A new buffer layer material evaluation method for solving the buffer layer ablation problem

Abstract

The application discloses a novel buffer layer material evaluation method for solving the buffer layer ablation problem, and belongs to the electrical engineering field.The method adopts CT scanning to obtain two-dimensional tomographic slices of the buffer layer material, and carries out AI artifact removal, filtering and three-value processing, and reconstructs a micro model reflecting real pores and fiber structures through a volume data method; the micro model is discretized into a finite element grid and optimized, and is embedded into a cable macro model to construct a macro-micro coupled three-dimensional finite element model; local grid refinement, parallel computing, model order reduction or reasonable selection of a solving method are adopted to accelerate simulation, and the electric field and current distribution are calculated under an electric-thermal-force multi-physical field coupling environment; the ablation resistance of the novel buffer material is compared with that of a polyester non-woven fiber water-blocking tape as a benchmark, and whether the ablation resistance of the novel material meets the standard is judged.The application realizes quantitative evaluation from a real micro structure to system performance, and has the advantages of high precision, good efficiency and strong engineering applicability.

Description

Technical Field

[0001] This invention belongs to the field of electrical engineering technology, and specifically relates to a novel method for evaluating buffer layer materials to address the problem of buffer layer ablation. Background Technology

[0002] With the increasing load on urban power grids and the ever-growing demands for power supply reliability, power cables are gradually replacing traditional overhead transmission lines due to their advantages such as high safety, space saving, and environmental aesthetics. However, operational experience shows that in cross-linked polyethylene (XLPE) insulated cables with voltage levels of 110kV and above, there is a risk of cable breakdown due to buffer layer erosion. Dissection analysis reveals that such faults are often accompanied by corrosion of the inner surface of the aluminum sheath, the appearance of white powdery residue on the surface of the buffer layer, and the formation of a discharge channel between the buffer layer and the insulating shielding layer. In severe cases, this can even lead to ablation holes in the main insulation layer.

[0003] To address the aforementioned issues, existing technologies primarily focus on two aspects: firstly, structural improvements, such as replacing the wrinkled aluminum sheath with a smooth one, to enhance the uniformity of contact between the buffer layer and the aluminum sheath, thereby reducing the risk of partial discharge and current-induced heating; secondly, material optimization, which enhances the performance of the buffer layer through the composite use of multiple materials. Currently, the buffer layer materials used in engineering mainly include polyester nonwoven fiber cloth, polyester nonwoven fiber water-blocking tape, copper wire fiber woven cloth, and semi-conductive butyl rubber tape, and various combination schemes have been derived.

[0004] However, the aforementioned existing technologies still have significant shortcomings: First, there are limitations to structural improvements. Specifically, while smooth aluminum sheaths can improve contact, their low mechanical strength and poor bending performance make them unsuitable for the mechanical requirements of large-scale power projects, especially those involving crossing distances. Therefore, corrugated aluminum sheathed cables are still widely used, and the problem of buffer layer ablation has not been fundamentally solved. Second, there is a lack of material evaluation methods. Specifically, although there are various combinations of buffer layer materials, there is currently a lack of systematic, efficient, and effective evaluation methods that truly reflect the microstructure of materials and their performance in actual engineering environments. In engineering practice, it is difficult to scientifically select the best material from numerous combinations, leading to reliance on experience in material selection, inaccurate performance predictions, and an inability to effectively prevent ablation failures. Summary of the Invention

[0005] The object of the present invention is to provide a new evaluation method for buffer layer materials to address the problem of buffer layer ablation, so as to solve the problem in the prior art that it is difficult to scientifically select the best from numerous buffer layer material combinations due to the lack of a systematic and efficient evaluation means. The present invention aims to establish a multi-scale simulation model that can truly reflect the microscopic pores and fiber structure of buffer layer materials by integrating CT scanning, AI image processing, three-dimensional reconstruction, and multi-physics field coupling simulation technologies. On this basis, to evaluate the ablation resistance performance of new buffer layer materials, taking the ablation resistance performance of polyester non-woven fiber water-blocking tapes as a benchmark, by comparing ablation thresholds such as electric field strength and current density, the ablation resistance performance of the material is determined according to whether the threshold parameters are exceeded. If the threshold is exceeded, it proves that its performance is qualified; if the threshold is not exceeded, it proves that the ablation resistance performance of the material is unqualified, thereby realizing the systematic and objective evaluation and optimization of the ablation resistance performance of buffer layer materials.

[0006] In a first aspect, an embodiment of the present invention provides a new evaluation method for buffer layer materials to address the problem of buffer layer ablation, including:

[0007] Using a CT scanning device to scan a buffer layer material sample to be evaluated to obtain multiple two-dimensional slice images;

[0008] Successively perform artifact removal based on AI and median filtering denoising on the two-dimensional slice images, and use the bottom-hat transformation method to trinary the images to distinguish the material matrix, pores, and transition regions; use the volume data method to register and resample the processed slice sequence to generate a three-dimensional volume data field composed of voxel values; through transfer function mapping and direct volume rendering based on the voxel values, reconstruct a three-dimensional microscopic structure model of the buffer layer that can truly reflect the internal pore distribution and fiber structure of the material;

[0009] Discretize the three-dimensional microscopic structure model of the buffer layer into a finite element mesh model, and optimize the mesh quality of the finite element mesh model; construct a cable macroscopic model, embed and replace the original buffer layer macroscopic structure in the cable macroscopic model with the optimized mesh model, and ensure close contact with the aluminum sheath and insulation shielding layer in the cable macroscopic model to construct a cable macro-microscopic three-dimensional finite element model that integrates the real microscopic structure of the buffer layer;

[0010] For the cable macro-microscopic coupled three-dimensional simulation model, perform optimization processing using at least one of local mesh refinement, multi-core parallel computing, proper orthogonal decomposition-based model reduction method, or reasonable selection of solution methods; the reasonable selection of solution methods is used for: analyzing the mathematical matrix properties of the cable macro-microscopic coupled three-dimensional simulation model, and selecting a suitable solution algorithm according to the mathematical matrix properties, and configuring a preconditioner and convergence error for the selected algorithm;

[0011] A simulation environment with electro-thermal-mechanical multiphysics coupling was established on the optimized model to calculate and obtain the electric field intensity and current density distribution in the buffer layer region. Taking the ablation resistance of polyester nonwoven fiber water-blocking tape as a reference benchmark, the maximum electric field intensity and the maximum current density were extracted from the electric field intensity and current density distribution. It was then determined whether the two maximum values ​​exceeded the preset ablation threshold. If neither exceeded the threshold, the buffer layer material was deemed qualified; otherwise, an ablation risk was identified.

[0012] Optionally, the three-dimensional microstructure model of the buffer layer is discretized into a finite element mesh model, including:

[0013] Based on the volume data constituting the three-dimensional microstructure model of the buffer layer, the voxels in the volume data are mapped into a set of hexahedral finite element elements according to their spatial arrangement relationship, thereby generating an initial finite element mesh.

[0014] Determine the grayscale value range corresponding to different material categories in the volume data; for each mesh cell in the initial finite element mesh, determine the material category of the mesh cell according to the grayscale value range to which the grayscale value of the mesh cell's location belongs, and assign corresponding material properties to the mesh cell according to the determined material category;

[0015] Export the mesh elements and node information that have been assigned material properties to generate a mesh file that can be recognized by the finite element model, and import it into the finite element model to construct a three-dimensional finite element mesh model.

[0016] Optionally, the finite element mesh model undergoes mesh quality optimization, including:

[0017] The finite element mesh model was evaluated to identify mesh elements with quality below a preset standard.

[0018] The identified mesh cells are repaired using a mesh smoothing method to improve their cell shape; wherein the mesh smoothing method includes at least one of Laplacian smoothing, angle-weighted smoothing, or uniform elastic smoothing.

[0019] The repaired mesh model was re-evaluated to ensure it met the quality requirements of simulation analysis.

[0020] Optionally, constructing the macroscopic model of the cable includes:

[0021] A macroscopic model of the cable is constructed based on the high-voltage cable structure parameter table. The macroscopic model of the cable includes at least the structure of the conductor, conductor shielding layer, XLPE insulation layer, insulation shielding layer, aluminum sheath, and outer sheath. The geometric dimensions and material properties of each layer are set according to the parameter table.

[0022] Optionally, the macro-micro coupled 3D simulation model of the cable is optimized using local mesh refinement, including:

[0023] Identify the key areas that need to be analyzed in the macro-micro coupled three-dimensional simulation model of the cable. The key areas include at least the optimized mesh model as a buffer layer component and its contact area with the aluminum sheath and the insulating shielding layer.

[0024] The overall mesh size is limited by controlling the mesh size of the micro-buffer layer model and the macro-cable model respectively; the mesh is locally refined in the boundary area between the micro-buffer layer model and the macro-cable model; and the gradual change of mesh size from fine to coarse is controlled by setting the spatial distribution of finite element elements and nodes at the boundary, so as to achieve a continuous transition between meshes of different scales and to achieve a smooth transition between coarse and fine meshes.

[0025] A fine mesh is set within the critical area, and a coarse mesh is set in the rest of the model outside the critical area. A mesh transition zone is set at the boundary between the fine mesh area and the coarse mesh area; wherein the mesh size in the mesh transition zone gradually changes from the fine mesh to the coarse mesh.

[0026] The mesh quality of the cable macro-micro coupled three-dimensional simulation model after local mesh refinement and transition matching processing is evaluated and verified to ensure that its mesh quality meets the requirements of simulation analysis and to obtain a mesh model that can be used for calculation.

[0027] Optionally, multi-core parallel computing is used to optimize the macro-micro coupled three-dimensional simulation model of the cable, including:

[0028] Parallel computing is enabled in the finite element model to decompose the computational domain of the macro-micro coupled three-dimensional simulation model of the cable into multiple subdomains;

[0029] Configure a parallel solution method to distribute the computational tasks of each subdomain to multiple processor cores for simultaneous computation;

[0030] The computation time and speedup were tested with different numbers of processor cores, and the number of processor cores used for the final simulation was determined based on the number of cores with the highest computational efficiency in the test results.

[0031] Optionally, the macro-micro coupled three-dimensional simulation model of the cable is optimized using a model order reduction method based on intrinsic orthogonal decomposition, including:

[0032] The macro-micro coupled three-dimensional simulation model of the cable was sampled and calculated multiple times under the electro-thermal-mechanical multi-physics coupling environment. By changing the key parameters, multiple sets of different solution vectors and their corresponding parameters were collected to form a training sample set.

[0033] The solution vectors in the training sample set are preprocessed by centering and normalizing.

[0034] The dimensionality of the preprocessed solution vector set is reduced by using the eigenorthogonal decomposition method to construct a reduced-order basis.

[0035] Projecting the control equations of the macro-micro coupled three-dimensional simulation model of the cable onto the reduced-order basis, a reduced-order model that can capture the main features of the system and has fewer degrees of freedom than the macro-micro coupled three-dimensional simulation model of the cable is obtained.

[0036] The reduced-order model is trained using the training sample set, and the trained reduced-order model is verified using the test set parameters. The verification results are compared with the calculation results of the cable macro-micro coupling three-dimensional simulation model to verify the accuracy of the reduced-order model.

[0037] Optionally, the control equations of the macro-micro coupled three-dimensional simulation model of the cable are numerically solved using a solution method to obtain model parameter solutions that satisfy the convergence conditions, specifically including:

[0038] Analyze the properties of the mathematical matrices generated by the macro-micro coupled three-dimensional simulation model of the cable in the electro-thermal-mechanical multiphysics coupling simulation;

[0039] Based on the properties of the mathematical matrix, a suitable solution method is selected, and a preconditioner is configured for the solution method to accelerate the convergence of the iteration.

[0040] Set the convergence error for the selected solution method.

[0041] Optionally, the step of using a CT scanning device to scan the buffer layer material sample to be evaluated to obtain multiple two-dimensional slice images includes:

[0042] Place the buffer layer material sample in the scanning area of ​​the CT scanning device;

[0043] The X-ray source and detector are controlled to perform multi-angle projection scanning on the sample, and the detector receives the X-ray signal after penetrating the sample.

[0044] The X-ray signal is photoelectrically converted and amplified to obtain projection data at each projection angle.

[0045] The projection data is reconstructed and calculated, and then converted into a digital image reflecting the material attenuation coefficient at each point inside the sample, thereby obtaining multiple grayscale two-dimensional slice images.

[0046] Secondly, embodiments of the present invention also provide an electronic device, comprising:

[0047] At least one processor;

[0048] Memory for storing the at least one processor-executable instruction;

[0049] The at least one processor is configured to execute the instructions to implement the method described in the first aspect.

[0050] The technical solution provided by this invention firstly digitizes the real microscopic morphology based on CT scanning and AI image processing technology, enabling the evaluation to break away from macroscopic experience assumptions and ensuring the model's authenticity from the source. Secondly, by embedding the microscopic model into the macroscopic system and establishing close contact, a coupled simulation model that can realistically replicate the flow distribution in the field is constructed. Furthermore, to evaluate the ablation resistance of the new buffer layer material, the ablation resistance of polyester nonwoven fiber water-blocking tape is used as a benchmark. By comparing ablation thresholds such as electric field strength and current density, the ablation resistance of the material is determined based on whether the threshold parameters are exceeded. If the threshold is exceeded, the material's performance is deemed qualified; if the threshold is not exceeded, the material's ablation resistance is deemed unqualified. This constructs a quantifiable and comparable selection system, transforming the selection decision from subjective experience to objective data support. Simultaneously, by comprehensively utilizing acceleration strategies such as local mesh refinement, parallel computing, and model order reduction, the engineering bottleneck of low computational efficiency in high-fidelity models is effectively solved, giving this method good engineering applicability.

[0051] In summary, the method provided by this invention enables a full-chain, quantitative evaluation of buffer layer materials, from microstructure characterization to macroscopic system performance prediction, providing a powerful technical tool for the scientific selection and reliability design of high-voltage cable buffer layers. Attached Figure Description

[0052] Figure 1 A flowchart illustrating the overall technical solution provided in the embodiments of the present invention;

[0053] Figure 2 A flowchart illustrating a novel buffer layer material evaluation method for addressing buffer layer ablation problems, provided as an embodiment of the present invention;

[0054] Figure 3 A schematic diagram of a CT scanning device provided in an embodiment of the present invention;

[0055] Figure 4(a) is a schematic diagram of the three-dimensional microstructure of the polyester nonwoven fiber water-blocking tape sample; Figure 4(b) is a schematic diagram of the finite element mesh model of the polyester nonwoven fiber water-blocking tape sample.

[0056] Figure 5(a) is a schematic diagram of the structure of the macro-micro three-dimensional finite element model of the cable; Figure 5(b) is a schematic diagram of the mesh of the macro-micro three-dimensional finite element model of the cable. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are merely some embodiments of this invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0058] There are currently no unified standards for the selection of buffer layer structures, and the ablation resistance of different buffer layer materials varies. The following four types of buffer layer materials are used in domestic cables: polyester nonwoven fiber cloth, polyester nonwoven fiber water-blocking tape, copper wire fiber braided cloth, and semi-conductive butyl rubber tape. Among them, polyester nonwoven fiber water-blocking tape is widely used in China due to its longitudinal water-blocking function; copper wire fiber braided cloth is not used as a standalone buffer layer material due to its constituent materials and functions.

[0059] To address the problem of buffer layer erosion, various material combinations for the buffer layer have been developed, such as: double-layer polyester nonwoven fabric, double-layer polyester nonwoven water-blocking tape, composites of copper wire woven fabric and polyester nonwoven fabric, composites of copper wire woven fabric and polyester nonwoven water-blocking tape, composites of copper wire woven fabric and semi-conductive butyl rubber tape, and composites of copper wire woven fabric, polyester nonwoven water-blocking tape, and semi-conductive butyl rubber tape. Given these various combinations, a systematic evaluation to determine which combination offers the best overall performance is a critical issue that urgently needs to be addressed in current engineering practice. To clearly illustrate the different combinations, the composition of the buffer layer is summarized in Table 1.

[0060] Table 1. Buffer layer composition structure of different schemes

[0061]

[0062] Based on the aforementioned issue of selecting the best buffer layer material, this invention proposes a novel buffer layer material evaluation method to address the problem of buffer layer ablation. Specifically, this invention provides a buffer layer material evaluation method based on CT scanning, AI image processing, three-dimensional reconstruction, and multiphysics field coupling simulation to achieve a systematic evaluation of the performance of different buffer layer structures.

[0063] To ensure the plan is clearly described, the following will combine... Figure 1 The overall technical solution of the embodiments of the present invention will be described in detail, such as... Figure 1 As shown, it may include the following steps:

[0064] The first step is to select different types of buffer layer materials.

[0065] Based on engineering requirements, a sample of buffer layer material to be evaluated is selected from a variety of candidate materials or material combinations (e.g., double-layer polyester nonwoven fabric, copper wire fiber woven fabric and semi-conductive butyl rubber tape composite, etc.).

[0066] The second step is to construct a three-dimensional micro-buffer layer model.

[0067] This step aims to obtain a digital microstructure model of the selected material, which is achieved through the following sub-steps:

[0068] The first sub-step is CT scanning imaging: the selected buffer layer material sample is scanned using a CT scanning device to obtain a series of two-dimensional tomographic images of its internal structure.

[0069] The second sub-step is CT slice image processing: using artificial intelligence (AI) based technology to perform preprocessing such as artifact removal and noise filtering on the acquired two-dimensional slice images.

[0070] The third sub-step involves using volumetric data methods to register, resample, and directly render the processed slice sequence, reconstructing a three-dimensional microscopic buffer layer model that can realistically reflect the internal pores and fiber structure of the material.

[0071] The third step is mesh construction and model embedding. This includes the following two sub-steps:

[0072] The first sub-step is mesh construction. The three-dimensional micro-buffer layer model obtained in the second step is transformed into a high-quality finite element mesh model by generating a mesh using voxels as units, assigning material properties based on grayscale values, and performing mesh repair and quality optimization.

[0073] The second sub-step is model embedding: the high-quality finite element mesh model mentioned above is embedded as a component into the macroscopic finite element model of the cable constructed based on the standard parameters of the cable, replacing the original macroscopic structure of the buffer layer, and ensuring close contact with the aluminum sheath and the insulation shielding layer, ultimately forming a macro-micro three-dimensional finite element model of the cable that takes into account the pore structure of the buffer layer.

[0074] The fourth step is model processing to improve computational efficiency.

[0075] Specifically, the macro-micro 3D finite element model of the cable constructed in the third step undergoes accelerated preprocessing. Based on computational resources and analysis requirements, one or more methods are employed in combination, such as local mesh refinement, multi-core parallel computing, training a reduced-order model, and appropriately selecting solution methods, to significantly reduce the computational complexity and time cost of model solving.

[0076] The fifth step is to conduct multi-physics field coupling simulation calculations involving electricity, heat, and force.

[0077] Based on the model optimized in the fourth step, a simulation environment coupling electric field, temperature field and stress field is established, actual operating conditions are set, and calculations are performed to obtain detailed electric field intensity and current density distribution results in the buffer layer region.

[0078] Step 6: Performance evaluation and judgment.

[0079] Specifically, the electric field strength and current density distribution obtained from the simulation calculation in the fifth step are compared and analyzed with the ablation threshold set in advance based on the material properties.

[0080] If neither the electric field strength nor the current density exceeds the threshold, the buffer layer material is deemed to meet the requirements and possess good ablation resistance. If either physical quantity exceeds the threshold, the buffer layer material is deemed to fail to meet the requirements and is at risk of ablation.

[0081] This invention is achieved through Figure 1 The complete technical process shown enables a systematic and high-precision evaluation of the ablation resistance of buffer layer materials. Its beneficial effects are specifically reflected in the following aspects: First, based on CT scanning and AI image processing technology, the true microscopic morphology is digitized, allowing the evaluation to move away from macroscopic empirical assumptions and ensuring the model's authenticity from the source. Second, by embedding the microscopic model into the macroscopic system and establishing close contact, a coupled simulation model that can realistically replicate the on-site flow distribution is constructed. Furthermore, to evaluate the ablation resistance of the new buffer layer material, the ablation resistance of polyester nonwoven fiber water-blocking tape is used as a benchmark. By comparing ablation thresholds such as electric field strength and current density, the evaluation is based on whether the threshold parameters are exceeded. The ablation resistance of materials is determined by numerical analysis. If the resistance exceeds a threshold, the material is deemed to be qualified; if it does not exceed the threshold, the material is deemed to be unqualified. This constructs a quantifiable and comparable selection system, transforming the selection decision from subjective experience to objective data support. Simultaneously, by comprehensively utilizing acceleration strategies such as local mesh refinement, parallel computing, and model order reduction, the engineering bottleneck of low computational efficiency in high-fidelity models is effectively solved, giving the method good engineering applicability. Ultimately, a complete technical solution is formed, encompassing material selection, microscopic characterization, multi-scale modeling, efficient simulation, and quantitative evaluation, providing a reliable means for the scientific selection and reliability improvement of high-voltage cable buffer layers.

[0082] After a detailed description of the complete technical process of the present invention, the following is in conjunction with the appendix. Figure 2 This invention provides a detailed description of its embodiments. This embodiment offers a novel method for evaluating buffer layer materials to address the problem of buffer layer ablation. This method achieves an objective and quantitative evaluation of the ablation resistance performance of buffer layer materials by constructing a high-fidelity multi-scale simulation model. Specifically, it includes the following steps:

[0083] S210 uses a CT scanning device to scan the buffer layer material sample to be evaluated, obtaining multiple two-dimensional slice images.

[0084] In this step, an industrial CT scanning device is used to perform non-destructive testing on the buffer layer material sample to be evaluated (e.g., polyester nonwoven water-blocking tape). The specific operation is as follows: The sample is fixed on the scanning platform of the CT device, and an X-ray beam emitted from an X-ray source penetrates the sample, while a detector receives the attenuated signal. By rotating the sample or the X-ray source-detector system at multiple angles (typically 360°), a series of projection data at different projection angles are acquired. Subsequently, these projection data are processed using reconstruction algorithms such as filtered backprojection to ultimately obtain a series of two-dimensional tomographic grayscale slice images reflecting the internal structure of the sample.

[0085] It should be noted that while computed tomography (CT) technology has been widely used in research in medicine, geology, and other disciplines, its application in the electrical field is currently limited. CT scanning technology can provide a non-destructive and cost-effective method for constructing microstructures. It utilizes Bell's law based on X-ray intensity attenuation and Radon's formula for establishing grayscale images through projection, and uses X-rays to perform tomographic scanning of the model to generate three-dimensional images of people or objects, such as... Figure 3 The image shows a schematic diagram of a CT scanning device.

[0086] from Figure 3 It can be seen that CT scanning equipment mainly consists of an X-ray source, a moving platform, a detector, and an image processing system. Figure 3 In this diagram, 1 represents the core working components of the CT scanning device: 1 is the X-ray generator, 2 is the moving platform of the CT scanning device, 3 is the test sample, and 4 is the X-ray detector. Before preparation, the test sample is placed on the moving platform of the CT scanning device, which has the function of translating along the X, Y, and Z axes and rotating along the Z axis. During operation, the X-ray generator emits X-rays onto the test sample placed on the moving platform. When the X-rays pass through objects of different densities, due to optical properties, light refraction, scattering, diffraction, and the absorption of light by the test sample's own material, the intensity of the X-rays will decrease to varying degrees. In addition, the buffer layer sample is irradiated with X-rays at different angles. After the rays pass through the sample, the X-ray detector detects the residual intensity of the X-rays. The projection data at each projection angle is obtained through photoelectric signal conversion and amplification processing. The projection data is reconstructed and calculated, and converted into digital images reflecting the material attenuation coefficients at various points inside the sample, thereby obtaining multiple grayscale two-dimensional slice images.

[0087] (1)

[0088] (2)

[0089] (3)

[0090] Equation (1) in the above formula expresses Bell's Law, where I This represents the intensity of X-rays after they penetrate an object. I 0 represents the initial intensity of the X-rays. μ The attenuation coefficient is related to the radiation intensity. E Equivalent atomic number Z and density ρ Related. Equation (2) represents the formula for the two-dimensional inverse Fourier transform, which is the basis of the Radon inversion principle. In the formula... f ( x,y ) is a two-dimensional function in the spatial domain, representing a point on the tomographic image after CT reconstruction. x,y The grayscale value at () F ( k x ,k y ) is a two-dimensional function in the frequency domain, and is f ( x,y The result of the two-dimensional Fourier transform represents the image. f ( x,y The spectral distribution of a frequency component represents the intensity of its specific frequency components, and the phase represents the spatial positional relationship of those frequency components; k x ,k y ) represents the spatial frequency, indicating the frequency of image variation in the horizontal and vertical directions. Equation (3) expresses the Radon inversion formula. f ( x,y ) is the reconstructed image function, representing the tomographic grayscale image of the test sample; θ The projection angle of the X-ray. s These are the detector's position coordinates, representing the projection values ​​at different positions under a fixed projection angle; Rf ( θ,s () represents projection data obtained from X-ray irradiation projection. h ( s ) is the impulse response of the ramp filter. In equation (3), the impulse response is based on the X-ray at various angles of the sample under test. θ The projection is illuminated to obtain projection data, which is then processed through a filter and based on each pixel in the image. x,y The processed projection data is then appended to each pixel to complete the reconstruction of the CT image.

[0091] S220 sequentially performs AI-based artifact removal and median filtering denoising on the two-dimensional slice image, and uses the bottom-hat transform method to trivalue the image to distinguish the material matrix, pores, and transition regions. The processed slice sequence is registered and resampled using the volume data method to generate a three-dimensional volume data field composed of voxel values. Through transfer function mapping based on voxel values ​​and direct volume rendering, a three-dimensional microstructure model of the buffer layer that can realistically reflect the internal pore distribution and fiber structure of the material is reconstructed.

[0092] Intelligent processing and 3D reconstruction of the series of two-dimensional slice images obtained in step S210 are key to achieving high-fidelity modeling in this embodiment. First, a pre-trained AI model based on deep learning is used to automatically identify and process the images, effectively removing interference such as ring artifacts introduced by the device or environment. Second, digital image processing methods such as median filtering are used to reduce noise and improve the image signal-to-noise ratio. Subsequently, morphological image processing algorithms such as bottom-hat transformation are applied, combined with threshold segmentation technology, to accurately classify the pixels in each two-dimensional slice image into three categories: material matrix, pores, and transition regions, completing the ternary processing of the image and laying the foundation for subsequent accurate material property assignment.

[0093] Based on this, a volumetric data method is used for 3D reconstruction. Specifically, all the preprocessed 2D slices are registered, aligned, and resampled in 3D space to form a continuous 3D volumetric data field consisting of voxels and their corresponding voxel values ​​(i.e., linear attenuation coefficients). Then, by defining and applying a transfer function, voxel values ​​of different ranges are mapped to specific color and opacity parameters. Finally, direct volume rendering algorithms such as ray casting are used to sample and integrate the 3D volumetric data field, rendering and reconstructing a 3D microstructure model that realistically reflects the internal pore distribution and fiber structure of the material. This model completely preserves the microscopic morphology features of the original sample, achieving a non-destructive and high-precision conversion from a physical sample to a digital model.

[0094] S230 discretizes the three-dimensional microstructure model of the buffer layer into a finite element mesh model and optimizes the mesh quality of the finite element mesh model; constructs a cable macro model, embeds the optimized mesh model into and replaces the original macro structure of the buffer layer in the cable macro model, and ensures close contact with the aluminum sheath and insulation shielding layer in the cable macro model, so as to construct a cable macro-micro three-dimensional finite element model that integrates the real microstructure of the buffer layer.

[0095] Specifically, this step aims to combine the microstructure model with the cable system model to construct a multi-scale finite element model that can be used for multiphysics simulation. This step includes two core sub-processes:

[0096] The first core sub-process is micro-meshing: the three-dimensional microstructure model of the buffer layer obtained in step S220 is discretized into a finite element mesh. Specifically, based on the volume data of the model, each voxel is used as a basic unit, and a regular hexahedral mesh is generated by determining its nodes, thus establishing the initial micro-finite element mesh. Then, according to the original voxel value (grayscale value) corresponding to the center point of each mesh unit or the area it covers, and referring to the pre-defined material grayscale range, the units are classified as materials (e.g., labeled as solid materials or pores), and each type of unit is assigned corresponding physical properties such as electrical conductivity and thermal conductivity. Due to the complexity of the microstructure, the initially generated mesh may have problems such as unit distortion and poor quality. Therefore, mesh repair techniques need to be introduced, using methods such as Laplace smoothing, angle-weighted smoothing, or homogenized elastic smoothing to adjust and optimize low-quality meshes, ensuring that the mesh quality meets the computational requirements of finite element analysis, and finally obtaining a high-precision micro-finite element mesh model of the buffer layer. For example, after processing the polyester nonwoven fiber water-blocking tape sample through the above process, its three-dimensional microstructure image and corresponding finite element mesh model can be obtained, as shown in Figure 4(a) and Figure 4(b), which intuitively demonstrate the transformation result from micromorphology to regular computational mesh.

[0097] The second core sub-process is the construction and assembly of the macroscopic model: Based on the standard design parameters of high-voltage cables, a parameter table similar to Table 2 can be referenced. Table 2 shows the structural parameters of a 110kV high-voltage cable. A macroscopic finite element model of the cable, including the conductor (copper core conductor), conductor shielding layer, XLPE insulation layer, insulation shielding layer, corrugated aluminum sheath, and outer sheath, is constructed in the finite element model. The buffer layer in this model is initially a homogeneous and simplified macroscopic structure.

[0098] Table 2

[0099]

[0100] Next, the optimized micro-mesh model obtained from the first core sub-process is imported as a buffer layer component into the constructed macro-cable model, replacing the original simplified macro-structure of the buffer layer. During assembly, precise geometric positioning and contact settings ensure a tight contact between the micro-mesh component and the inner surface of the corrugated aluminum sheath and the outer surface of the insulation shield in the macro-model, thus successfully constructing a macro-micro three-dimensional finite element model of the cable that incorporates the real pore structure. Figure 5(a) shows a schematic diagram of the structure of the macro-micro three-dimensional finite element model of the cable, and Figure 5(b) shows a schematic diagram of the mesh of the macro-micro three-dimensional finite element model of the cable. This model includes both the macro-geometry and material information of the cable system and embeds details reflecting the real micro-uniformity of the material in the buffer layer region, laying the foundation for subsequent accurate simulation.

[0101] S240 optimizes the three-dimensional simulation model of cable macro-micro coupling by employing at least one of the following methods: local mesh refinement, multi-core parallel computing, model order reduction based on intrinsic orthogonal decomposition, or a reasonably selected solution method.

[0102] Among them, the reasonable selection of solution methods is used to: analyze the mathematical matrix properties of the macro-micro coupling three-dimensional simulation model of the cable, select an appropriate solution algorithm based on the mathematical matrix properties, and configure preconditioners and convergence errors for the selected algorithm.

[0103] Because the macro-micro coupled model constructed in step S230 has an extremely high number of meshes and computational degrees of freedom, directly performing full-order simulation calculations would be extremely costly. To improve simulation efficiency and make it practical for engineering applications, this embodiment employs one or more of the following techniques for accelerated optimization and preprocessing of the model:

[0104] 1. Local Mesh Refinement: Identify key areas in the model (such as the buffer layer itself and its contact interface with the aluminum sheath / shielding layer), refine the mesh only in these areas, use a coarser mesh in non-critical areas, and set a smooth transition zone to significantly reduce the total number of elements while ensuring the accuracy of key areas.

[0105] 2. Multi-core parallel computing: Configure a parallel computing environment in the finite element model, automatically decompose the computational domain of the entire model into multiple subdomains, and allocate them to multiple CPU cores for simultaneous operation, making full use of the hardware resources of modern computers to shorten the computation time.

[0106] 3. Model reduction based on intrinsic orthogonal decomposition: First, the full-order model is sampled and calculated multiple times to collect solution vectors under different parameters to form a training set; then, the intrinsic orthogonal decomposition method is used to reduce the dimensionality of the training set and extract the principal feature modes to construct the reduced-order basis; finally, the control equations are projected onto the reduced-order basis to obtain a simplified model that can capture the main features of the system and has very low degrees of freedom, which can be used for subsequent rapid analysis and parameter scanning.

[0107] 4. Select the appropriate solution method: For large linear equation systems of a specific type (such as sparse and asymmetric) generated by electro-thermal-mechanical coupling simulation, select the most suitable solution method (such as the generalized minimum residual method) and configure it with efficient preconditioners (such as incomplete LU decomposition). At the same time, set a reasonable convergence error to accelerate the iterative convergence speed of the solution process.

[0108] The above optimizations significantly reduce the time and hardware resource requirements of simulation calculations.

[0109] S250: On the optimized model, establish a simulation environment with multi-physics coupling of electric, thermal, and mechanical fields, calculate and obtain the electric field strength and current density distribution in the buffer layer area; using the ablation resistance of polyester nonwoven fiber water-blocking tape as a reference benchmark, extract the maximum electric field strength and the maximum current density from the electric field strength and current density distribution; determine whether the two maximum values ​​exceed the preset ablation threshold: if neither exceeds, the buffer layer material is deemed qualified; otherwise, there is a risk of ablation.

[0110] On the model optimized in step S240, a two-way coupled simulation environment of electro-thermal-mechanical multiphysics is established. Boundary conditions and loads (such as rated voltage and load current) consistent with actual operating conditions are applied, and steady-state or transient simulation calculations are performed. After the calculation is completed, the electric field intensity distribution cloud map and current density distribution cloud map in the buffer layer region are extracted and analyzed.

[0111] This embodiment uses a comparative evaluation method to determine performance. First, polyester nonwoven fiber water-blocking tape, which is widely used in engineering and has relatively complete performance data, is selected as the benchmark material. Modeling and simulation are performed on it according to the complete process from S210 to S250 above, and the simulation results (field distribution characteristics, maximum values, etc.) are established as the performance benchmark.

[0112] Then, the simulation results of other buffer layer materials or material combinations to be evaluated are systematically compared with this benchmark. The core evaluation indicators are the maximum values ​​of electric field strength and current density. The specific acceptance criteria are as follows: extract the maximum electric field strength from the electric field strength distribution and the maximum current density from the current density distribution, and determine whether they are lower than the ablation threshold preset based on the material's long-term tolerance and safety margin; if both are lower than the ablation threshold, the buffer layer material is considered qualified and has good ablation resistance; conversely, if either maximum value exceeds the ablation threshold, the material is considered unqualified, has a high risk of ablation, and is not suitable for use in engineering.

[0113] The technical solution provided by this invention firstly digitizes the real microscopic morphology based on CT scanning and AI image processing technology, enabling the evaluation to break away from macroscopic experience assumptions and ensuring the model's authenticity from the source. Secondly, by embedding the microscopic model into the macroscopic system and establishing close contact, a coupled simulation model that can realistically replicate the flow distribution in the field is constructed. Furthermore, to evaluate the ablation resistance of the new buffer layer material, the ablation resistance of polyester nonwoven fiber water-blocking tape is used as a benchmark. By comparing ablation thresholds such as electric field strength and current density, the ablation resistance of the material is determined based on whether the threshold parameters are exceeded. If the threshold is exceeded, the material's performance is deemed qualified; if the threshold is not exceeded, the material's ablation resistance is deemed unqualified. This constructs a quantifiable and comparable selection system, transforming the selection decision from subjective experience to objective data support. Simultaneously, by comprehensively utilizing acceleration strategies such as local mesh refinement, parallel computing, and model order reduction, the engineering bottleneck of low computational efficiency in high-fidelity models is effectively solved, giving this method good engineering applicability.

[0114] In summary, the method provided by this invention enables a full-chain, quantitative evaluation of buffer layer materials, from microstructure characterization to macroscopic system performance prediction, providing a powerful technical tool for the scientific selection and reliability design of high-voltage cable buffer layers.

[0115] exist Figure 2 Based on the illustrated embodiment, as a specific implementation of step S230 "micro-meshing" in this embodiment of the invention, discretizing the three-dimensional microstructure model of the buffer layer into a finite element mesh model may include the following steps:

[0116] Step a1: Based on the volume data constituting the three-dimensional microstructure model of the buffer layer, the voxels in the volume data are mapped into a set of hexahedral finite element elements according to their spatial arrangement relationship, thereby generating the initial finite element mesh.

[0117] Specifically, the volume data constituting the three-dimensional microstructure model of the buffer layer is used as the basis for generating the mesh. Utilizing the regular arrangement of each voxel in three-dimensional space, each voxel is directly converted into a hexahedral (cube) mesh element. By determining the vertices of all voxels as mesh nodes and defining elements according to their spatial connectivity, an initial finite element mesh with a large number of elements but regular structure, perfectly consistent with the geometry of the original three-dimensional model, is quickly generated. This method avoids the difficulties of meshing complex surfaces and preserves the original details of the microstructure to the greatest extent possible.

[0118] Step a2: Determine the grayscale value range corresponding to different material categories in the volume data. For each mesh element in the initial finite element mesh, determine the material category of the mesh element based on the grayscale value range to which the grayscale value of the mesh element's location belongs, and assign the corresponding material properties to the mesh element based on the determined material category.

[0119] First, based on the classification criteria established through the previous image triarization process, the grayscale value ranges corresponding to different material categories (such as "material matrix," "pores," and "transition regions") in the volumetric data are determined. Then, for each mesh element in the initial finite element mesh generated in step a1, its grayscale value range is determined based on the grayscale value of its center point or the area it covers in the original volumetric data, thus determining the material category of that mesh element. For example, elements with grayscale values ​​falling into the high range are identified as solid material matrices, and those falling into the low range are identified as pores. Finally, based on the determined material category, corresponding physical material properties are assigned to each mesh element, such as assigning equivalent electrical conductivity and thermal conductivity of polyester fibers to solid material elements, and corresponding air properties to pore elements. This process achieves accurate mapping of material microscopic inhomogeneities in the finite element model.

[0120] Step a3: Export the mesh elements and node information that have been assigned material properties to generate a mesh file that can be recognized by the finite element model, and import it into the finite element model to construct a three-dimensional finite element mesh model.

[0121] Specifically, all mesh element information (including element type and node connection relationships) and node coordinate information with assigned material properties are organized and exported according to a standard format to generate a mesh file that can be directly read by the finite element model. This mesh file is then imported into the selected finite element model. After successfully reading the file, the model constructs a complete 3D finite element mesh model containing complete geometry, mesh, and material information within its own environment, ready for simulation and subsequent model assembly and multiphysics analysis.

[0122] The above three steps systematically complete the conversion from a three-dimensional volume data model to a high-fidelity, attribute-based finite element mesh model, which is the core link in achieving accurate microscale modeling in this embodiment.

[0123] exist Figure 2 Based on the illustrated embodiment, as a specific implementation of step S230 "mesh quality optimization" in this embodiment of the invention, mesh quality optimization of the finite element mesh model may include the following steps:

[0124] Step b1: Evaluate the finite element mesh model and identify mesh elements whose quality is lower than the preset standard.

[0125] A comprehensive mesh quality assessment is performed on the initial 3D finite element mesh model constructed in step a3. Geometric quality indices for all elements in the mesh model are calculated, including element distortion, aspect ratio, interior angle range, and Jacobian matrix determinant value. Based on the numerical stability requirements of the simulation analysis model for the type of problem being solved (e.g., electro-thermal-mechanical coupling), acceptable standards (thresholds) for each quality index are preset. By systematically comparing the quality index of each element with the preset standards, all mesh elements with quality below the preset standards are automatically identified and marked. These elements typically exhibit excessive shape distortion, aberration, or abrupt size changes, which are the main causes of decreased computational accuracy, convergence difficulties, and even solution failure.

[0126] Step b2 involves repairing the identified mesh cells using a mesh smoothing method to improve their shape. The mesh smoothing method includes at least one of Laplacian smoothing, angle-weighted smoothing, or uniform elastic smoothing.

[0127] For the low-quality mesh cells identified in step b1, a mesh smoothing method is used to optimize and adjust their node positions to improve their cell shape and enhance the overall mesh quality. In practice, one or more of the following smoothing algorithms can be selected based on the type and degree of cell distortion:

[0128] (1) Laplace smoothing: Adjusting the position of the problem node to the centroid (average value) of its adjacent node positions is a fast and simple homogenization method.

[0129] (2) Angle-weighted smoothing: When adjusting the position of a node, the angle between it and the line connecting it to the adjacent node is considered. The unit with abnormal angle is optimized first, which is especially effective in improving the distribution of internal angles of the unit.

[0130] (3) Homogenization and elastic smoothing: The mesh element is regarded as an elastic network. By simulating the virtual elastic mechanical equilibrium process, the node position is naturally relaxed to a more uniform state under the constraint conditions, which can effectively handle complex local distortions.

[0131] By iteratively applying the selected smoothing algorithm, the node coordinates of low-quality cells are gradually adjusted until their shape parameters are significantly improved, meeting or approaching the preset quality standards.

[0132] Step b3 involves re-evaluating the quality of the repaired mesh model to ensure it meets the quality requirements of the simulation analysis.

[0133] After completing the smoothing repair in step b2, a comprehensive quality assessment is performed again on the entire repaired mesh model. Using the same quality indicators and assessment procedures as in step b1, all elements, especially those that were repaired, are examined to verify whether their quality has been improved to meet the requirements. The purpose of this assessment is to ensure that the optimized mesh model as a whole meets the stringent mesh quality requirements for subsequent electro-thermal-mechanical multiphysics coupling simulation analysis, guaranteeing the accuracy, convergence, and stability of the calculations. Only mesh models that pass this verification will be used in subsequent "model embedding" and simulation calculation steps.

[0134] The above steps b1 to b3 constitute a complete "evaluation-repair-verification" quality assurance closed loop, which is a key link in ensuring the high computational reliability of the microscale finite element mesh in the embodiments of this invention, and lays a solid mesh foundation for subsequent high-precision multiphysics simulation.

[0135] exist Figure 2 Based on the illustrated embodiment, as a specific implementation of step S240 "optimization processing using local mesh refinement" in this embodiment of the invention, the local mesh refinement of the macro-micro coupled three-dimensional simulation model of the cable can be further refined into the following steps:

[0136] Step c1: Identify the key areas that need to be analyzed in the macro-micro coupled 3D simulation model of the cable. The key areas include at least the optimized mesh model as a buffer layer component and its contact area with the aluminum sheath and insulation shield.

[0137] Specifically, the macro-micro coupled three-dimensional simulation model of the cable constructed in step S230 is analyzed to identify key areas in the model that require focused analysis. These areas are usually the parts where the physical field change gradient is large and has a decisive impact on the accuracy of the simulation results. Specifically, the key areas include at least: (1) the optimized mesh model itself as a buffer layer component, whose complex micro-pore structure is the root cause of electric field distortion and current concentration; (2) the contact area between the buffer layer component and the inner surface of the corrugated aluminum sheath, where contact resistance and interface discharge may occur; (3) the contact area between the buffer layer component and the outer surface of the insulation shield, where the electric field may be relatively concentrated. Using the selection or marking function of the finite element model, the above areas are precisely defined as the objects of subsequent mesh refinement operations.

[0138] Step c2: By controlling the mesh size of the micro-buffer layer model and the cable macro model respectively, the overall mesh size is limited; the mesh is locally refined in the boundary area between the micro-buffer layer model and the cable macro model, and the gradual change of mesh size from fine to coarse is controlled by setting the spatial distribution of finite element elements and nodes at the boundary, so as to achieve a continuous transition between meshes of different scales and to achieve a smooth transition between coarse and fine meshes.

[0139] Specifically, since macroscopic cable models typically use coarser meshes to control computational scale, while microscopic buffer layer models use finer meshes to characterize pore structures, direct coupling between the two can lead to computational instability or decreased accuracy due to abrupt changes in mesh size. Therefore, it is necessary to establish a continuous, gradual transition in mesh size from fine to coarse at the boundary between macroscopic and microscopic models to achieve compatibility between meshes of different scales and a smooth transition between coarse and fine meshes, thereby ensuring the feasibility and reliability of multiphysics coupled simulations.

[0140] Step c3 involves setting a fine mesh within the critical region and a coarse mesh for the rest of the model outside the critical region, with a mesh transition zone established at the boundary between the fine and coarse mesh areas. Within the mesh transition zone, the mesh size gradually transitions from fine to coarse.

[0141] Specifically, within the identified critical areas, a smaller mesh size (i.e., fine mesh) is set to ensure sufficient spatial resolution to capture details in these areas of complex physical field variations. Meanwhile, in the remaining parts of the model outside the critical areas (such as cable conductors, XLPE insulation layers, and outer sheaths), a relatively larger mesh size (i.e., coarse mesh) is set to control the overall number of elements in the model.

[0142] To avoid abrupt changes in element size, deterioration in element quality, and computational instability caused by direct adjacency between fine and coarse mesh regions, a mesh transition zone is created at the boundary between these regions. Within this transition zone, the mesh size smoothly transitions from fine to coarse in a gradual manner. This design ensures the continuity of mesh changes and maintains a good element shape, which is key to the successful application of local mesh refinement techniques.

[0143] Step c4 involves evaluating and verifying the mesh quality of the cable macro-micro coupled 3D simulation model after local mesh refinement and transition matching processing to ensure that the mesh quality meets the simulation analysis requirements and obtains a mesh model that can be used for calculation.

[0144] Specifically, after defining the local mesh size and generating the transition zone in step c3, a new round of mesh quality evaluation and verification is performed on the entire macro-micro coupled 3D simulation model of the cable after local mesh refinement. The key points of the check include: whether the elements in the fine mesh region reach the expected density, whether the mesh gradient in the transition zone is smooth, and whether there are any low-quality elements (such as excessively distorted elements) newly introduced due to local refinement in the overall model. This verification step aims to ensure that the locally refined mesh model as a whole meets the stringent requirements for mesh quality in electro-thermal-mechanical multiphysics coupling simulation analysis. Only models that pass verification and confirm that their mesh quality meets the requirements of simulation analysis are considered final mesh models that can be used for stable and accurate calculations and proceed to subsequent simulation calculations.

[0145] Steps c1 to c4 above detail the specific implementation process of local mesh refinement. The core advantage of this method lies in its intelligent allocation of computational resources (refining in critical regions and coarsening in non-critical regions), which significantly reduces the total degrees of freedom and computational complexity of the macro-micro coupled model without sacrificing simulation accuracy. This effectively improves the efficiency of subsequent simulation solutions, enabling high-fidelity multiphysics analysis to be completed within the time and resource constraints of engineering practice.

[0146] exist Figure 2 Based on the illustrated embodiment, as a specific implementation of step S240 "using multi-core parallel computing for optimization processing" in this embodiment of the invention, the optimization processing of the macro-micro coupled three-dimensional simulation model of the cable can be further refined into the following steps:

[0147] Step d1: Enable parallel computing in the finite element model to decompose the computational domain of the macro-micro coupled 3D simulation model of the cable into multiple subdomains.

[0148] In the selected finite element model, its built-in parallel computing module is enabled. For the macro-micro coupled 3D simulation model of the cable constructed in step S230, the model's large computational domain (i.e., the space composed of all mesh elements and nodes requiring numerical solutions) is automatically or semi-automatically decomposed into multiple geometrically continuous subdomains with a computationally balanced load. Each subdomain will serve as an independent computational task unit. This decomposition process aims to transform the large-scale overall computational task into multiple smaller, parallelizable tasks, laying the data foundation for parallel computing using multi-core processors.

[0149] Step d2: Configure a parallel solution method to distribute the computational tasks of each subdomain to multiple processor cores for simultaneous computation.

[0150] Based on the mathematical characteristics of the electro-thermal-mechanical multiphysics coupling problem to be solved (which typically generates large sparse linear equation systems), configure a solution method that supports distributed memory or shared memory parallelism in the model, such as a parallel version of the preconditioned conjugate gradient method or the generalized minimum residual method. After configuration, start the parallel solution process. At this point, the solution method will automatically distribute the computational tasks of each subdomain generated in step d1 to multiple processor cores of the computer. Each core will independently be responsible for local matrix assembly, local solution, and other computational tasks within its assigned subdomain, and will perform necessary inter-core data synchronization and communication through message passing interfaces or shared memory, thereby achieving true simultaneous parallel computing and accelerating the solution process from the hardware level.

[0151] Step d3: Test the computation time and speedup ratio under different numbers of processor cores, and determine the number of processor cores to be used for the final simulation based on the number of cores with the highest computational efficiency in the test results.

[0152] To fully leverage the parallel computing potential of specific computer hardware and achieve optimal acceleration, parallel performance testing is necessary. The specific method involves repeatedly running the simulation calculation in step d2 using different numbers of processor cores (e.g., 2, 4, 8, 16 cores), recording and comparing the total computation time for each configuration. The computation speedup ratio (i.e., the ratio of single-core computation time to multi-core computation time) is calculated based on the computation time to quantify parallel efficiency. Analyzing these test data identifies the number of cores corresponding to the shortest computation time or a stable speedup ratio; this number represents the most computationally efficient configuration given the current hardware and problem scale. Finally, based on these test results, the optimal number of processor cores is determined and used for subsequent high-fidelity coupled simulations to ensure reliable results are obtained in the shortest possible time.

[0153] Steps d1 to d3 above constitute the complete implementation process of multi-core parallel computing optimization. The core value of this method lies in intelligently decomposing large-scale computing tasks and mapping them to multi-core processors for parallel execution, thereby significantly shortening the originally lengthy serial computing time. This is one of the key enabling technologies for completing the simulation of high-precision macro-micro coupled models within an engineering-feasible timeframe, greatly improving the practicality and efficiency of this evaluation method.

[0154] exist Figure 2 Based on the illustrated embodiment, as a specific implementation of step S240 "optimization processing using a model order reduction method based on intrinsic orthogonal decomposition" in this embodiment of the invention, the optimization processing of the macro-micro coupled three-dimensional simulation model of the cable can be further refined into the following steps:

[0155] Step e1 involves performing multiple sampling calculations on the macro-micro coupled three-dimensional simulation model of the cable under an electro-thermal-mechanical multi-physics coupling environment. By changing key parameters, multiple sets of different solution vectors and their corresponding parameters are collected to form a training sample set.

[0156] The macro-micro coupled 3D simulation model of the cable constructed in step S230 is used as a high-fidelity full-order model, and a simulation environment of electro-thermal-mechanical multiphysics coupling is established on it. To capture the system response characteristics of the model under different operating conditions, one or more key input parameters (e.g., the equivalent conductivity range of the buffer layer material, the applied voltage amplitude, ambient temperature boundary conditions, etc.) are systematically changed, and the full-order model is sampled and calculated multiple times. Each calculation corresponds to a specific set of parameters and outputs the complete field solution under that operating condition (e.g., the potential, temperature, displacement, etc. of each node). These different parameter combinations and their corresponding full-order solution vectors are collected to form a training sample set containing multiple response modes of the system. This sample set aims to cover as many of the main operating conditions the model may encounter in practical applications as possible.

[0157] Step e2 involves centering and normalizing the solution vectors in the training sample set as preprocessing steps.

[0158] To improve the numerical stability and efficiency of subsequent dimensionality reduction processing, all solution vectors in the training sample set collected in step e1 are preprocessed. First, centering is performed, i.e., the mean vector of all sample solution vectors is calculated, and this mean is subtracted from each solution vector to eliminate the DC component of the data. Then, normalization is performed, typically scaling the processed data to a specific range (e.g., [-1, 1]) to make field variables of different physical dimensions or orders of magnitude comparable and to improve the convergence of the dimensionality reduction algorithm.

[0159] Step e3: Use the eigenorthogonal decomposition method to reduce the dimensionality of the preprocessed solution vector set and construct a reduced-order basis.

[0160] The preprocessed solution vector set is statistically analyzed using the intrinsic orthogonal decomposition method. POD extracts a set of eigenvectors, called POD modes, by solving the eigenvalue problem of the sample covariance matrix. These modes are sorted in descending order of their corresponding eigenvalues, representing the dominant to secondary vibrational modes in the system response. The top k POD modes with the highest energy proportions (usually requiring their cumulative energy contribution to exceed a preset threshold, such as 99.9%) are selected, and a low-dimensional subspace is spanned by them. This set of orthogonal bases is the reduced-order basis. Essentially, this step projects the high-dimensional full-order solution space onto a subspace with significantly reduced dimensionality (k is much smaller than the full-order degrees of freedom) that captures the main dynamic characteristics of the system.

[0161] Step e4: Project the control equations of the cable macro-micro coupled three-dimensional simulation model onto the reduced-order basis to obtain a reduced-order model that can capture the main features of the system and has fewer degrees of freedom than the cable macro-micro coupled three-dimensional simulation model.

[0162] The original governing equations (partial differential equations or their spatially discretized ordinary differential / algebraic equations) describing the macro-micro coupling 3D simulation model of the cable are projected onto the reduced-order basis constructed in step e3. This projection operation is achieved through methods such as Galerkin projection or least squares projection, and its mathematical essence is to use the reduced-order basis to perform variable substitution and dimensionality reduction on the original equations. Finally, a simplified ordinary differential equation or algebraic equation system with very low degrees of freedom is obtained, with the reduced-order basis coefficients (generalized coordinates) as state variables, i.e., the reduced-order model. This model inherits the main physical characteristics of the full-order model within the operating conditions covered by the training sample set, but the significant reduction in its solution dimensionality indicates an order-of-magnitude improvement in computational speed.

[0163] Step e5: Train the reduced-order model using the training sample set, and verify the trained reduced-order model using the test set parameters. Compare the verification results with the calculation results of the cable macro-micro coupling 3D simulation model to verify the accuracy of the reduced-order model.

[0164] The reduced-order model obtained in step e4 is trained using the training sample set constructed in step e1. The training process typically involves determining certain parameters in the reduced-order model or establishing a mapping relationship from input parameters to the reduced-order coefficients (e.g., using interpolation or regression methods). After training, to evaluate the generalization ability and accuracy of the reduced-order model under unseen operating conditions, a new set of parameters not used in training (test set parameters) is used to perform simulation calculations using both the reduced-order and full-order models. The prediction results of the reduced-order model are quantitatively compared with the baseline calculation results of the full-order model on key physical quantities (such as the maximum electric field strength of the buffer layer and the average current density), and their relative errors or correlation coefficients are calculated. This comparison verifies the accuracy of the reduced-order model, ensuring that its prediction error is within the acceptable error range for engineering applications. Only the validated reduced-order model will be used for subsequent large-scale parameter scanning or rapid performance evaluation, thereby achieving ultra-real-time simulation or rapid optimization design.

[0165] Steps e1 to e5 above systematically illustrate the entire implementation process of the model order reduction method based on intrinsic orthogonal decomposition. The core advantage of this method lies in its ability to extract a computationally efficient surrogate model from the computationally expensive full-order model through data-driven and mathematical projection. While maintaining the accuracy of key physical predictions, it increases the simulation speed by several orders of magnitude, providing a groundbreaking computational tool for performance evaluation and optimization of buffer layer materials that require extensive iterations or real-time analysis.

[0166] exist Figure 2Based on the illustrated embodiment, as a specific implementation of step S240 "optimization processing by reasonably selecting a solution method" in this embodiment of the invention, the control equations of the macro-micro coupled three-dimensional simulation model of the cable are numerically solved by a solution method to obtain model parameter solutions that satisfy the convergence conditions. Specifically, this can be further refined into the following steps:

[0167] Step f1: Analyze the properties of the mathematical matrix generated by the macro-micro coupling three-dimensional simulation model of the cable in the electro-thermal-mechanical multiphysics coupling simulation.

[0168] This paper delves into the underlying mathematical problems arising from the macro-micro coupled 3D simulation model of the cable constructed in step S230 during the electro-thermal-mechanical multiphysics coupling simulation. After spatially and temporally discretizing the governing equations using the finite element method, the final solution requires solving one or a series of large, sparse linear algebraic equations. The mathematical properties of the system matrices (stiffness matrix, Jacobian matrix, etc.) directly influence the choice and efficiency of the solution method. The analysis focuses on determining whether the matrices are symmetric, positive definite, exhibiting a distribution pattern of non-zero elements (sparse structure), the size of the condition number, and the block structure characteristics of the matrices resulting from multiphysics coupling. For example, electrostatic field equations may produce symmetric positive definite matrices, while strongly coupled thermo-electric problems may lead to asymmetric matrices. This analysis forms the basis for the scientific selection of solution strategies.

[0169] Step f2: Based on the properties of the mathematical matrix, select a suitable solution method and configure a preconditioner for the solution method to accelerate the convergence of the iteration.

[0170] Based on the determination of the system matrix properties in step f1, a solution method that is mathematically compatible, can guarantee convergence, and is highly efficient is selected from the algorithm library provided by the finite element model. For example, for symmetric positive definite matrices, the preconditioned conjugate gradient method is preferred; for asymmetric matrices, the generalized minimum residual method or the double conjugate gradient stabilization method can be selected.

[0171] After selecting the master solution method, an efficient preconditioner must be configured to further accelerate its iterative convergence. The choice of preconditioner also depends on the matrix properties and aims to improve the spectral distribution of the original system. Common choices include: incomplete LU decomposition is suitable for general asymmetric problems; algebraic multigrids are extremely efficient for matrices derived from elliptic partial differential equations (such as heat conduction); and for specific block structures, block Jacobi or block Gauss-Seidel preconditioners can also be used. This step is one of the most critical technical decisions for improving solution efficiency.

[0172] Step f3 sets the convergence error for the selected solution method.

[0173] The selected solution method undergoes parameter tuning to achieve the optimal balance between computational accuracy and time cost. A key setting is the convergence error: a sufficiently small positive number is set as the criterion for iterative convergence. The solution is considered converged when the relative residual norm of the solution vectors between two iterations is less than this error. Setting the error too loosely will affect the accuracy of the results, while setting it too strictly will unnecessarily increase computation time.

[0174] Through the systematic implementation of steps f1 to f3, this embodiment of the invention achieves underlying algorithm optimization for the solution process of large-scale coupled simulations. This method significantly improves the solution characteristics of the system matrix by matching the solution method, preconditioners, and parameters, thereby enabling the acquisition of high-precision numerical solutions with fewer iterations and shorter computation time. This directly improves the computational efficiency and reliability of the overall simulation process at the numerical algorithm level without changing the physical model and mesh, and is an important component in addressing the enormous computational challenges of macro-micro coupled models.

[0175] This invention also provides an electronic device, comprising:

[0176] At least one processor;

[0177] Memory for storing the at least one processor-executable instruction;

[0178] The at least one processor is configured to execute the instructions to implement the method described in the above embodiments.

Claims

1. A novel buffer layer material evaluation method for dealing with buffer layer ablation problems, characterized by, include: The buffer layer material sample to be evaluated was scanned using a CT scanning device to obtain multiple two-dimensional slice images; The two-dimensional slice image is sequentially processed with AI-based artifact removal and median filtering for noise reduction. The image is then ternary-valued using the bottom-hat transform method to distinguish the material matrix, pores, and transition regions. The processed slice sequence is registered and resampled using the volume data method to generate a three-dimensional volume data field composed of voxel values. Through transfer function mapping based on the voxel values ​​and direct volume rendering, a three-dimensional microstructure model of the buffer layer that can realistically reflect the internal pore distribution and fiber structure of the material is reconstructed. The three-dimensional microstructure model of the buffer layer is discretized into a finite element mesh model, and the mesh quality of the finite element mesh model is optimized. A macroscopic model of the cable is constructed, and the optimized mesh model is embedded and replaced with the original macroscopic structure of the buffer layer in the macroscopic model of the cable. At the same time, it is ensured that it is in close contact with the aluminum sheath and insulation shielding layer in the macroscopic model of the cable, so as to construct a macro-micro three-dimensional finite element model of the cable that integrates the microscopic structure of the real buffer layer. For the aforementioned macro-micro coupled three-dimensional simulation model of the cable, optimization is performed using at least one of the following methods: local mesh refinement, multi-core parallel computing, model order reduction based on intrinsic orthogonal decomposition, or a reasonably selected solution method. The reasonably selected solution method is used to: analyze the mathematical matrix properties of the macro-micro coupled three-dimensional simulation model of the cable, select an appropriate solution algorithm based on the mathematical matrix properties, and configure preconditioners and convergence errors for the selected algorithm. A simulation environment with electro-thermal-mechanical multiphysics coupling was established on the optimized model to calculate and obtain the electric field intensity and current density distribution in the buffer layer region. Using the ablation resistance of polyester nonwoven fiber water-blocking tape as a reference benchmark, the maximum electric field strength and the maximum current density are extracted from the electric field strength and current density distribution. It is then determined whether the two maximum values ​​exceed the preset ablation threshold. If neither exceeds the threshold, the buffer layer material is deemed qualified; otherwise, there is a risk of ablation.

2. The method of claim 1, wherein, The three-dimensional microstructure model of the buffer layer is discretized into a finite element mesh model, including: Based on the volume data constituting the three-dimensional microstructure model of the buffer layer, the voxels in the volume data are mapped into a set of hexahedral finite element elements according to their spatial arrangement relationship, thereby generating an initial finite element mesh. Determine the grayscale value range corresponding to different material categories in the volume data; for each mesh cell in the initial finite element mesh, determine the material category of the mesh cell according to the grayscale value range to which the grayscale value of the mesh cell's location belongs, and assign corresponding material properties to the mesh cell according to the determined material category; Export the mesh elements and node information that have been assigned material properties to generate a mesh file that can be recognized by the finite element model, and import it into the finite element model to construct a three-dimensional finite element mesh model.

3. The method according to claim 1, characterized in that, The finite element mesh model undergoes mesh quality optimization, including: The finite element mesh model was evaluated to identify mesh elements with quality below a preset standard. The identified mesh cells are repaired using a mesh smoothing method to improve their cell shape; wherein the mesh smoothing method includes at least one of Laplacian smoothing, angle-weighted smoothing, or uniform elastic smoothing. The repaired mesh model was re-evaluated to ensure it met the quality requirements of simulation analysis.

4. The method of claim 1, wherein, The construction of the macroscopic model of the cable includes: A macroscopic model of the cable is constructed based on the high-voltage cable structure parameter table. The macroscopic model of the cable includes at least the structure of the conductor, conductor shielding layer, XLPE insulation layer, insulation shielding layer, aluminum sheath, and outer sheath. The geometric dimensions and material properties of each layer are set according to the parameter table.

5. The method of claim 1, wherein, The macro-micro coupled 3D simulation model of the cable is optimized using local mesh refinement, including: Identify the key areas that need to be analyzed in the macro-micro coupled three-dimensional simulation model of the cable. The key areas include at least the optimized mesh model as a buffer layer component and its contact area with the aluminum sheath and the insulating shielding layer. The overall mesh size is limited by controlling the mesh size of the micro-buffer layer model and the macro-cable model respectively; the mesh is locally refined in the boundary area between the micro-buffer layer model and the macro-cable model; and the gradual change of mesh size from fine to coarse is controlled by setting the spatial distribution of finite element elements and nodes at the boundary, so as to achieve a continuous transition between meshes of different scales and to achieve a smooth transition between coarse and fine meshes. A fine mesh is set within the critical area, and a coarse mesh is set in the rest of the model outside the critical area. A mesh transition zone is set at the boundary between the fine mesh area and the coarse mesh area; wherein the mesh size in the mesh transition zone gradually changes from the fine mesh to the coarse mesh. The mesh quality of the cable macro-micro coupled three-dimensional simulation model after local mesh refinement and transition matching processing is evaluated and verified to ensure that its mesh quality meets the requirements of simulation analysis and to obtain a mesh model that can be used for calculation.

6. The method of claim 1, wherein, The macro-micro coupled three-dimensional simulation model of the cable is optimized using multi-core parallel computing, including: Parallel computing is enabled in the finite element model to decompose the computational domain of the macro-micro coupled three-dimensional simulation model of the cable into multiple subdomains; Configure a parallel solution method to distribute the computational tasks of each subdomain to multiple processor cores for simultaneous computation; The computation time and speedup were tested with different numbers of processor cores, and the number of processor cores used for the final simulation was determined based on the number of cores with the highest computational efficiency in the test results.

7. The method of claim 1, wherein, The macro-micro coupled three-dimensional simulation model of the cable is optimized using a model order reduction method based on intrinsic orthogonal decomposition, including: The macro-micro coupled three-dimensional simulation model of the cable was sampled and calculated multiple times under the electro-thermal-mechanical multi-physics coupling environment. By changing the key parameters, multiple sets of different solution vectors and their corresponding parameters were collected to form a training sample set. The solution vectors in the training sample set are preprocessed by centering and normalizing. The dimensionality of the preprocessed solution vector set is reduced by using the eigenorthogonal decomposition method to construct a reduced-order basis. By projecting the control equations of the macro-micro coupled three-dimensional simulation model of the cable onto the reduced-order basis, a reduced-order model that can capture the main features of the system and has fewer degrees of freedom than the macro-micro coupled three-dimensional simulation model of the cable is obtained. The reduced-order model is trained using the training sample set, and the trained reduced-order model is verified using the test set parameters. The verification results are compared with the calculation results of the cable macro-micro coupling three-dimensional simulation model to verify the accuracy of the reduced-order model.

8. The method according to claim 1, characterized in that, The control equations of the macro-micro coupled three-dimensional simulation model of the cable are numerically solved using a solution method to obtain model parameter solutions that satisfy the convergence condition, specifically including: Analyze the properties of the mathematical matrices generated by the macro-micro coupled three-dimensional simulation model of the cable in the electro-thermal-mechanical multiphysics coupling simulation; Based on the properties of the mathematical matrix, a suitable solution method is selected, and a preconditioner is configured for the solution method to accelerate the convergence of the iteration. Set the convergence error for the selected solution method.

9. The method of claim 1, wherein, The process involves using a CT scanning device to scan the buffer layer material sample to be evaluated, obtaining multiple two-dimensional slice images, including: Place the buffer layer material sample in the scanning area of ​​the CT scanning device; The X-ray source and detector are controlled to perform multi-angle projection scanning on the sample, and the detector receives the X-ray signal after penetrating the sample. The X-ray signal is photoelectrically converted and amplified to obtain projection data at each projection angle. The projection data is reconstructed and calculated, and then converted into a digital image reflecting the material attenuation coefficient at each point inside the sample, thereby obtaining multiple grayscale two-dimensional slice images.

10. An electronic device, comprising: include: At least one processor; Memory for storing the at least one processor-executable instruction; The at least one processor is configured to execute the instructions to implement the method according to any one of claims 1 to 9.

Images