Additive manufacturing electrodeposition synergistic construction of amorphous nanocrystalline composite lattice structure and machine learning driven performance optimization method thereof
By using laser powder bed melting and electrodeposition technology to synergistically fabricate amorphous nanocrystalline composite lattice structures, and combining machine learning optimization methods, the problem of balancing strength and plasticity in amorphous alloy lattice structures has been solved, achieving efficient performance optimization and customized component fabrication.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- BEIJING UNIV OF TECH
- Filing Date
- 2026-03-19
- Publication Date
- 2026-06-19
AI Technical Summary
Existing technologies are insufficient for efficiently preparing amorphous alloy lattice structures that combine high strength and excellent plasticity. Furthermore, traditional uniform coating designs are inefficient and fail to tap the performance potential of non-uniform coating distributions, thus failing to meet the application requirements of engineering structures.
Amorphous alloy lattice structure matrix was prepared by laser powder bed melting technology, and nanocrystalline twinned copper coating and nanocrystalline nickel-tungsten alloy coating were deposited sequentially by electrodeposition technology. Combined with machine learning-driven performance optimization methods, a Gaussian process regression surrogate model and a Bayesian optimization strategy were constructed to efficiently optimize multi-dimensional design parameters.
It achieves high strength and excellent plastic deformation capability of amorphous alloy lattice structure, significantly reduces experimental costs and R&D cycle, and provides a customized high-end component preparation method.
Smart Images
Figure CN122245553A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of advanced materials and structure manufacturing technology, and in particular relates to an additive manufacturing electrodeposition co-construction of amorphous nanocrystalline composite lattice structures and its machine learning-driven performance optimization method. Background Technology
[0002] Lattice structures, as lightweight, high-strength, and multifunctional structures, can achieve structural lightweighting while maintaining excellent mechanical properties through reasonable configuration design, showing broad application prospects in fields such as machinery, electronics, medicine, and aerospace. Amorphous alloys, also known as metallic glasses, possess excellent mechanical and physical properties such as high strength, large elastic strain limit, and corrosion resistance. Combining the material properties of amorphous alloys with the design flexibility of lattice structures holds promise for obtaining structural materials that can withstand high loads, are corrosion-resistant, lightweight, and have high specific strength. However, amorphous alloys have extremely poor plasticity and toughness, making them prone to brittle fracture under external loads. This limits amorphous alloy lattice structures to functional components without load-bearing requirements, such as catalyst carriers, rather than meeting the application needs of engineering structural construction. Laser powder bed melting technology can achieve near-net-shape three-dimensional forming, with cooling rates reaching 10⁻⁶. 4 High K / s ratios offer technical advantages for preparing amorphous alloy lattice structures, but the inherent differences in cooling and heating rates can lead to partial crystallization. Electrodeposition, as a surface modification technique, can uniformly deposit metal coatings on complex three-dimensional substrates, improving the mechanical properties of amorphous alloys by suppressing shear band instability and propagation. Among these, the multifunctional gradient composite coating, consisting of an inner layer of nanotwinned copper and an outer layer of nanocrystalline nickel-tungsten alloy, significantly enhances plasticity compared to traditional single metal coatings. The inner copper coating acts as a flexible transition layer, mitigating local strain in the amorphous substrate through its own plastic deformation, while the outer nickel-tungsten alloy coating provides rigid constraints during substrate deformation, effectively hindering shear band propagation and forming a gradient stress field that significantly improves the mechanical properties of amorphous alloys. However, the three-dimensional morphology of the honeycomb lattice structure is complex, and the coating thickness gradient distribution and macroscopic plasticity and toughness constitute a multi-dimensional strong nonlinear relationship space. Traditional uniform coating design is not only inefficient and requires a lot of time, but also makes it difficult to explore the performance potential brought by non-uniform coating distribution. Relying solely on finite element analysis simulation cannot achieve high-throughput exploration and global optimization of the entire design space. There is an urgent need for a new design method that can deeply integrate physical simulation and data-driven approach to intelligently simulate and lock in the optimal gradient coating scheme. Summary of the Invention
[0003] To address the aforementioned technical problems, this invention proposes an additive manufacturing electrodeposition method for co-constructing amorphous nanocrystalline composite lattice structures and a machine learning-driven performance optimization method, thereby resolving the issues present in the prior art.
[0004] In a first aspect, to achieve the above objectives, the present invention provides an additive manufacturing electrodeposition method for co-constructing amorphous nanocrystalline composite lattice structures and a machine learning-driven performance optimization method thereof, comprising the following steps: S1. Amorphous alloy lattice structure matrix is prepared using laser powder bed melting technology; S2. Using electrodeposition technology, nano-twinned copper coating and nano-crystalline nickel-tungsten alloy coating are sequentially deposited on the surface of the amorphous alloy lattice structure substrate to form a composite coating. S3. Perform mechanical property tests on the lattice structure after deposition of composite coating to obtain mechanical property data; S4. Using the geometric design parameters of the lattice structure and the coating design parameters of the composite coating as inputs, and the mechanical performance data as outputs, construct a dataset; S5. Train a Gaussian process regression surrogate model based on the dataset, and iteratively update the model using a Bayesian optimization strategy; S6. Use a genetic algorithm or particle swarm optimization algorithm to search in the design space constructed by the Gaussian process regression surrogate model to obtain the optimal combination of design parameters for prediction performance. S7. Based on the optimal design parameter combination, additive manufacturing and electrodeposition processes are used to prepare a high-performance amorphous nanocrystalline composite lattice structure.
[0005] Optionally, the process of preparing the amorphous alloy lattice structure matrix in S1 includes: using zirconium-based amorphous alloy powder as raw material, and forming it into a matrix with a honeycomb lattice structure, body-centered cubic structure or diamond structure configuration by laser powder bed melting technology.
[0006] Optionally, the process of sequentially depositing a nano-twinned copper coating and a nano-crystalline nickel-tungsten alloy coating on the surface of the amorphous alloy lattice structure substrate in S2 includes: firstly, electroplating a nano-twinned copper coating on the surface of the amorphous alloy lattice structure substrate, and then electroplating a nano-crystalline nickel-tungsten alloy coating on the surface of the nano-twinned copper coating.
[0007] Optionally, in the electroplating process of forming nano-twinned copper coating and nano-crystalline nickel-tungsten alloy coating in S2, auxiliary anodic electroplating technology is adopted. By inserting an anode metal wire into each unit cell of the lattice structure, the deposition thickness of the coating at local locations on the amorphous alloy lattice structure substrate can be controlled.
[0008] Optionally, the process of constructing the dataset in S4 includes: generating several sets of sample points composed of the geometric design parameters and the coating design parameters using the Latin hypercube sampling method, and outputting the mechanical performance data corresponding to each set of sample points to form an initial dataset.
[0009] Optionally, the process of training the Gaussian process regression surrogate model in S5 and iteratively updating the model using a Bayesian optimization strategy includes: selecting the Matern kernel function to construct the Gaussian process regression surrogate model; using the desired improvement function as a guide, selecting a new design point for finite element simulation in each iteration; and adding the simulation data to the training set to dynamically update the model.
[0010] Optionally, the process of using a genetic algorithm or a particle swarm optimization algorithm in S6 to search in the design space constructed by the Gaussian process regression surrogate model includes: using the trained Gaussian process regression surrogate model as the objective function evaluator, setting the population size and number of generations and then running the genetic algorithm, or setting the particle swarm size and number of iterations and then running the particle swarm optimization algorithm, and outputting the combination of geometric design parameters and coating design parameters with the best prediction performance.
[0011] Optionally, the geometric design parameters in S4 vary depending on the lattice unit cell configuration. When the unit cell configuration is a honeycomb structure, the geometric design parameters include the unit cell side length and wall thickness; when the unit cell configuration is a body-centered cubic structure or a diamond structure, the geometric design parameters include the rod diameter and the unit cell side length.
[0012] Secondly, the present invention also provides a computer terminal device, comprising: One or more processors; A memory, coupled to the processor, for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the steps of the additive manufacturing electrodeposition co-construction of amorphous nanocrystalline composite lattice structures and the machine learning-driven performance optimization method in the first aspect above.
[0013] Thirdly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, it implements the steps of the additive manufacturing electrodeposition co-construction of amorphous nanocrystalline composite lattice structures and its machine learning-driven performance optimization method in the first aspect described above.
[0014] Compared with the prior art, the present invention has the following advantages and technical effects: This invention provides an additive manufacturing method for the co-construction of amorphous and nanocrystalline composite lattice structures via electrodeposition and a machine learning-driven performance optimization method. Through the synergistic fabrication of laser powder bed melting and electrodeposition technologies, an amorphous alloy lattice structure substrate and a nanocrystalline copper and nanocrystalline nickel-tungsten alloy composite coating are integrated. While retaining the lightweight and high specific strength characteristics of the lattice structure and the high hardness and corrosion resistance of the amorphous alloy, the flexible transition of the inner copper coating alleviates local strain in the substrate, and the rigid constraint of the outer nickel-tungsten alloy coating effectively suppresses shear band instability propagation. This overcomes the inherent poor plasticity and toughness of amorphous alloys, enabling the composite lattice structure to possess both high strength and excellent plastic deformation capability. Based on this, a machine learning-driven high-throughput optimization method is introduced. By constructing a nonlinear mapping model of lattice geometry, coating structure distribution, and plastic deformation capability, combined with Bayesian optimization and global search algorithms, efficient optimization of multi-dimensional strongly correlated design parameters is achieved. This transforms the traditional inefficient R&D model relying on trial and error into a data-driven, precise design paradigm, significantly reducing experimental costs and R&D cycles. The general performance optimization paradigm formed by this collaborative manufacturing and intelligent optimization strategy can be extended to various lattice configurations such as honeycomb, body-centered cubic, and diamond, providing a customized component preparation method with ultra-lightweight, high-strength, and high-toughness characteristics for high-end fields such as aerospace precision vibration reduction structures and biomedical implants. Attached Figure Description
[0015] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 The X-ray diffraction pattern of a zirconium-based amorphous honeycomb lattice structure prepared using LPBF technology is shown in this embodiment of the invention. Figure 2 Images showing the microstructure of a Cu / Ni-W composite coating deposited in a honeycomb lattice structure according to an embodiment of the present invention. Figure 3 This is a schematic diagram of the additive manufacturing electrodeposition co-construction of amorphous nanocrystalline composite lattice structures and its machine learning-driven performance optimization method according to an embodiment of the present invention. Figure 4 The stress-strain curves of the amorphous nanocrystalline gradient composite honeycomb lattice structure with optimized parameters according to an embodiment of the present invention are shown under quasi-static compression conditions. Detailed Implementation
[0016] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.
[0017] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.
[0018] like Figure 3 As shown, this embodiment provides an additive manufacturing electrodeposition co-construction method for amorphous nanocrystalline composite lattice structures and a machine learning-driven performance optimization method therefor, including: S1. Amorphous alloy lattice structure matrix is prepared using laser powder bed melting technology; S2. Using electrodeposition technology, nano-twinned copper coating and nano-crystalline nickel-tungsten alloy coating are sequentially deposited on the surface of the amorphous alloy lattice structure substrate to form a composite coating. S3. Perform mechanical property tests on the lattice structure after deposition of composite coating to obtain mechanical property data; S4. Using the geometric design parameters of the lattice structure and the coating design parameters of the composite coating as inputs, and the mechanical performance data as outputs, construct a dataset; S5. Train a Gaussian process regression surrogate model based on the dataset, and iteratively update the model using a Bayesian optimization strategy; S6. Use a genetic algorithm or particle swarm optimization algorithm to search in the design space constructed by the Gaussian process regression surrogate model to obtain the optimal combination of design parameters for prediction performance. S7. Based on the optimal design parameter combination, additive manufacturing and electrodeposition processes are used to prepare a high-performance amorphous nanocrystalline composite lattice structure.
[0019] Furthermore, the process of preparing the amorphous alloy lattice structure matrix in S1 includes: using zirconium-based amorphous alloy powder as raw material, and forming it into a matrix with a honeycomb lattice structure, body-centered cubic structure or diamond structure configuration by laser powder bed melting technology.
[0020] Furthermore, the process of sequentially depositing a nano-twinned copper coating and a nano-crystalline nickel-tungsten alloy coating on the surface of the amorphous alloy lattice structure substrate in S2 includes: firstly, electroplating a nano-twinned copper coating on the surface of the amorphous alloy lattice structure substrate, and then electroplating a nano-crystalline nickel-tungsten alloy coating on the surface of the nano-twinned copper coating.
[0021] Furthermore, in the electroplating process of forming nano-twinned copper coating and nano-crystalline nickel-tungsten alloy coating in S2, auxiliary anodic electroplating technology is adopted. By inserting an anode metal wire into each unit cell of the lattice structure, the deposition thickness of the coating at local positions on the amorphous alloy lattice structure substrate can be controlled.
[0022] Furthermore, the process of constructing the dataset in S4 includes: using the Latin hypercube sampling method to generate several sets of sample points composed of the geometric design parameters and the coating design parameters, and using the mechanical performance data corresponding to each set of sample points as output to form the initial dataset.
[0023] Furthermore, the process of training the Gaussian process regression surrogate model in S5 and iteratively updating the model using a Bayesian optimization strategy includes: selecting the Matern kernel function to construct the Gaussian process regression surrogate model; using the desired improvement function as a guide, selecting a new design point for finite element simulation in each iteration; and adding the simulation data to the training set to dynamically update the model.
[0024] Furthermore, the process of searching in the design space constructed by the Gaussian process regression surrogate model using genetic algorithm or particle swarm optimization algorithm in S6 includes: using the trained Gaussian process regression surrogate model as the objective function evaluator, setting the population size and number of generations and then running the genetic algorithm, or setting the particle swarm size and number of iterations and then running the particle swarm optimization algorithm, and outputting the combination of geometric design parameters and coating design parameters with the best prediction performance.
[0025] Furthermore, the geometric design parameters in S4 vary depending on the lattice unit cell configuration. When the unit cell configuration is a honeycomb structure, the geometric design parameters include the unit cell side length and wall thickness; when the unit cell configuration is a body-centered cubic structure or a diamond structure, the geometric design parameters include the rod diameter and the unit cell side length.
[0026] In this embodiment, a computer terminal device is provided, including: One or more processors; A memory, coupled to the processor, for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the steps of the above-described additive manufacturing electrodeposition co-construction of amorphous nanocrystalline composite lattice structures and its machine learning-driven performance optimization method.
[0027] In this embodiment, a computer-readable storage medium is also provided, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the above-described additive manufacturing electrodeposition co-construction of amorphous nanocrystalline composite lattice structures and its machine learning-driven performance optimization method.
[0028] Example 1 The following are the specific steps for forming a zirconium-based amorphous alloy honeycomb lattice structure using LPBF technology (EOS M100 equipment): Zr... 41.2 Ti 13.8 Cu 12.5 Ni 10 Be22.5 Amorphous alloy powder was placed in a vacuum drying oven. After the pressure inside the oven was reduced to -0.1 MPa by a mechanical pump, it was kept at 60-80℃ for 4-6 hours to thoroughly remove moisture and prevent oxidation. A Ti6Al4V (TC4) substrate, which exhibits excellent bonding with zirconium-based amorphous alloy systems, was used. The substrate preheating temperature was approximately 350 K. Laser forming process parameters were set as follows: laser power 80 W, scanning rate 1600 mm / s, and scanning spacing 0.1 mm. A honeycomb lattice structure with a wall thickness of 0.2 mm, a unit cell side length of 2 mm, and an overall shape of 18 mm × 10 mm × 8 mm was fabricated. Its X-ray diffraction pattern is shown below. Figure 1 As shown. Mechanical property tests were performed on the zirconium-based amorphous alloy honeycomb lattice structure, and its stress-strain curves are shown below. Figure 4 As shown, the specific strength is 4.9 × 10⁻⁶. 5 The strength is Nm / kg, the compressive strength is 3.64 MPa, and the fracture strain is 7.8%.
[0029] A constant-temperature heated magnetic stirrer and a DC regulated power supply were used to perform electrodeposition surface treatment on the honeycomb lattice structure. Before electrodeposition, a pretreatment procedure was required for the amorphous alloy lattice structure substrate, with the following steps: The sample was ultrasonically cleaned in acetone for 5 minutes at 20°C, followed by rinsing with anhydrous ethanol and deionized water. The amorphous alloy lattice structure substrate was then immersed in an alkaline solution composed of 20 g / L sodium hydroxide (NaOH), 30 g / L sodium bicarbonate (Na2HCO3), and 20 g / L sodium phosphate (Na3PO4) at 70°C for 30 minutes, followed by rinsing with deionized water. Finally, the amorphous alloy lattice structure substrate was immersed in 10% dilute sulfuric acid (H2SO4) at 20°C for 5 minutes for acid pickling and activation, followed by rinsing with deionized water for 2 minutes to complete the pretreatment before electroplating. A nano-twinned copper metal coating was deposited using an acidic bright copper plating bath. The pure Cu coating solution bath consisted of copper sulfate pentahydrate (CuSO4·5H2O), sulfuric acid (H2SO4, >98% purity), and chloride ions (Cl). - Composition: Prepared using analytical grade reagents and deionized water, consisting of 110 g / L CuSO4·5H2O, 90 g / L H2SO4, and 65 ppm Cl-. -The electroplating solution (pH≈3.0) was prepared using phosphorus copper containing 0.3% phosphorus as the anode. Specific electroplating parameters were: temperature 25℃, electroplating voltage 0.3V, electroplating time 50min, and stirring speed 280rpm. A nanotwinned copper metal coating with an average thickness of 27.2μm was finally obtained. A nickel-tungsten metal coating was deposited using a nickel-tungsten alloy plating bath composed of nickel sulfate (NiSO4·6H2O), sodium tungstate (Na2WO42H2O), sodium citrate (Na3C6H5O72H2O), and boric acid (H3BO3). An electroplating solution (pH≈3.0) consisting of 100 g / L nickel sulfate, 30 g / L sodium tungstate, 60 g / L sodium citrate, and 30 g / L boric acid was prepared using analytical grade reagents and deionized water. Specific electroplating parameters were: temperature 55℃, current density, electroplating voltage 2.4V, electroplating time 60min, and stirring speed 280rpm. A nickel-tungsten metal coating with an average thickness of 30.5 μm was finally obtained, and the composite coating ranged from 50 to 65 μm. The microstructure morphology of the deposited Cu / Ni-W composite coating is shown in the image below. Figure 2 As shown, the amorphous / nanocrystalline composite honeycomb structure was mechanically tested, and its compressive strength was 11.2 MPa and its fracture strain was 14.1%.
[0030] Machine learning methods were used to co-optimize the geometric and coating parameters of zirconium-based amorphous alloy honeycomb lattice structures to obtain optimal mechanical properties. The specific steps are as follows: Design variables and their value ranges for the honeycomb lattice structure were defined. Geometric design parameters included the cell side length L (1.0-3.5 mm) and wall thickness T (0.1-0.5 mm); coating design parameters included the thickness of the inner nanotwinned copper layer (10-30 μm), the thickness of the outer nanocrystalline nickel-tungsten alloy layer (10-30 μm), and the linear gradient decay coefficient G (0.5-2.0, where G>1 indicates coating concentration towards the nodes) along the composite coating thickness at the nodes; the optimization objectives were compressive strength and fracture strain; and 35 initial sample points were generated in the aforementioned five-dimensional design space using Latin hypercube sampling.
[0031] A finite element model was established for 35 initial sample points, and quasi-static axial compression simulation was performed to output the corresponding stress-strain curves, thus forming the initial dataset for machine learning. A Gaussian process regression surrogate model was constructed using Gaussian process regression, and the Matern kernel function was selected. The initial dataset was divided into training and test sets in a 4:1 ratio for model training. The prediction determination coefficient R0 of the initial model on the test set was calculated. 2The accuracy was 0.87. To improve model accuracy more efficiently, a Bayesian optimization framework was introduced for active learning. Guided by the expected improvement function, after 5 iterations, 4 new design points with the greatest expected improvement were selected in each iteration for finite element simulation, and new data were added to the training set to dynamically update the model. After 5 rounds of active learning, the total sample size reached 55, at which point the prediction accuracy R of the Gaussian process regression surrogate model was [value missing]. 2 With a value of 0.93, it possesses high-precision performance prediction capabilities.
[0032] A genetic algorithm was used to perform a global search in the design space constructed by the Gaussian process regression surrogate model. The algorithm was set to a population size of 150 and evolved for 300 generations. The optimal combination of design parameters was found to be: unit cell side length L = 2.5 mm, wall thickness T = 0.2 mm, nanocrystalline copper coating thickness of 28.7 μm, nanocrystalline nickel-tungsten alloy coating thickness of 25.4 μm, and gradient coefficient G = 1.6. This combination predicted a compressive strength of 11.8 MPa and a fracture strain of 15.2%.
[0033] Using the predicted optimal parameters, an amorphous / nanocrystalline composite honeycomb lattice structure was re-fabricated. The additive manufacturing process parameters remained consistent with those in Example 1, with the following adjustments to the electrodeposition process parameters: the voltage for the nanotwinned copper coating was adjusted to 0.35V, and the voltage for the nanocrystalline nickel-tungsten alloy coating was adjusted to 2.3V. An auxiliary anode technique was used to control the coating gradient distribution at G=1.6. Quasi-mechanical tests were performed on the amorphous / nanocrystalline composite honeycomb structure, and its stress-strain curve is shown below. Figure 4 As shown, the compressive strength is 12.1 MPa and the fracture strain is 15.8%, which is about 8.0% higher than the original sample in terms of compressive strength and about 12.1% higher in terms of fracture strain.
[0034] Example 2 The body-centered cubic lattice structure of zirconium-based amorphous alloy was formed using LPBF technology, with Zr powder as the powder. 53 Cu 19 Ni 10 Al 14 Y4Ti 32.8 The substrate was TC4, preheated to 350K. Key process parameters: laser power 100W, scanning rate 1600 mm / s, scanning spacing 0.09 mm. Samples with a rod diameter of 0.3 mm, a unit cell side length of 2 mm, and an external dimension of 10 mm × 10 mm × 10 mm were fabricated. Mechanical properties of the zirconium-based amorphous alloy body-centered cubic lattice structure were tested, showing a compressive strength of 18.4 MPa and a fracture strain of 6.8%.
[0035] Electrodeposition was performed on the body-centered cubic structure using the same pretreatment process and plating solution composition as in Example 1. The deposition parameters for the nanotwinned copper coating were: temperature 25°C, plating voltage 0.4V, plating time 50 min, and stirring speed 280 rpm. A nanotwinned copper metal coating with an average thickness of 26.4 μm was obtained. The deposition parameters for the nickel-tungsten alloy coating were: temperature 55°C, plating voltage 2.5V, plating time 60 min, and stirring speed 280 rpm. A nanotwinned copper metal coating with an average thickness of 24.2 μm was obtained. Mechanical testing was performed on the amorphous / nanocrystalline composite body-centered cubic structure, revealing a compressive strength of 25.6 MPa and a fracture strain of 12.2%.
[0036] Optimization was performed using the same machine learning framework as in Example 1. Geometric design parameters included rod diameter d (0.3-0.7 mm) and unit cell side length a (1.5-3.0 mm); coating design parameters and optimization objectives were the same as in Example 1. The optimal design parameter combination was calculated to be: rod diameter d = 0.45 mm, unit cell side length a = 2.2 mm, nanotwinned copper coating thickness of 26.6 μm, nanocrystalline nickel-tungsten alloy coating thickness of 24.3 μm, and gradient coefficient G = 1.2. This combination predicted a compressive strength of 29.2 MPa and a fracture strain of 12.6%.
[0037] Using the predicted optimal parameters, an amorphous / nanocrystalline composite honeycomb lattice structure was re-prepared. The additive manufacturing process parameters remained consistent with those of Example 1, with the following adjustments to the electrodeposition process parameters: the voltage for the nanotwinned copper coating was adjusted to 0.38V, and the voltage for the nanocrystalline nickel-tungsten alloy coating was adjusted to 2.45V. An auxiliary anode technique was used to control the coating gradient distribution at G=1.2. Quasi-mechanical tests were performed on the amorphous / nanocrystalline composite body-centered cubic structure, revealing a compressive strength of 28.9 MPa and a fracture strain of 13.9%, representing an increase of approximately 12.9% in compressive strength and approximately 13.8% in fracture strain compared to the unoptimized sample.
[0038] Example 3 The body-centered cubic lattice structure of zirconium-based amorphous alloy was formed using LPBF technology, with Zr powder as the powder. 52.5 Al 10 Cu 15 Ni 10 Be 12.5 The substrate was TC4, preheated to 350K. Key process parameters: laser power 100W, scanning rate 1100 mm / s, scanning spacing 0.1 mm. Samples with a rod diameter of 0.5 mm, a unit cell side length of 2 mm, and an external dimension of 10 mm × 10 mm × 10 mm were fabricated. Mechanical properties of the zirconium-based amorphous alloy diamond lattice structure were tested, showing a compressive strength of 19.6 MPa and a fracture strain of 8.5%.
[0039] The diamond structure was electrodeposited using the same pretreatment process and plating solution composition as in Example 1. The deposition parameters for the nanotwinned copper coating were: temperature 25°C, plating voltage 0.45V, plating time 50 min, and stirring speed 280 rpm. A nanotwinned copper metal coating with an average thickness of 24.9 μm was obtained. The deposition parameters for the nickel-tungsten alloy coating were: temperature 55°C, plating voltage 2.6V, plating time 60 min, and stirring speed 280 rpm. A nanotwinned copper metal coating with an average thickness of 24.4 μm was obtained. Mechanical testing of the amorphous / nanocrystalline composite diamond structure showed a compressive strength of 30.2 MPa and a fracture strain of 14.5%.
[0040] The same machine learning framework as in Example 1 was used for optimization. Geometric design parameters included rod diameter d (0.3-0.7 mm) and unit cell side length a (1.5-3.5 mm); coating design parameters and optimization objectives were the same as in Example 1. Optimization calculations yielded the following optimal design parameter combination: rod diameter d = 0.6 mm, unit cell side length a = 2.5 mm, nanotwinned copper coating thickness of 24.5 μm, nanocrystalline nickel-tungsten alloy coating thickness of 22.7 μm, and gradient coefficient G = 1.4. This combination predicted a compressive strength of 31.2 MPa and a fracture strain of 15.4%.
[0041] Using the predicted optimal parameters, an amorphous / nanocrystalline composite honeycomb lattice structure was re-prepared. The additive manufacturing process parameters remained consistent with those of Example 1, with the following adjustments to the electrodeposition process parameters: the voltage for the nanotwinned copper coating was adjusted to 0.4V, and the voltage for the nanocrystalline nickel-tungsten alloy coating was adjusted to 2.55V. An auxiliary anode technique was used to control the coating gradient distribution at G=1.4. Quasi-mechanical tests were performed on the amorphous / nanocrystalline composite diamond structure, revealing a compressive strength of 32.4 MPa and a fracture strain of 15.7%, representing an increase of approximately 7.3% in compressive strength and approximately 8.3% in fracture strain compared to the unoptimized sample.
[0042] This invention provides an additive manufacturing method for the co-construction of amorphous and nanocrystalline composite lattice structures via electrodeposition and a machine learning-driven performance optimization method. Through the synergistic fabrication of laser powder bed melting and electrodeposition technologies, an amorphous alloy lattice structure substrate and a nanocrystalline copper and nanocrystalline nickel-tungsten alloy composite coating are integrated. While retaining the lightweight and high specific strength characteristics of the lattice structure and the high hardness and corrosion resistance of the amorphous alloy, the flexible transition of the inner copper coating alleviates local strain in the substrate, and the rigid constraint of the outer nickel-tungsten alloy coating effectively suppresses shear band instability propagation. This overcomes the inherent poor plasticity and toughness of amorphous alloys, enabling the composite lattice structure to possess both high strength and excellent plastic deformation capability. Based on this, a machine learning-driven high-throughput optimization method is introduced. By constructing a nonlinear mapping model of lattice geometry, coating structure distribution, and plastic deformation capability, combined with Bayesian optimization and global search algorithms, efficient optimization of multi-dimensional strongly correlated design parameters is achieved. This transforms the traditional inefficient R&D model relying on trial and error into a data-driven, precise design paradigm, significantly reducing experimental costs and R&D cycles. The general performance optimization paradigm formed by this collaborative manufacturing and intelligent optimization strategy can be extended to various lattice configurations such as honeycomb, body-centered cubic, and diamond, providing a customized component preparation method with ultra-lightweight, high-strength, and high-toughness characteristics for high-end fields such as aerospace precision vibration reduction structures and biomedical implants.
[0043] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.
Claims
1. A method for additive manufacturing electrodeposition to co-construct amorphous nanocrystalline composite lattice structures and its machine learning-driven performance optimization, characterized in that, Includes the following steps: S1. Amorphous alloy lattice structure matrix is prepared using laser powder bed melting technology; S2. Using electrodeposition technology, nano-twinned copper coating and nano-crystalline nickel-tungsten alloy coating are sequentially deposited on the surface of the amorphous alloy lattice structure substrate to form a composite coating. S3. Perform mechanical property tests on the lattice structure after deposition of composite coating to obtain mechanical property data; S4. Using the geometric design parameters of the lattice structure and the coating design parameters of the composite coating as inputs, and the mechanical performance data as outputs, construct a dataset; S5. Train a Gaussian process regression surrogate model based on the dataset, and iteratively update the model using a Bayesian optimization strategy; S6. Use a genetic algorithm or particle swarm optimization algorithm to search in the design space constructed by the Gaussian process regression surrogate model to obtain the optimal combination of design parameters for prediction performance. S7. Based on the optimal combination of design parameters predicted for the best performance, additive manufacturing and electrodeposition processes are used to prepare amorphous nanocrystalline composite lattice structures with optimized performance.
2. The method according to claim 1, characterized in that, The process of preparing the amorphous alloy lattice structure matrix in S1 includes: using zirconium-based amorphous alloy powder as raw material, and forming it into a matrix with a honeycomb lattice structure, body-centered cubic structure or diamond structure configuration by laser powder bed melting technology.
3. The method according to claim 2, characterized in that, The process of sequentially depositing a nano-twinned copper coating and a nano-crystalline nickel-tungsten alloy coating on the surface of the amorphous alloy lattice structure substrate in S2 includes: firstly, electroplating a nano-twinned copper coating on the surface of the amorphous alloy lattice structure substrate, and then electroplating a nano-crystalline nickel-tungsten alloy coating on the surface of the nano-twinned copper coating.
4. The method according to claim 3, characterized in that, In the electroplating process of forming nano-twinned copper coating and nano-crystalline nickel-tungsten alloy coating in S2, auxiliary anodic electroplating technology is adopted. By inserting an anode metal wire into each unit cell of the lattice structure, the deposition thickness of the coating at local positions on the amorphous alloy lattice structure substrate can be controlled.
5. The method according to claim 1, characterized in that, The process of constructing the dataset in S4 includes: using the Latin hypercube sampling method to generate several sets of sample points composed of the geometric design parameters and the coating design parameters, and using the mechanical performance data corresponding to each set of sample points as output to form the initial dataset.
6. The method according to claim 1, characterized in that, The process of training a Gaussian process regression surrogate model in S5 and iteratively updating the model using a Bayesian optimization strategy includes: selecting the Matern kernel function to construct the Gaussian process regression surrogate model; using the desired improvement function as a guide, selecting a new design point for finite element simulation in each iteration; and adding the simulation data to the training set to dynamically update the model.
7. The method according to claim 1, characterized in that, The process of searching in the design space constructed by the Gaussian process regression surrogate model using genetic algorithm or particle swarm optimization algorithm in S6 includes: using the trained Gaussian process regression surrogate model as the objective function evaluator, setting the population size and number of generations and then running the genetic algorithm, or setting the particle swarm size and number of iterations and then running the particle swarm optimization algorithm, and outputting the combination of geometric design parameters and coating design parameters with the best prediction performance.
8. The method according to claim 1, characterized in that, The geometric design parameters in S4 vary depending on the lattice unit cell configuration. When the unit cell configuration is a honeycomb structure, the geometric design parameters include the unit cell side length and wall thickness. When the unit cell configuration is a body-centered cubic structure or a diamond structure, the geometric design parameters include the rod diameter and the unit cell side length.
9. A computer terminal device, characterized in that, include: One or more processors; A memory, coupled to the processor, for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors perform the steps of the method as described in any one of claims 1-8.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1-8.