice image; S13. Perform 3D reconstruction and voxel alignment on the high-resolution grayscale slice image, remove background noise, and extract the region of interest containing aggregate, mortar and pore features. S14. Serialize and store the image data of the region of interest according to spatial coordinates to construct a standardized microstructure image dataset.

4. The method for mechanical optimization analysis and design of concrete pole connections according to claim 3, characterized in that, S2 specifically includes: S21. Traverse each voxel in the microstructure image dataset and obtain the gray intensity value and gray gradient vector of that voxel. S22. Construct an anisotropic voxel permeability mapping function, which is configured as follows: establish a positive correlation mapping between gray values ​​and virtual permeability, wherein high gray intensity values ​​representing high-density aggregates are mapped to high permeability values ​​to simulate the characteristics of easy stress transmission; and low gray intensity values ​​representing pores or cracks are mapped to permeability values ​​approaching zero to simulate stress blocking characteristics. S23. Based on the direction of the gray-level gradient vector, construct a local rotation matrix to determine the principal axis direction of the local permeability tensor of the voxel point, so that the flow dominance direction of the virtual fluid is consistent with the material texture direction. S24. Combine the local permeability tensors of all voxel points to generate a global permeability tensor field that can characterize the anisotropy of stress transmission within heterogeneous materials.

5. The method for mechanical optimization analysis and design of concrete pole connections according to claim 4, characterized in that, S3 specifically includes: S31. Load the anisotropic permeability tensor field data into the GPU's video memory to construct a voxel grid, i.e., the corresponding virtual flow field calculation grid. S32. Set a constant high-pressure boundary condition at one end of the connection of the virtual flow field mesh and a constant low-pressure boundary condition at the other end of the connection to form a virtual pressure difference that drives the flow of virtual fluid. S33. The collision step and migration step calculations of the lattice Boltzmann method are executed in parallel using the CUDA core. During the calculation, a matrix resistance term based on Darcy's law is introduced. The distribution function of fluid particles is corrected according to the local permeability tensor of each voxel point to simulate the obstruction effect of different material media on the flow. S34. Perform iterative calculations until the relative error of the macroscopic fluid quantities across the entire field is less than the preset steady-state threshold. Extract the fluid velocity vector and flow density of each grid point to generate a visual stress flux distribution heat map.

6. The method for mechanical optimization analysis and design of concrete pole connections according to claim 5, characterized in that, S5 specifically includes: S51. Obtain the optimized geometric reinforcement parameters and update the material distribution definition in the calculation model constructed based on the microstructure image dataset; S52. Calculate the updated total energy dissipation rate of the flow field. If the change in the total energy dissipation rate is less than the preset convergence threshold, the optimization is considered complete. S53. If the change in total energy dissipation rate is greater than or equal to the preset convergence threshold, then return to execute steps S2 to S4 for the next iteration. S54. After the optimization is completed, convert the final geometric reinforcement parameters into CAD engineering drawing format and output the final optimized design drawings.

7. The method for mechanical optimization analysis and design of concrete pole connections according to claim 5, characterized in that, When the parallel flow field solving module performs calculations: The distribution function of fluid particles is stored in double-precision floating-point format, and the shared memory block of the GPU is used for the exchange and synchronization of local mesh data. For image data whose resolution exceeds the preset video memory limit, a multi-GPU spatial decomposition strategy is adopted to divide the anisotropic permeability tensor field into multiple sub-regions, which are then computed in parallel on different GPU nodes, and the computation results are merged through a boundary exchange protocol.

Images