Porous ceramic 3D printing parameter optimization method and system adopting digital twinning

Abstract

The invention discloses a porous ceramic 3D printing parameter optimization method and system adopting digital twinning, and relates to the technical field of digital twinning The method comprises the steps that design information and initial printing parameters of a target forming structure are obtained, and a digital twinning model is correspondingly constructed; taking the model overhead as a constraint, and extracting a model from a physical property calculation model library to construct a multi-field coupling simulation module; inversely selecting a model library according to an extraction result, and performing data-driven modeling to generate a forming performance prediction module; and the digital twinborn model, the multi-field coupling simulation module and the forming performance prediction module are fused to evaluate the performance of a printed product, and printing process parameters are optimized. And therefore, the technical effects of reducing the printing parameter optimization cost, improving the parameter adjustment efficiency and improving the quality of a printed ceramic product are achieved.

Description

Technical Field

[0001] This invention relates to the field of digital twin technology, and in particular to a method and system for optimizing parameters in porous ceramic 3D printing using digital twins. Background Technology

[0002] Traditional methods for optimizing parameters in porous ceramic 3D printing rely heavily on extensive experimentation. These methods involve gradually adjusting parameters by testing the performance of the printed products under different printing parameters to achieve the desired printing results. This approach is not only time-consuming and labor-intensive, but also struggles to accurately predict potential problems during the printing process, especially when dealing with complex porous ceramic structures. This results in low success rates and inconsistent product quality. Summary of the Invention

[0003] This invention provides a method and system for optimizing parameters in porous ceramic 3D printing using digital twins, in order to solve the technical problems of high cost, low efficiency and impact on the quality of printed ceramic products in the prior art, and to achieve the technical effects of reducing the cost of printing parameter optimization, improving the efficiency of parameter adjustment and improving the quality of printed ceramic products.

[0004] In a first aspect, the present invention provides a method for optimizing parameters in porous ceramic 3D printing using digital twins, wherein the method for optimizing parameters in porous ceramic 3D printing using digital twins includes: Obtain the design information of the target structure and the initial parameters of the printing process, and construct a digital twin model of the printing process.

[0005] With model overhead as a constraint, the preset physical property calculation model library is traversed to perform model extraction, and multi-field coupling simulation modules are constructed according to the model extraction results.

[0006] Based on the model extraction results, the physical property calculation model library is reverse-selected, and data-driven modeling is performed with the reverse-selection results as the target to obtain the molding performance prediction module.

[0007] The digital twin model, the multi-field coupling simulation module, and the molding performance prediction module are combined to evaluate the performance of the printed product of the target molding structure, and the initial parameters of the printing process are optimized based on the performance evaluation results.

[0008] In one feasible implementation, the design information includes at least a structural model and porous feature parameters, and the initial parameters of the printing process include at least heat source power, scanning speed, scanning spacing, layer thickness, and printing path strategy.

[0009] In one feasible implementation, the design information of the target molding structure and the initial parameters of the printing process are obtained, and a digital twin model of the printing process is constructed, including: Based on the design information, the structural model and the porous feature parameters are extracted, and the porous feature parameters are merged into the structural model to construct an ideal moldable structural model.

[0010] The ideal molded structure model is sliced ​​and analyzed based on the initial parameters of the printing process.

[0011] Based on the slice analysis results, the external wall trajectory information is extracted, and the surface reconstruction of the ideal formed structure model is performed according to the external wall trajectory information. The surface reconstruction result is the digital twin model.

[0012] In one feasible implementation, model extraction is performed by traversing a pre-defined physical property calculation model library, constrained by model overhead. Based on the model extraction results, a multi-field coupling simulation module is constructed, including: Define a model overhead evaluation factor and calculate typical model overhead by combining historical calculation logs of the physical property calculation model obtained from big data.

[0013] Obtain the scene optimization resource information of the target scene, and calculate the relative model cost in combination with the typical model cost.

[0014] The relative model overhead of multiple property calculation models is serialized, and the cumulative overhead is calculated in combination with a preset optimization resource allocation coefficient. Based on the cumulative overhead calculation result, model extraction is performed in the property calculation model library.

[0015] By combining the parameter transfer relationships between the aforementioned physical property calculation models and coupling the model extraction results, the multi-field coupling simulation module is obtained.

[0016] In one feasible implementation, the physical property calculation model library is inversely selected based on the model extraction results, and data-driven modeling is performed with the inverse selection results as the target to obtain a molding performance prediction module, which further includes: The difference between the model extraction result and the physical property calculation model library is used to obtain the inverse selection result.

[0017] Using the inverse selection result as the target modeling object, corresponding physical property sample data are obtained by combining big data.

[0018] Training multiple molding performance prediction models is performed based on the physical property sample data, and the multiple molding performance prediction models are integrated to generate the molding performance prediction module, wherein each molding performance prediction model corresponds to a physical property calculation model in the inverse selection result.

[0019] In one feasible implementation, the digital twin model, the multi-field coupling simulation module, and the molding performance prediction module are combined to evaluate the printed product performance of the target molding structure, and the initial parameters of the printing process are optimized based on the printed product performance evaluation results, including: The field distribution of the target forming structure is simulated by combining the digital twin model with the multi-field coupling simulation module, and the shape performance is evaluated based on the field distribution simulation results.

[0020] The parameter performance of the target molding structure is predicted by combining the digital twin model with the molding performance prediction module, and the parameter performance is evaluated based on the parameter performance prediction results.

[0021] The evaluation results of the shape performance and the parameter performance evaluation are used to determine whether the molding performance requirements of the target molding structure are met. If any molding performance requirement is not met, the initial parameters of the printing process are adjusted and the performance evaluation of the printed product is iterated until all molding performance requirements are met.

[0022] One feasible implementation also includes: The optimized initial parameters of the printing process are applied to the process control module of the printing equipment to drive the 3D printing process.

[0023] Real-time monitoring data is collected during the 3D printing process, and the digital twin model, the multi-field coupling simulation module, and the molding performance prediction module are updated based on the monitoring data. The real-time monitoring data includes at least temperature distribution, structural deformation, pore evolution, and performance parameters.

[0024] Secondly, the present invention also provides a parameter optimization system for porous ceramic 3D printing using digital twins, wherein the parameter optimization system for porous ceramic 3D printing using digital twins includes: The twin model building component is used to obtain the design information of the target molding structure and the initial parameters of the printing process, and to build a digital twin model of the printing process.

[0025] The coupling simulation screening component is used to extract models by traversing a pre-defined physical property calculation model library with model cost as a constraint, and to construct multi-field coupling simulation modules based on the model extraction results.

[0026] The performance prediction modeling component is used to perform reverse selection on the physical property calculation model library based on the model extraction results, and to perform data-driven modeling with the reverse selection results as the target to obtain the molding performance prediction module.

[0027] The performance evaluation and parameter optimization component is used to evaluate the performance of the printed product of the target molded structure by combining the digital twin model, the multi-field coupling simulation module and the molding performance prediction module, and optimize the initial parameters of the printing process based on the performance evaluation results.

[0028] This invention discloses a method and system for optimizing parameters in porous ceramic 3D printing using digital twins, comprising: acquiring design data of the target molding structure and its corresponding initial printing process parameters, and constructing a digital twin model for simulating the printing process; performing model extraction operations in a preset physical property calculation model library with model computational overhead as a constraint, and constructing a corresponding multi-field coupled simulation module based on the extraction results; performing reverse screening of the physical property calculation model library based on the model extraction results, and using the reverse screening results as input to perform data-driven modeling and construct a molding performance prediction module; integrating the digital twin model, the multi-field coupled simulation module, and the molding performance prediction module to evaluate the performance of the printed product of the target molding structure, and optimizing and adjusting the initial printing process parameters based on the evaluation results.

[0029] Beneficial effects: This invention acquires the design data of the target molding structure and its corresponding initial printing process parameters to construct a digital twin model for simulating the printing process; using model computational overhead as a constraint, it performs model extraction operations in a pre-set physical property calculation model library and constructs a corresponding multi-field coupled simulation module based on the extraction results; it performs reverse screening of the physical property calculation model library based on the model extraction results and uses the reverse selection results as input to perform data-driven modeling and construct a molding performance prediction module; it integrates the digital twin model, the multi-field coupled simulation module, and the molding performance prediction module to evaluate the performance of the printed product of the target molding structure and optimize and adjust the initial printing process parameters based on the evaluation results, thereby achieving the technical effects of reducing the cost of printing parameter optimization, improving the efficiency of parameter adjustment, and improving the quality of printed ceramic products. Attached Figure Description

[0030] Figure 1 This is a flowchart illustrating the parameter optimization method for porous ceramic 3D printing using digital twins according to the present invention.

[0031] Figure 2 This is a schematic diagram of the structure of the porous ceramic 3D printing parameter optimization system using digital twins according to the present invention.

[0032] The components represented by each number in the attached diagram are explained below: The components include: twin model construction component 11, coupled simulation screening component 12, performance prediction modeling component 13, and performance evaluation and parameter optimization component 14. Detailed Implementation

[0033] The above technical solutions will now be described in detail with reference to the accompanying drawings and specific embodiments to provide a better understanding of them. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be understood that the present invention is not limited to the exemplary embodiments used only to explain the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. Furthermore, it should be noted that, for ease of description, only the parts related to the present invention are shown in the drawings, not all of them.

[0034] Example 1, as Figure 1 This is a flowchart illustrating the parameter optimization method for porous ceramic 3D printing using digital twins according to the present invention. The method includes: S100: Obtain the design information of the target molding structure and the initial parameters of the printing process, and build a digital twin model of the printing process.

[0035] Specifically, the target formed structure refers to the final three-dimensional solid model expected to be manufactured through the 3D printing process. This target formed structure involves design information such as the geometry, structural features, porosity distribution, and material properties of the porous ceramic. The digital twin model is used to reflect the forming process of the target formed structure in real time in a digital environment, facilitating subsequent prediction and optimization.

[0036] Specifically, firstly, design information for the target porous ceramic structure can be obtained through CAD modeling or reverse engineering, including the three-dimensional geometric model, porosity distribution, structural dimensions, and tolerance requirements. Next, initial printing parameters are set based on the selected ceramic material and equipment type, such as a nozzle diameter of 0.4 mm, a layer thickness of 50 μm, a printing speed of 20 mm / s, and a sintering temperature of 1200℃. Then, based on this input information, a digital twin model is constructed in a twin simulation platform to simulate the physical behavior during the printing process, such as material deposition path, surface condition, porosity evolution, heat conduction, sintering shrinkage, and warpage deformation.

[0037] In some embodiments, the design information includes at least a structural model and porous feature parameters, and the initial parameters of the printing process include at least heat source power, scanning speed, scanning spacing, layer thickness, and printing path strategy.

[0038] Specifically, the design information includes the structural model and porous characteristic parameters of the target molding structure. The structural model is the three-dimensional geometry of the molded part, such as the existing three-dimensional model in the design data or the three-dimensional model reconstructed based on two-dimensional drawings. The porous characteristic parameters include the porosity and pore size distribution of the target molding structure, which are key factors affecting material properties such as mechanical strength, thermal conductivity, and air permeability.

[0039] Specifically, the initial parameters of the printing process are the original process control variables set for the target shaped structure, including heat source power (affecting nozzle temperature), scanning speed (i.e., printing speed), scanning spacing, layer thickness, and printing path strategy (affecting the scanning path), etc.

[0040] In some embodiments, obtaining design information of the target molded structure and initial parameters of the printing process, and constructing a digital twin model of the printing process includes: Based on the design information, the structural model and the porous feature parameters are extracted, and the porous feature parameters are merged into the structural model to construct an ideal molding structural model; the ideal molding structural model is sliced ​​and analyzed according to the initial parameters of the printing process; the external wall trajectory information is extracted based on the slice analysis results, and the surface reconstruction of the ideal molding structural model is performed according to the external wall trajectory information, and the surface reconstruction result is output as the digital twin model.

[0041] Specifically, slicing analysis is a key transformation step in 3D printing. It is used to generate two-dimensional cross-sectional views of a three-dimensional model layer by layer along the Z-axis, and then generate corresponding scanning paths according to the printing path strategy to guide the layer-by-layer printing path. For example, the slicing analysis results are represented by G-code.

[0042] Specifically, the external wall trajectory information refers to the printing path data of the model's outer contour in each slice of the slice analysis results, which determines the printing accuracy and surface quality. Based on the external wall trajectory information mentioned above, the surface of the constructed model can be fitted or reconstructed, thereby more accurately reflecting the surface morphology after material deposition during the actual printing process, and providing a more realistic geometric basis for the digital twin model (i.e., the surface reconstruction process).

[0043] Specifically, firstly, the structural model and porous feature parameters are extracted based on the design information, and then the porous feature parameters are merged into the structural model to construct an ideal shaped structural model. For example, for an aero-engine blade with a complex internal cavity heat dissipation structure, its structural model can be considered as the complete three-dimensional configuration of the blade's shape and the relatively large internal main heat dissipation channel. The porous feature parameters correspond to the micropore size, distribution density, etc., of the inner wall of the heat dissipation channel of the blade. By merging the porous feature parameters with the blade structural model, an ideal shaped structural model containing detailed porous features can be formed.

[0044] Specifically, the ideal molded structure model is then analyzed by slicing based on the initial parameters of the printing process: First, the ideal molded structure model is decomposed into multiple two-dimensional slices based on the layer thickness parameters, with each slice corresponding to a processing layer in the printing process. Then, each slice in the slice analysis results is traversed, the outer contour path of the slice is identified, and the corresponding external wall trajectory information is extracted.

[0045] Furthermore, actual printing may produce surface textures such as layer patterns with different characteristics. Therefore, based on the extracted exterior wall trajectory information, a triangular mesh-based interpolation algorithm is used to reconstruct the surface of the ideal formed structure model, resulting in a smooth and continuous surface model. The surface reconstruction result is then output as a digital twin model. This digital twin model can accurately map the surface morphology and internal porous features of the formed structure during the actual printing process, thus providing a high-precision digital foundation for subsequent simulation and optimization.

[0046] The digital twin model constructed through the above process can accurately reflect the geometry, internal porous features, and surface morphology corresponding to the printing process parameters of the target molding structure during the printing process. This provides a reliable digital foundation for subsequent multi-field coupling simulations, molding performance predictions, and other operations, ensuring the accuracy and authenticity of the simulation results, thereby effectively improving the efficiency and precision of the entire parameter optimization method.

[0047] S200: With model overhead as a constraint, it traverses the preset physical property calculation model library to perform model extraction, and constructs a multi-field coupling simulation module based on the model extraction results.

[0048] Specifically, model overhead refers to the comprehensive consumption of computing resources, computing time, and model complexity by the physical property calculation model during the calculation process. This model overhead is quantitatively evaluated through the model overhead evaluation factor.

[0049] Specifically, the physical property calculation model library is a pre-set collection of models related to the calculation of physical properties of printing materials, which can be used to calculate the physical properties of materials such as heat conduction, heat accumulation, stress and strain during the printing process; the multi-field coupling simulation module is a comprehensive simulation tool that can simultaneously simulate multiple interrelated physical fields, such as temperature field and stress field. It is a model tool that analyzes the interaction of multiple physical phenomena during the printing process through existing explicit physical property knowledge.

[0050] In some embodiments, model extraction is performed by traversing a preset physical property calculation model library under the constraint of model cost, and a multi-field coupling simulation module is constructed based on the model extraction results, including: Define a model cost evaluation factor and calculate typical model costs by combining historical calculation logs of physical property calculation models obtained from big data; obtain scene optimization resource information of the target scene and calculate relative model costs by combining the typical model costs; serialize the relative model costs of multiple physical property calculation models, calculate cumulative costs by combining preset optimization resource allocation coefficients, and perform model extraction in the physical property calculation model library based on the cumulative cost calculation results; combine the parameter transfer relationship between the physical property calculation models, couple the model extraction results, and obtain the multi-field coupling simulation module.

[0051] Specifically, a material property calculation model refers to a mathematical or physical model used to simulate the physical properties of materials, such as a heat transfer model, a material constitutive relation model, and a stress-strain relation module; a multi-field coupling simulation module is a simulation system that links and solves multiple material property calculation models together, which can reflect the interaction of multiple physical mechanisms in the manufacturing process, such as the interaction between the temperature field and the strain field.

[0052] Specifically, relative model overhead refers to the ratio of model performance overhead to resource constraints in the current target scenario, which is calculated by mapping typical model overhead to resource constraints. This ratio is used for model selection under resource-constrained conditions. Optimized resource allocation coefficients are weighted parameters used to allocate computational resources during model selection, reflecting the biased strategies for computational accuracy (i.e., scenario resources allocated to the multi-field coupled simulation module) and efficiency (i.e., scenario resources allocated to the performance prediction module) in different scenarios.

[0053] Specifically, firstly, a model cost evaluation factor is defined. This factor needs to comprehensively consider the consumption of computing resources, model complexity, and computation time to ensure that the selected physical property calculation model can efficiently complete the analysis and calculation tasks.

[0054] Specifically, the following steps involve combining the historical computation logs of the physical property computation model obtained from big data, and statistically obtaining the average computational cost of each type of typical model based on the computational resource consumption, model complexity, and computation time in the historical computation logs, and outputting the typical model cost.

[0055] Furthermore, the resource configuration parameters of the target scenario are obtained, and the ratio of the weighted typical model overhead to the overhead evaluation factors in the resource configuration parameters is used to obtain the relative model overhead. The smaller this value is, the lighter the corresponding physical property calculation model is, and the more suitable it is to be selected first in the current scenario.

[0056] Furthermore, the relative model costs of multiple models in the property calculation model library are serialized, and cumulative cost calculation is performed in combination with a preset optimization resource allocation coefficient. That is, multiple relative model cost values ​​are accumulated one by one from small to large until the accumulated result just exceeds the optimization resource allocation coefficient. At this point, N relative model cost values ​​are accumulated. Then, the property calculation models corresponding to the first N-1 relative model cost values ​​are extracted from the property calculation model library as the property calculation models that meet the model cost expectations. In addition, the parameter transfer relationship between the property calculation models, such as the correlation between heat conduction parameters and stress and strain parameters, is combined to couple the model extraction results and construct a multi-field coupled simulation module.

[0057] Among them, the optimization resource allocation coefficient is determined according to the parameter optimization requirements of the target scenario and can be configured between 0.5 and 0.7, such as 0.6 for the multi-field coupling simulation module and 0.4 for the molding performance prediction module.

[0058] The multi-field coupling simulation module constructed through the above process can accurately simulate the interactions of various physical fields in the porous ceramic printing process while meeting model overhead constraints, providing support for subsequent performance evaluation and parameter optimization.

[0059] S300: Based on the model extraction results, the physical property calculation model library is reverse-selected, and data-driven modeling is performed with the reverse-selection results as the target to obtain the molding performance prediction module.

[0060] Specifically, deselection is the process of removing selected elements from a complete set to obtain the remaining portion. In other words, it involves removing models selected for multi-field coupling simulations from the physical property calculation model library to obtain the set of unselected models. Data-driven modeling refers to using a large amount of historical data samples and machine learning or statistical modeling methods to automatically learn the mapping relationship between inputs and outputs, replacing the traditional physical modeling process.

[0061] Through the above process, a data-driven molding performance prediction channel can be constructed under resource-constrained conditions to replace the resource-intensive physical property calculation model, achieving complementary synergy between physical modeling and data modeling. That is, when some physical properties are difficult to model or computational resources are insufficient, high-precision prediction results can be obtained through data learning.

[0062] In some embodiments, the module for inverse selection of the physical property calculation model library based on the model extraction results, and data-driven modeling with the inverse selection results as the target, to obtain a molding performance prediction module, further includes: The difference between the model extraction result and the physical property calculation model library is taken to obtain the inverse selection result; the inverse selection result is used as the target modeling object, and the corresponding physical property sample data is obtained by combining big data; the training of multiple molding performance prediction models is performed according to the physical property sample data, and multiple molding performance prediction models are integrated to generate the molding performance prediction module, wherein each molding performance prediction model corresponds to a physical property calculation model in the inverse selection result.

[0063] Specifically, a molding performance prediction model is a model that uses material or structural parameters as input to predict the performance parameters of the molding result. It can be constructed using regression neural networks, ensemble learning methods, mathematical fitting methods, etc. Physical property sample data refers to material physical property data obtained based on experimental or sensor data acquisition, including input features and output labels (i.e., performance parameters).

[0064] Specifically, firstly, based on the results extracted from the previous stage of physical property calculation models, such as the selected heat conduction physical property calculation model and stress physical property calculation model, a difference operation is performed with the entire physical property calculation model library to obtain a set of unselected physical property calculation models as the inverse selection result; then, the physical properties described by multiple physical property calculation models in the inverse selection result are used as the target modeling objects, and corresponding physical property sample data are collected from historical printing databases or experimental measurement records using big data technology.

[0065] Furthermore, machine learning algorithms or statistical analysis methods are used to train corresponding molding performance prediction models for each physical property calculation model in the inverse selection results. For example, for sintering shrinkage rate prediction, XGBoost, LightGBM, and deep neural networks are used to train molding performance prediction models. Similarly, a molding performance prediction model for predicting the bending strength of the printed part is trained for the elastic modulus model, and a molding performance prediction model for predicting the pore connectivity of the printed part is trained for the fluid permeation model. Finally, the multiple molding performance prediction models obtained above are merged and output to obtain the molding performance prediction module, so as to realize the rapid prediction and compensation of physical properties that are not physically modeled (i.e., target modeling objects).

[0066] Through the above process, the molding performance prediction module is trained based on physical property sample data. This can accurately predict various performance indicators of porous ceramic printed parts while reducing the performance burden of the expensive physical property calculation model on the target scenario, thereby helping to improve the overall performance evaluation and parameter optimization efficiency.

[0067] S400: Combine the digital twin model, the multi-field coupling simulation module, and the molding performance prediction module to evaluate the performance of the printed product of the target molding structure, and optimize the initial parameters of the printing process based on the performance evaluation results.

[0068] In some embodiments, the performance evaluation of the printed product of the target molded structure is performed by combining the digital twin model, the multi-field coupling simulation module, and the molding performance prediction module, and the initial parameters of the printing process are optimized based on the performance evaluation results, including: The field distribution of the target molding structure is simulated by combining the digital twin model with the multi-field coupling simulation module, and the shape performance is evaluated based on the field distribution simulation results. The parametric performance of the target molding structure is predicted by combining the digital twin model with the molding performance prediction module, and the parametric performance is evaluated based on the parametric performance prediction results. The molding performance requirements of the target molding structure are met by judging the shape performance evaluation results and parametric performance evaluation results. If any molding performance requirement is not met, the initial parameters of the printing process are adjusted and the performance evaluation of the printed product is iteratively performed until all molding performance requirements are met.

[0069] Specifically, field distribution simulation is performed on a digital twin model using a multi-field coupling simulation module to calculate the distribution of physical fields such as temperature and stress fields, which is used to predict the material's response behavior in the spatial and temporal dimensions during the printing process. Shape performance evaluation is a quantitative analysis of the geometric accuracy, warpage, and dimensional deviations of the target structure based on the field distribution simulation results, in order to determine whether the initial parameters of the current printing process meet the design shape requirements.

[0070] Specifically, parametric performance prediction is the process of predicting the mechanical, electrical, and thermal properties of materials or structures under specific process parameters through a molding performance prediction module. By comparing the parametric performance prediction results with the design specifications, it can be determined whether the initial parameters of the current printing process meet the performance requirements. Molding performance requirements are determined by the design goals, and for example, include geometric accuracy, strength, porosity, and thermal conductivity.

[0071] Specifically, firstly, based on the digital twin model of the target molding structure and the multi-field coupling simulation module, the field distribution during the molding process is simulated, such as simulating the coupling of the thermal field and the stress field, thereby predicting the temperature gradient and stress concentration distribution of the structure after printing each layer, and identifying stress concentration areas based on the field distribution simulation results, and calculating their corresponding strain; then, the obtained strain is compared with the design tolerance to identify possible defects such as warping, collapse, misalignment, and cracking, and to complete the evaluation of the shape performance.

[0072] Furthermore, the digital twin model is combined with the molding performance prediction module. By inputting the same printing process parameters, the key performance parameters of the target molding structure, such as bending strength, density, and thermal conductivity, are predicted. The predicted values ​​(parameter performance prediction results) are then compared with the target performance indicators to complete the parameter performance evaluation.

[0073] Furthermore, if any performance indicator fails to meet the standard, such as warpage exceeding the tolerance or insufficient density, the optimization module is triggered to adjust the initial parameters of the printing process, such as reducing the scanning speed, increasing the preheating temperature, optimizing the scanning path, changing the layer height, etc. Based on the new printing process parameters, the field distribution simulation, shape performance evaluation, parameter performance prediction, and parameter performance evaluation are performed again. The above process is repeated until all evaluation results meet the molding performance requirements.

[0074] Through the above process, a closed-loop iterative optimization mechanism with performance as the objective is established. This allows for precise control of the forming process before printing through simulation and prediction, which helps improve the geometric accuracy and functional performance of printed products, while reducing trial and error costs and material waste in parameter optimization and improving the efficiency of printing process parameter optimization for complex structures.

[0075] In some embodiments, it also includes: The optimized initial parameters of the printing process are applied to the process control module of the printing equipment to drive the 3D printing process. Real-time monitoring data is collected during the 3D printing process, and the digital twin model, the multi-field coupling simulation module and the molding performance prediction module are updated based on the monitoring data. The real-time monitoring data includes at least temperature distribution, structural deformation, pore evolution and performance parameters.

[0076] Specifically, the process control module is the control unit in a 3D printing device. It is used to receive and execute printing process parameters such as laser power, scanning speed, layer thickness, and scanning path, and to control various operations in the printing process accordingly.

[0077] Specifically, real-time monitoring data refers to dynamic information related to the forming state collected during the printing process through sensors, camera systems, infrared thermal imagers, experimental equipment, etc., such as temperature distribution, structural deformation, porosity evolution, and performance parameters. By comparing the actual monitoring data with the outputs of the digital twin model, the multi-field coupling simulation module, and the forming performance prediction module, deviations can be identified and the model can be corrected accordingly, thereby improving prediction accuracy and responsiveness and maintaining consistency between the digital twin model and the physical entity.

[0078] Through the above process, a closed-loop control mechanism encompassing virtual optimization, physical execution, and real-time feedback is achieved. This enables the digital twin model to not only possess predictive capabilities before printing but also adaptive adjustment and dynamic response capabilities during printing. This helps to continuously improve the stability and consistency of the printing process, effectively address uncertainties such as material fluctuations and environmental changes, and ultimately enhance the forming quality and performance reliability of the printed product.

[0079] In summary, the parameter optimization method for porous ceramic 3D printing using digital twins provided by this invention has the following technical effects: By acquiring the design data of the target molding structure and its corresponding initial printing process parameters, a digital twin model for simulating the printing process is constructed. Using model computational overhead as a constraint, a model extraction operation is performed in a pre-defined physical property calculation model library, and a corresponding multi-field coupled simulation module is constructed based on the extraction results. The physical property calculation model library is then reverse-selected based on the model extraction results, and data-driven modeling is performed using this reverse selection result as input to construct a molding performance prediction module. By integrating the digital twin model, the multi-field coupled simulation module, and the molding performance prediction module, the performance of the printed product of the target molding structure is evaluated, and the initial printing process parameters are optimized and adjusted based on the evaluation results. This achieves the technical effects of reducing printing parameter optimization costs, improving parameter adjustment efficiency, and enhancing the quality of printed ceramic products.

[0080] Example 2, as Figure 2 This is a schematic diagram of the structural optimization system for porous ceramic 3D printing using digital twins, as described in this invention. For example, Figure 1 The flowchart of the method for optimizing parameters of porous ceramic 3D printing using digital twins in this invention can be seen as follows: Figure 2 The structure shown is implemented.

[0081] Based on the same concept as the porous ceramic 3D printing parameter optimization method using digital twins in the embodiments described above, the present invention also provides a porous ceramic 3D printing parameter optimization system using digital twins, comprising: The twin model building component 11 is used to obtain the design information of the target molding structure and the initial parameters of the printing process, and to build a digital twin model of the printing process.

[0082] The coupling simulation screening component 12 is used to perform model extraction by traversing the preset physical property calculation model library with model cost as a constraint, and to construct a multi-field coupling simulation module according to the model extraction results.

[0083] The performance prediction modeling component 13 is used to perform reverse selection on the physical property calculation model library based on the model extraction results, and to perform data-driven modeling with the reverse selection results as the target to obtain the molding performance prediction module.

[0084] The performance evaluation and parameter optimization component 14 is used to evaluate the performance of the printed product of the target molding structure by combining the digital twin model, the multi-field coupling simulation module and the molding performance prediction module, and optimize the initial parameters of the printing process based on the performance evaluation results of the printed product.

[0085] In some embodiments, the twin model building component 11 includes: An ideal molding structure model construction unit is used to extract the structure model and the porous feature parameters according to the design information, and merge the porous feature parameters into the structure model to construct an ideal molding structure model.

[0086] The slicing analysis unit is used to perform slicing analysis on the ideal molded structure model based on the initial parameters of the printing process.

[0087] The digital twin model generation unit is used to extract the external wall trajectory information based on the slice analysis results, and perform surface reconstruction of the ideal formed structure model according to the external wall trajectory information, and output the surface reconstruction result as the digital twin model.

[0088] In some embodiments, the coupling simulation screening component 12 includes: The typical model cost calculation unit is used to define model cost evaluation factors and calculate typical model costs by combining historical calculation logs of physical property calculation models obtained based on big data.

[0089] The relative model cost calculation unit is used to obtain scene optimization resource information of the target scene and calculate the relative model cost in combination with the typical model cost.

[0090] The model extraction unit is used to serialize the relative model costs of multiple property calculation models, perform cumulative cost calculation in combination with preset optimization resource allocation coefficients, and perform model extraction in the property calculation model library based on the cumulative cost calculation results.

[0091] The multi-field coupling simulation module coupling unit is used to combine the parameter transfer relationship between the physical property calculation models, couple the model extraction results, and obtain the multi-field coupling simulation module.

[0092] In some embodiments, the performance prediction modeling component 13 includes: The inverse selection result acquisition unit is used to obtain the difference between the model extraction result and the physical property calculation model library to obtain the inverse selection result.

[0093] The physical property sample data acquisition unit is used to acquire corresponding physical property sample data by combining the inverse selection result as the target modeling object and big data.

[0094] The molding performance prediction module generation unit is used to train multiple molding performance prediction models according to the physical property sample data, and integrate the multiple molding performance prediction models to generate the molding performance prediction module, wherein each molding performance prediction model corresponds to a physical property calculation model in the inverse selection result.

[0095] In some embodiments, the performance evaluation and parameter optimization component 14 includes: The shape performance evaluation unit is used to combine the digital twin model with the multi-field coupling simulation module to simulate the field distribution of the target forming structure, and to evaluate the shape performance based on the field distribution simulation results.

[0096] The parameter performance evaluation unit is used to combine the digital twin model with the molding performance prediction module to predict the parameter performance of the target molding structure, and to evaluate the parameter performance based on the parameter performance prediction results.

[0097] The printing process parameter optimization and iterative evaluation unit is used to determine whether the shape performance evaluation results and parameter performance evaluations meet the molding performance requirements of the target molded structure. If any molding performance requirement is not met, the initial parameters of the printing process are adjusted and the performance evaluation of the printed product is iteratively performed until all molding performance requirements are met.

[0098] In some implementations, the design information includes at least a structural model and porous feature parameters, and the initial parameters of the printing process include at least heat source power, scanning speed, scanning spacing, layer thickness, and printing path strategy.

[0099] In some implementations, the execution steps of the performance evaluation and parameter optimization component 14 also include: The optimized initial parameters of the printing process are applied to the process control module of the printing equipment to drive the 3D printing process. Real-time monitoring data is collected during the 3D printing process, and the digital twin model, the multi-field coupling simulation module, and the molding performance prediction module are updated based on this monitoring data. The real-time monitoring data includes at least temperature distribution, structural deformation, pore evolution, and performance parameters.

[0100] It should be understood that the focus of the embodiments mentioned in this specification is their difference from other embodiments. The specific embodiments in the aforementioned Embodiment 1 are also applicable to the porous ceramic 3D printing parameter optimization system using digital twins described in Embodiment 2. For the sake of brevity, they will not be elaborated further here.

[0101] It should be understood that the embodiments disclosed in this invention and the above description enable those skilled in the art to implement this invention. However, this invention is not limited to the embodiments mentioned above. It should be understood that those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this invention, and should all be included within the protection scope of this invention.

Claims

1. A method for optimizing parameters in porous ceramic 3D printing using digital twins, characterized in that, include: Obtain the design information of the target molding structure and the initial parameters of the printing process, and construct a digital twin model of the printing process; With model overhead as a constraint, the preset physical property calculation model library is traversed to perform model extraction, and multi-field coupling simulation modules are constructed according to the model extraction results. Based on the model extraction results, the physical property calculation model library is reverse-selected, and data-driven modeling is performed with the reverse selection results as the target to obtain the molding performance prediction module. The digital twin model, the multi-field coupling simulation module, and the molding performance prediction module are combined to evaluate the performance of the printed product of the target molding structure, and the initial parameters of the printing process are optimized based on the performance evaluation results.

2. The method for optimizing parameters of porous ceramic 3D printing using digital twins as described in claim 1, characterized in that, The design information includes at least a structural model and porous feature parameters, and the initial parameters of the printing process include at least heat source power, scanning speed, scanning spacing, layer thickness, and printing path strategy.

3. The method for optimizing parameters of porous ceramic 3D printing using digital twins as described in claim 2, characterized in that, Obtain the design information and initial parameters of the printing process for the target structure, and construct a digital twin model of the printing process, including: Based on the design information, the structural model and the porous feature parameters are extracted, and the porous feature parameters are merged into the structural model to construct an ideal moldable structural model; The ideal molded structure model is sliced ​​and analyzed based on the initial parameters of the printing process. Based on the slice analysis results, the external wall trajectory information is extracted, and the surface reconstruction of the ideal formed structure model is performed according to the external wall trajectory information. The surface reconstruction result is the digital twin model.

4. The method for optimizing parameters of porous ceramic 3D printing using digital twins as described in claim 3, characterized in that, Constrained by model overhead, the system iterates through a pre-defined library of physical property calculation models to extract models, and constructs a multi-field coupled simulation module based on the model extraction results, including: Define model overhead evaluation factors and calculate typical model overhead by combining historical calculation logs of physical property calculation models obtained from big data. Obtain scene optimization resource information for the target scene, and calculate the relative model cost in conjunction with the typical model cost; The relative model overhead of multiple property calculation models is serialized, and the cumulative overhead is calculated in combination with the preset optimization resource allocation coefficient. Based on the cumulative overhead calculation result, model extraction is performed in the property calculation model library. By combining the parameter transfer relationships between the aforementioned physical property calculation models and coupling the model extraction results, the multi-field coupling simulation module is obtained.

5. The method for optimizing parameters of porous ceramic 3D printing using digital twins as described in claim 4, characterized in that, Based on the model extraction results, the physical property calculation model library is inversely selected, and data-driven modeling is performed with the inverse selection results as the target to obtain a molding performance prediction module, which also includes: The difference between the model extraction result and the physical property calculation model library is used to obtain the inverse selection result; Using the inverse selection results as the target modeling object, and combining big data to obtain the corresponding physical property sample data; Training multiple molding performance prediction models is performed based on the physical property sample data, and the multiple molding performance prediction models are integrated to generate the molding performance prediction module, wherein each molding performance prediction model corresponds to a physical property calculation model in the inverse selection result.

6. The method for optimizing parameters of porous ceramic 3D printing using digital twins as described in claim 5, characterized in that, The performance of the printed product of the target molded structure is evaluated by combining the digital twin model, the multi-field coupling simulation module, and the molding performance prediction module. Based on the performance evaluation results, the initial parameters of the printing process are optimized, including: The field distribution of the target forming structure is simulated by combining the digital twin model with the multi-field coupling simulation module, and the shape performance is evaluated by combining the field distribution simulation results. The parameter performance of the target molding structure is predicted by combining the digital twin model with the molding performance prediction module, and the parameter performance is evaluated based on the parameter performance prediction results. The evaluation results of the shape performance and the parameter performance evaluation are used to determine whether the molding performance requirements of the target molding structure are met. If any molding performance requirement is not met, the initial parameters of the printing process are adjusted and the performance evaluation of the printed product is iterated until all molding performance requirements are met.

7. The method for optimizing parameters of porous ceramic 3D printing using digital twins as described in claim 1, characterized in that, Also includes: The optimized initial parameters of the printing process are applied to the process control module of the printing equipment to drive the 3D printing process. Real-time monitoring data is collected during the 3D printing process, and the digital twin model, the multi-field coupling simulation module, and the molding performance prediction module are updated based on the monitoring data. The real-time monitoring data includes at least temperature distribution, structural deformation, pore evolution, and performance parameters.

8. A parameter optimization system for porous ceramic 3D printing using digital twins, characterized in that, The system is used to implement the parameter optimization method for porous ceramic 3D printing using digital twins as described in any one of claims 1 to 7, including: The twin model building component is used to obtain the design information of the target molding structure and the initial parameters of the printing process, and to build a digital twin model of the printing process; The coupling simulation screening component is used to extract models by traversing a pre-defined physical property calculation model library with model cost as a constraint, and to construct multi-field coupling simulation modules based on the model extraction results. The performance prediction modeling component is used to perform reverse selection on the physical property calculation model library based on the model extraction results, and to perform data-driven modeling with the reverse selection results as the target to obtain the molding performance prediction module. The performance evaluation and parameter optimization component is used to evaluate the performance of the printed product of the target molded structure by combining the digital twin model, the multi-field coupling simulation module and the molding performance prediction module, and optimize the initial parameters of the printing process based on the performance evaluation results.

Images