An aluminum-based surface multilayer interface enhancement treatment method and system

Abstract

The present application relates to the technical field of material processing, and in particular to a kind of aluminum-based surface multilayer interface enhancement processing method and system, by collecting aluminum-based substrate surface topography scanning data and matrix material attribute data, utilize multi-task learning network to extract surface concave-convex feature and active site distribution and generate space distribution model representing interface bonding capacity, based on the model and matrix attribute data, predict interface stress evolution trend and peeling risk distribution, accordingly generate adaptive layer construction strategy to dynamically optimize each functional interface layer thickness distribution and transition layer component gradient, finally implement multilayer interface enhancement processing and measure bonding strength, effectively reduce interface stress concentration degree and improve interface bonding strength.

Description

Technical Field

[0001] This invention relates to the field of material processing technology, and in particular to a method and system for multilayer interface enhancement treatment of aluminum-based surfaces. Background Technology

[0002] Aluminum-based materials are widely used in aerospace, automotive manufacturing, and electronic packaging due to their lightweight and high specific strength. However, their strong surface chemical inertness and insufficient interfacial adhesion with coatings or functional layers often lead to cracking or peeling failure in multilayer interfacial systems during use. Existing conventional processing methods mainly rely on surface pretreatments such as mechanical grinding, chemical etching, or anodizing to improve mechanical bonding by increasing surface roughness or forming a porous oxide layer. Subsequently, a primer or transition layer is applied to improve chemical compatibility. These methods typically employ empirical process flows, such as first fixing the grinding grit and then setting the chemical treatment time, followed by depositing functional materials layer by layer at a fixed thickness.

[0003] Conventional methods have significant drawbacks. The correlation between surface treatment parameters and interfacial bonding performance lacks quantitative analysis, and mechanical anchoring effects and chemical bonding activity are assessed independently, making synergistic optimization difficult. For example, while increased roughness can enhance mechanical bonding, it may lead to localized stress concentration, while uneven distribution of chemically active sites weakens molecular-level connections between coatings. Existing technologies cannot simultaneously characterize the combined impact of these two factors on interfacial bonding capabilities on a spatial scale. Therefore, existing methods face bottlenecks in addressing the need for multilayer interfacial reinforcement of aluminum substrates under complex conditions, exhibiting poor synergy and insufficient adaptability. A treatment scheme capable of global optimization based on surface microstructures and material properties is urgently needed. Summary of the Invention

[0004] This invention provides a method and system for multilayer interface enhancement treatment of aluminum-based surfaces, which can solve the problems in the prior art.

[0005] A first aspect of the present invention provides a method for multilayer interface enhancement treatment of an aluminum-based surface, comprising: Collect surface morphology scanning data and matrix material property data of aluminum substrate; The surface morphology scanning data is collaboratively feature extracted by a multi-task learning network, and the surface unevenness features that affect mechanical anchoring and the distribution of active sites that affect chemical bonding are identified simultaneously. A spatial distribution model characterizing the interfacial bonding ability is generated. The spatial distribution model includes the mechanical anchoring strength value and chemical bonding activity value at each surface location. Based on the spatial distribution model and the matrix material property data, the interfacial stress evolution trend of the multilayer interface system is predicted and the distribution of interfacial delamination risk is calculated. An adaptive layer construction strategy is generated based on the interface peeling risk distribution. The adaptive layer construction strategy dynamically optimizes the thickness distribution of each functional interface layer and the component gradient design of the interface transition layer, so as to minimize the stress concentration of the multi-layer interface system. According to the adaptive layer construction strategy, a multi-layer interface enhancement treatment is performed on the surface of the aluminum substrate, and the interfacial bonding strength of the multi-layer interface system after the treatment is measured.

[0006] The surface morphology scanning data is collaboratively feature-extracted using a multi-task learning network. This simultaneously identifies surface irregularities affecting mechanical anchoring and the distribution of active sites influencing chemical bonding, generating a spatial distribution model characterizing interfacial bonding capabilities. A dual-encoder architecture for a multi-task learning network is constructed, comprising a geometric topology encoder and an energy potential field encoder; In the geometric topology encoder, based on the surface topography scanning data, the Gaussian curvature field and the average curvature field of the surface are calculated, the saddle point region and peak-valley region of the surface are identified according to the Gaussian curvature field, and the depression depth gradient of each peak-valley region is calculated according to the average curvature field. The effective containment volume of each peak and valley region is obtained by spatial integration of the depression depth gradient. The curvature direction of the saddle point region is analyzed to obtain the principal curvature direction and the secondary curvature direction. The stress concentration direction of the surface is identified based on the angle distribution between the principal and secondary curvature directions. In the energy potential field encoder, based on the surface topography scanning data, the local curvature variation coefficient at each position on the surface is calculated and geometric abrupt change points are marked. Spatial clustering is performed on the geometric abrupt change points to identify the clustering regions of the geometric abrupt change points. The clustering regions correspond to the grain boundary network distribution of the aluminum substrate. The effective containment volume, the stress concentration direction, and the grain boundary network distribution are weighted and fused to generate mechanical anchoring strength and chemical bonding activity values, which are then registered in a three-dimensional spatial coordinate system to generate a spatial distribution model.

[0007] Based on the surface morphology scanning data, the local curvature variation coefficient at each location on the surface is calculated and geometric abrupt change points are marked. Spatial clustering is performed on these geometric abrupt change points to identify clustered regions. These clustered regions correspond to the grain boundary network distribution of the aluminum substrate, including: Based on the surface morphology scanning data, the surface normal vector of each sampling point is calculated, and the position where the angle between the normal vectors of adjacent sampling points exceeds the lattice orientation tolerance angle is marked as the orientation abrupt change region. A curvature profile is established along the direction of normal vector change within the orientation abrupt change region, and curvature extrema points on the profile are extracted. These curvature extrema points correspond to surface ridges and grooves at grain boundaries. The distribution density of curvature extrema points in the neighborhood of each sampling point is statistically analyzed, and the location where the density exceeds the grain boundary characteristic density threshold is taken as the peak center of the local curvature variation coefficient and marked as the geometric abrupt change point. Extract the three-dimensional coordinates of the geometric abrupt change points, perform ray tracing along the principal curvature direction of each point, and connect the geometric abrupt change points where the rays intersect to form a spatial chain, which serves as the linear skeleton of the aggregation region; Identify the triangular nodes and closed loops in the linear skeleton. The triangular nodes correspond to the intersection of three grains, and the closed loops correspond to the boundaries of a single grain. Reconstruct the grain boundary network distribution of the aluminum substrate based on the spatial distribution relationship of the triangular nodes and closed loops.

[0008] Based on the spatial distribution model and the matrix material property data, the evolution trend of interfacial stress in the multilayer interface system is predicted and the distribution of interfacial delamination risk is calculated, including: The mechanical anchoring strength and chemical bonding activity values ​​of each surface location are extracted from the spatial distribution model, and the elastic modulus and thermal expansion coefficient of the aluminum substrate are extracted from the matrix material property data. Calculate the spatial gradient modulus of the mechanical anchoring strength value, mark the position where the spatial gradient modulus exceeds the bearing change threshold as the stress accumulation zone, and in the stress accumulation zone, multiply the chemical bonding activity value with the elastic modulus of the aluminum substrate to obtain the interface stiffness matching coefficient, and mark the position where the interface stiffness matching coefficient is lower than the coordination critical coefficient as the stiffness mismatch zone. A time-domain evolution sequence of interface stress is established within the stiffness mismatch region. The time-domain evolution sequence records the changes of interface normal stress and interface tangential stress with temperature cycling. The time and spatial location of stress peaks in the time-domain evolution sequence are extracted. At the stress peak location, the ratio of the interface normal stress to the chemical bonding activity value is calculated as the bonding failure tendency factor. The cumulative number of times the bonding failure tendency factor exceeds the failure initiation factor during temperature cycling at each location is counted, and a normalized mapping is performed in the three-dimensional spatial coordinate system to generate the interface peeling risk distribution.

[0009] An adaptive layer construction strategy is generated based on the interface peeling risk distribution. This strategy dynamically optimizes the thickness distribution of each functional interface layer and the component gradient design of the interface transition layer to minimize the stress concentration of the multi-layer interface system. This includes: The spatial distribution of cumulative occurrences is extracted from the risk distribution of the interface peeling, and the spatial gradient of the cumulative occurrences is calculated to obtain the risk propagation direction field. The risk propagation direction field characterizes the propagation path of interface peeling damage in the multi-layer interface system. A stress transmission chain is established along the risk propagation direction field. The stress transmission chain connects the locations of adjacent target accumulation times in space. The stress transmission efficiency at each node on the stress transmission chain is calculated. The stress transmission efficiency is equal to the weighted sum of the mechanical anchoring strength value and the chemical bonding activity value at the node location. The nodes in the stress transmission chain whose stress transmission efficiency is lower than the transmission blocking threshold are identified and marked as stress stagnation points. An interface transition layer is inserted at the stress stagnation point, and the composition gradient of the interface transition layer is determined by the difference in chemical bonding activity values ​​between adjacent nodes. Extract the start and end nodes of the stress transmission chain, establish a thickness distribution optimization model between the start and end nodes, minimize the stress gradient integral value on the stress transmission chain path as the objective function, and solve the thickness distribution optimization model to obtain the thickness distribution scheme of each functional interface layer. The thickness allocation scheme is registered with the component gradient of the interface transition layer in a three-dimensional coordinate system to generate an adaptive layer construction strategy.

[0010] A thickness distribution optimization model is established between the start and end nodes, with the objective function being to minimize the stress gradient integral value along the stress transmission chain path. Solving the thickness distribution optimization model yields the thickness distribution schemes for each functional interface layer, including: All intermediate nodes between the starting node and the ending node on the stress transmission chain are extracted, and a local stress tensor field is constructed. Tensor decomposition is performed based on the mechanical anchoring strength value and chemical bonding activity value at the node position to obtain the stress tensor distribution sequence on the stress transmission chain path. Gradient calculation is performed on the stress tensor distribution sequence along the path direction, and the spatial rate of change of the principal component of the stress tensor is extracted as the stress gradient field. The position where the gradient magnitude exceeds the gradient jump threshold in the stress gradient field is identified and marked as the stress transmission inflection point. Each stress transmission inflection point is used as the boundary marker of the functional interface layer. The stress gradient field is integrated within the path segment between adjacent stress transmission inflection points. The path integral value represents the degree of stress concentration accumulation within the corresponding functional interface layer. A thickness allocation optimization model is established, with the thickness of each functional interface layer as the decision variable and the objective function being the weighted sum of the path integral values ​​of all functional interface layers. The thickness allocation optimization model is solved by gradient descent iterative method. In each iteration, the thickness allocation is adjusted according to the partial derivative of the objective function with respect to the thickness of each functional interface layer. The iteration is terminated when the change in the objective function value between two adjacent iterations is lower than the convergence criterion, and the thickness allocation scheme of each functional interface layer is output.

[0011] According to the adaptive layer construction strategy, a multi-layer interface enhancement treatment is performed on the surface of the aluminum substrate, and the interfacial bonding strength of the multi-layer interface system after the treatment is measured, including: Based on the thickness allocation scheme of each functional interface layer and the component gradient of the interface transition layer in the adaptive layer construction strategy, the spatial coordinate positioning information and deposition sequence information of each functional interface layer on the aluminum substrate surface are extracted to generate the process execution sequence of multi-layer interface enhancement treatment. According to the process execution sequence, each functional interface layer is deposited sequentially on the surface of the aluminum substrate, and an interface transition layer is deposited between adjacent functional interface layers. The composition of the interface transition layer is continuously adjusted along the thickness direction according to the composition gradient, so that the composition of the interface transition layer transitions from the main component of the previous functional interface layer to the main component of the next functional interface layer. The interfacial bonding strength of the deposited multi-layer interface system was measured. The interfacial bonding strength was obtained by applying a normal tensile load to the surface of the multi-layer interface system and recording the critical load value when the interface peels off. The critical load value was normalized to the unit interface area to obtain the interfacial bonding strength.

[0012] A second aspect of the present invention provides a multilayer interface enhancement treatment system for aluminum-based surfaces, comprising: The data acquisition unit is used to acquire surface morphology scanning data and matrix material property data of aluminum substrate; The spatial distribution modeling unit is used to perform collaborative feature extraction on the surface morphology scanning data through a multi-task learning network, simultaneously identify the surface unevenness features that affect mechanical anchoring and the distribution of active sites that affect chemical bonding, and generate a spatial distribution model that characterizes the interfacial bonding ability. The spatial distribution model includes the mechanical anchoring strength value and chemical bonding activity value at each surface location. The risk distribution prediction unit is used to predict the interfacial stress evolution trend of the multilayer interface system and calculate the interfacial delamination risk distribution based on the spatial distribution model and the matrix material property data. An adaptive layer strategy construction unit is used to generate an adaptive layer construction strategy based on the interface stripping risk distribution. The adaptive layer construction strategy dynamically optimizes the thickness distribution of each functional interface layer and the component gradient design of the interface transition layer to minimize the stress concentration of the multi-layer interface system. An execution unit is used to perform multi-layer interface enhancement treatment on the surface of the aluminum substrate according to the adaptive layer construction strategy, and to measure the interfacial bonding strength of the multi-layer interface system after the treatment is completed.

[0013] A third aspect of the present invention provides an electronic device, comprising: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0014] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0015] This invention achieves dual enhancement of mechanical anchoring and chemical bonding by synergistically extracting surface unevenness features and active site distribution, effectively improving the overall load-bearing capacity of the interface. Accurate prediction of stress evolution trends reduces the peak value of interlayer shear stress at the interface, enhancing long-term service reliability.

[0016] The establishment of the spatial distribution model enables the quantitative characterization of surface heterogeneity. The mechanical anchoring strength and chemical bonding activity at each location are accurately labeled. The multi-task learning network processes morphological and chemical information simultaneously, avoiding the limitations of single feature analysis and effectively improving the accuracy of identifying weak areas at the interface. Based on the layer construction strategy generated by this model, the thickness allocation of the functional interface layer and the gradient of the transition layer components are optimally matched.

[0017] The adaptive layer construction strategy dynamically optimizes the thickness distribution of each interface layer, transferring stress concentration areas and dispersing them over a wider range. The transition layer composition gradient design matches the difference in thermal expansion coefficients between the aluminum substrate and the functional layers. The measured interfacial bonding strength after treatment directly verifies the effectiveness of the strategy, and the process stability is significantly improved. This method achieves closed-loop control throughout the entire process from data acquisition to final performance verification, eliminating redundancy and deficiencies in empirical design. The interface failure mode of the multi-layer interface system changes from interface peeling to matrix ductile fracture, resulting in a leap in reliability. Attached Figure Description

[0018] Figure 1 This is a schematic flowchart of a multilayer interface enhancement treatment method for aluminum-based surfaces according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating the method for generating an adaptive layer construction strategy according to an embodiment of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0021] Figure 1 This is a schematic flowchart of a multilayer interface enhancement treatment method for an aluminum-based surface according to an embodiment of the present invention. The embodiment of the present invention provides a multilayer interface enhancement treatment method for an aluminum-based surface, comprising: Collect surface morphology scanning data and matrix material property data of aluminum substrate; The surface morphology scanning data is collaboratively feature extracted by a multi-task learning network, and the surface unevenness features that affect mechanical anchoring and the distribution of active sites that affect chemical bonding are identified simultaneously. A spatial distribution model characterizing the interfacial bonding ability is generated. The spatial distribution model includes the mechanical anchoring strength value and chemical bonding activity value at each surface location. Based on the spatial distribution model and the matrix material property data, the interfacial stress evolution trend of the multilayer interface system is predicted and the distribution of interfacial delamination risk is calculated. An adaptive layer construction strategy is generated based on the interface peeling risk distribution. The adaptive layer construction strategy dynamically optimizes the thickness distribution of each functional interface layer and the component gradient design of the interface transition layer, so as to minimize the stress concentration of the multi-layer interface system. According to the adaptive layer construction strategy, a multi-layer interface enhancement treatment is performed on the surface of the aluminum substrate, and the interfacial bonding strength of the multi-layer interface system after the treatment is measured.

[0022] The surface morphology scanning data is collaboratively feature-extracted using a multi-task learning network. This simultaneously identifies surface irregularities affecting mechanical anchoring and the distribution of active sites influencing chemical bonding, generating a spatial distribution model characterizing interfacial bonding capabilities. A dual-encoder architecture for a multi-task learning network is constructed, comprising a geometric topology encoder and an energy potential field encoder; In the geometric topology encoder, based on the surface topography scanning data, the Gaussian curvature field and the average curvature field of the surface are calculated, the saddle point region and peak-valley region of the surface are identified according to the Gaussian curvature field, and the depression depth gradient of each peak-valley region is calculated according to the average curvature field. The effective containment volume of each peak and valley region is obtained by spatial integration of the depression depth gradient. The curvature direction of the saddle point region is analyzed to obtain the principal curvature direction and the secondary curvature direction. The stress concentration direction of the surface is identified based on the angle distribution between the principal and secondary curvature directions. In the energy potential field encoder, based on the surface topography scanning data, the local curvature variation coefficient at each position on the surface is calculated and geometric abrupt change points are marked. Spatial clustering is performed on the geometric abrupt change points to identify the clustering regions of the geometric abrupt change points. The clustering regions correspond to the grain boundary network distribution of the aluminum substrate. The effective containment volume, the stress concentration direction, and the grain boundary network distribution are weighted and fused to generate mechanical anchoring strength and chemical bonding activity values, which are then registered in a three-dimensional spatial coordinate system to generate a spatial distribution model.

[0023] The dual-encoder architecture of the multi-task learning network consists of two parallel feature extraction pathways: a geometric topology encoder and an energy potential field encoder. Both share the underlying convolutional feature mapping layer but maintain independent task-specific branches in higher-level feature representation. The geometric topology encoder extracts 3D topology information directly related to mechanical anchoring from surface topography scanning data, while the energy potential field encoder focuses on identifying local energy state distributions related to chemical bonding. The outputs of the two encoders are integrated through a weighted fusion layer to ultimately generate a spatial distribution model covering the entire surface of the aluminum substrate. This architectural design allows the two types of features to share underlying texture information during extraction while maintaining independent optimization at the task objective level, avoiding the accuracy degradation caused by feature competition in multi-target extraction by a single encoder.

[0024] In the computational flow of a geometric topology encoder, a 3D point cloud or height map constructed from surface topography scanning data is used as input. For each surface sampling point, its local Gaussian curvature field and mean curvature field are calculated. Let the two principal curvatures at a certain sampling point be... and The Gaussian curvature at that point is... With mean curvature Defined respectively and Gaussian curvature fields are used to distinguish the local geometry of a surface: when When corresponding to peak or valley regions, When the corresponding saddle point region is reached, This corresponds to flat or cylindrical regions. By thresholding and labeling the connected components of the Gaussian curvature field across the entire surface, the spatial distribution of saddle point regions and peak-valley regions can be automatically determined without manual intervention.

[0025] For the identified peak and valley regions, the depression depth gradient within each region is further calculated using the average curvature field. The depression depth gradient reflects the abrupt change in curvature from the edge of the peak / valley region towards the center; a larger gradient indicates a steeper depression shape and a more significant mechanical interlocking effect on the coating material. The effective containment volume of the region is obtained by integrating the depression depth gradient within the spatial range of each peak / valley region. The physical meaning of this index is the maximum volume of coating material that the recessed structure can accommodate under ideal filling conditions. The larger the effective accommodating volume, the higher the potential strength of the mechanical anchoring. This index transforms discrete curvature data into a quantitative value of anchoring capacity with clear physical meaning, providing a comparable numerical basis for the subsequent construction of spatial distribution models.

[0026] For the saddle point region, further curvature direction analysis is conducted. The two principal curvatures at the saddle point are... and The opposite signs correspond to two orthogonal principal curvature directions. These two principal curvature directions are defined as the principal curvature direction and the secondary curvature direction, respectively, and the angle between them in space is calculated. .when A deviation from 90° indicates anisotropy in the geometry at the saddle point, suggesting that local stress is more likely to concentrate in a specific direction. This can be determined by statistically analyzing the included angles of the saddle point regions across the entire surface. The distribution can identify the overall stress concentration direction preference on the aluminum substrate surface, providing directional constraint information for subsequent prediction of interface stress evolution.

[0027] The energy potential field encoder also draws input from surface topography scanning data, but its analytical objective focuses on identifying the chemically active structure of grain boundary network distribution. It calculates the local curvature variation coefficient at each sampling location on the surface. It is defined as the ratio of the standard deviation to the mean of the curvature values ​​within a local neighborhood, i.e. ,in For the local curvature standard deviation, This represents the local curvature mean. Locations with a high coefficient of variation in curvature indicate significant abrupt changes in curvature over a short distance, corresponding to geometric abrupt change points on the aluminum substrate surface. These geometric abrupt change points physically correspond highly to the locations of crystal defects such as grain boundaries and dislocation outcrops, because the discontinuity of atomic arrangement at grain boundaries produces minute but high-resolution morphological jumps in the macroscopic morphology.

[0028] Spatial clustering analysis was performed on the marked geometric abrupt change points, and a density-based clustering method was used to identify the clustering regions of these points. The linear extension morphology of the abrupt change points within the clustering regions matches the mesh topology of the grain boundary network, thus mapping the clustering regions of geometric abrupt change points to the grain boundary network distribution map of the aluminum substrate. Due to the unsaturated atomic bonding state and high surface energy, the grain boundary regions are concentrated areas of active sites where chemical bonding reactions preferentially occur. Therefore, the grain boundary network distribution directly characterizes the spatial non-uniformity of chemical bonding activity on the aluminum substrate surface.

[0029] After completing feature extraction for both encoder paths, the effective volume will be accommodated. Three types of features—stress concentration direction information, grain boundary network distribution, and other characteristics—are weighted and fused. The weighting coefficients are automatically optimized during the training phase of a multi-task learning network based on measured data of interface bonding strength, maximizing the prediction accuracy of the fused result for interface bonding capability. The fused mechanical anchoring strength and chemical bonding activity values ​​are calculated separately for each surface sampling point and registered with the original surface morphology data in a three-dimensional coordinate system. The registration process employs an iterative nearest-point algorithm to ensure the accuracy of the correspondence between the two types of values ​​and their actual physical locations, avoiding positional errors in subsequent layer construction strategies due to coordinate offsets.

[0030] After registration, the mechanical anchoring strength value and chemical bonding activity value together constitute the core data layer of the spatial distribution model. This model uses a three-dimensional voxel mesh as its storage structure. Each voxel node records the coordinate information, mechanical anchoring strength value, and chemical bonding activity value of its corresponding surface location, forming a complete spatial description of the interfacial bonding capability of the entire aluminum substrate surface. The resolution of the spatial distribution model is determined by the sampling density of the surface morphology scan data. The higher the scanning density, the more refined the model's characterization of local interfacial bonding capability differences, and the higher the accuracy of its guidance for subsequent adaptive layer construction strategies. This model serves as the input basis for subsequent prediction of interfacial stress evolution trends and calculation of interfacial delamination risk distribution, ensuring the spatial specificity and accuracy of the multi-layer interfacial enhancement treatment scheme.

[0031] Based on the surface morphology scanning data, the local curvature variation coefficient at each location on the surface is calculated and geometric abrupt change points are marked. Spatial clustering is performed on these geometric abrupt change points to identify clustered regions. These clustered regions correspond to the grain boundary network distribution of the aluminum substrate, including: Based on the surface morphology scanning data, the surface normal vector of each sampling point is calculated, and the position where the angle between the normal vectors of adjacent sampling points exceeds the lattice orientation tolerance angle is marked as the orientation abrupt change region. A curvature profile is established along the direction of normal vector change within the orientation abrupt change region, and curvature extrema points on the profile are extracted. These curvature extrema points correspond to surface ridges and grooves at grain boundaries. The distribution density of curvature extrema points in the neighborhood of each sampling point is statistically analyzed, and the location where the density exceeds the grain boundary characteristic density threshold is taken as the peak center of the local curvature variation coefficient and marked as the geometric abrupt change point. Extract the three-dimensional coordinates of the geometric abrupt change points, perform ray tracing along the principal curvature direction of each point, and connect the geometric abrupt change points where the rays intersect to form a spatial chain, which serves as the linear skeleton of the aggregation region; Identify the triangular nodes and closed loops in the linear skeleton. The triangular nodes correspond to the intersection of three grains, and the closed loops correspond to the boundaries of a single grain. Reconstruct the grain boundary network distribution of the aluminum substrate based on the spatial distribution relationship of the triangular nodes and closed loops.

[0032] Accurate reconstruction of the grain boundary network distribution in aluminum substrates based on surface morphology scanning data is a crucial prerequisite for generating high-quality spatial distribution models. Aluminum alloys are composed of polycrystalline grains at the microscale. Due to discontinuous lattice orientations, the grain boundary regions often exhibit specific geometric features on their surface morphology, including microscopic ridges, grooves, and abrupt curvature changes. Through quantitative calculation and spatial analysis of these geometric features, the grain boundary network topology of aluminum substrates can be reconstructed from morphology scanning data without relying on additional imaging techniques.

[0033] For each sampling point in the surface morphology scanning data, the surface normal vector at that point is calculated based on its three-dimensional coordinate information. Specifically, taking the sampling point as the center, several adjacent sampling points within its neighborhood are selected. The least squares plane fitting method is used to determine the tangent plane of this local region, and the normal direction of the tangent plane is the surface normal vector of that sampling point. After obtaining the normal vectors of each sampling point, the angle between the normal vectors of adjacent sampling points is calculated one by one. When the angle between the normal vectors of two adjacent points exceeds the preset lattice orientation tolerance angle, this position is marked as an orientation abrupt change region. The value of the lattice orientation tolerance angle is determined based on the crystallographic properties of aluminum alloy materials, and is usually set with reference to the orientation difference critical value between low-angle and high-angle grain boundaries of aluminum alloys to ensure that the marking results can effectively distinguish the grain boundary region from the normal morphology fluctuations within the grain.

[0034] After identifying orientation abrupt change regions, curvature profiles are established along the direction of the most significant change in normal vector within each region. The direction of normal vector change is determined by calculating the gradient direction of the angle between the normal vectors of adjacent points; this direction is typically perpendicular to the grain boundary orientation, thus maximizing the capture of the cross-sectional features of the morphology at the grain boundary. On the established curvature profiles, the curvature values ​​of the profile curves are calculated point by point, and the maxima and minima of curvature are extracted, i.e., curvature extrema. These extrema physically correspond to the surface ridges and grooves caused by lattice mismatch at the grain boundary. The curvature at the ridges is a positive maxima, and the curvature at the grooves is a negative minima; together, they constitute the characteristic imprint of the grain boundary on the surface morphology.

[0035] For each sampling point, the spatial distribution density of curvature extrema within its neighborhood is statistically analyzed. The neighborhood radius is determined by considering both the spatial resolution of the scan data and the typical range of aluminum alloy grain sizes. When the distribution density of curvature extrema within the neighborhood of a sampling point exceeds a preset grain boundary characteristic density threshold, the sampling point is identified as the peak center of the local curvature variation coefficient and marked as a geometric abrupt change point. The determination of the grain boundary characteristic density threshold can be based on standard grain size distribution data of aluminum alloy materials, establishing a reasonable discrimination standard through statistical analysis to avoid misjudging local morphological fluctuations within grains as grain boundary features. After the above steps, a set of geometric abrupt change points distributed throughout the entire scan area is finally obtained, which spatially outlines the preliminary contour of the grain boundary network.

[0036] After extracting the 3D coordinates of all geometric abrupt change points, ray tracing is performed along the principal curvature direction at each point. The principal curvature direction is determined by the eigenvector of the curvature tensor at that point, representing the spatial direction where curvature changes most drastically, and usually coincides with the extension direction of the grain boundary. During ray tracing, rays are emitted along the principal curvature direction starting from the current geometric abrupt change point. When the spatial distance between the ray and other geometric abrupt change points is less than the set connection tolerance range, the two points are connected to form a line segment unit of the spatial chain. By sequentially performing ray tracing and connection operations on all geometric abrupt change points, a spatial chain network composed of line segment units is gradually constructed. These spatial chains constitute the linear skeleton of the clustered region, exhibiting a mesh topology corresponding to the grain boundary network on a macroscopic level.

[0037] After establishing the linear skeleton, the topological nodes in the skeleton network are identified and classified. When a node has exactly three skeleton segments converging, it is identified as a triangular node, corresponding to the triangular grain boundary where three grains intersect in the polycrystalline structure of aluminum alloy. Triangular nodes are the most common topological units in the grain boundary network of polycrystalline materials, and their spatial distribution density is directly related to the grain size; the higher the density, the finer the corresponding grain. When skeleton segments connect end to end to form a closed loop, this loop is identified as a closed loop, corresponding to the complete boundary profile of a single grain. The shape and area of ​​the closed loop reflect the geometry of the corresponding grain; approximately equiaxed closed loops correspond to equiaxed grains, while elongated closed loops correspond to deformed grains or columnar grains.

[0038] Based on the spatial distribution relationship between the three-way nodes and closed loops, the grain boundary network distribution of the aluminum substrate is reconstructed. During reconstruction, the three-way nodes are used as network vertices, the skeleton segments connecting the three-way nodes as network edges, and the regions enclosed by closed loops as network surfaces, thus constructing a complete grain boundary network topology. For isolated segments or dangling ends not connected to the three-way nodes in the skeleton network, corrections are made based on the peak distribution of the local curvature variation coefficient, filling in grain boundary breaks that may be caused by scanning noise or data loss, thereby improving the completeness and continuity of the grain boundary network reconstruction.

[0039] The reconstructed grain boundary network distribution data is mapped back to the spatial coordinate system of the aluminum substrate surface and superimposed and correlated with the mechanical anchoring strength and chemical bonding activity values ​​in the spatial distribution model. Due to the irregularity of atomic arrangement, grain boundary regions typically exhibit higher chemical activity, corresponding to local maxima of chemical bonding activity values; while the surface ridge structures at grain boundaries provide additional mechanical anchoring sites for the coating, corresponding to regions with enhanced mechanical anchoring strength. Integrating the grain boundary network distribution information into the spatial distribution model further improves the prediction accuracy of interface delamination risk distribution, especially for the stress concentration risk assessment in densely grain boundary regions, thus providing a more refined spatial basis for the composition gradient design of the interface transition layer in adaptive layer construction strategies.

[0040] Based on the spatial distribution model and the matrix material property data, the evolution trend of interfacial stress in the multilayer interface system is predicted and the distribution of interfacial delamination risk is calculated, including: The mechanical anchoring strength and chemical bonding activity values ​​of each surface location are extracted from the spatial distribution model, and the elastic modulus and thermal expansion coefficient of the aluminum substrate are extracted from the matrix material property data. Calculate the spatial gradient modulus of the mechanical anchoring strength value, mark the position where the spatial gradient modulus exceeds the bearing change threshold as the stress accumulation zone, and in the stress accumulation zone, multiply the chemical bonding activity value with the elastic modulus of the aluminum substrate to obtain the interface stiffness matching coefficient, and mark the position where the interface stiffness matching coefficient is lower than the coordination critical coefficient as the stiffness mismatch zone. A time-domain evolution sequence of interface stress is established within the stiffness mismatch region. The time-domain evolution sequence records the changes of interface normal stress and interface tangential stress with temperature cycling. The time and spatial location of stress peaks in the time-domain evolution sequence are extracted. At the stress peak location, the ratio of the interface normal stress to the chemical bonding activity value is calculated as the bonding failure tendency factor. The cumulative number of times the bonding failure tendency factor exceeds the failure initiation factor during temperature cycling at each location is counted, and a normalized mapping is performed in the three-dimensional spatial coordinate system to generate the interface peeling risk distribution.

[0041] When extracting the mechanical anchoring strength and chemical bonding activity values ​​at each surface location from the spatial distribution model, it is necessary to read the quantized values ​​stored in the model point by point according to the preset spatial sampling grid. The mechanical anchoring strength value reflects the supporting ability of the surface morphology for the physical interlocking of the coating, while the chemical bonding activity value characterizes the degree to which the active sites on the aluminum substrate surface promote the formation of interfacial chemical bonds. Simultaneously, the elastic modulus of the aluminum substrate is extracted from the matrix material property data. With coefficient of thermal expansion These two parameters are the fundamental mechanical inputs for subsequent interface stress calculations. Elastic modulus The coefficient of thermal expansion determines the aluminum substrate's resistance to deformation under stress. This determines the dimensional response of the aluminum substrate to temperature changes, and both factors together affect the stress state of the multilayer interface system under service conditions.

[0042] When calculating the spatial gradient modulus of mechanical anchorage strength, the spatial rate of change of mechanical anchorage strength values ​​within the neighborhood of each sampling point is numerically differentiated. Let the mechanical anchorage strength value at a certain sampling point be... Their partial derivatives in the three-dimensional spatial coordinate directions are respectively , , Then the spatial gradient magnitude at that point for .when Exceeding the preset mutation threshold At that time, this location was marked as a stress-prone area. Bearing sudden change threshold. The setting is determined comprehensively based on the material system of the aluminum substrate and the expected service load conditions, and is usually obtained through statistical analysis of a large number of samples to ensure the representativeness of the marking results. The physical meaning of the stress concentration zone is that the rapid spatial change in mechanical anchoring strength will lead to uneven distribution of interface load, thus creating potential conditions for stress concentration at these locations.

[0043] After identifying stress-prone regions, the interfacial stiffness matching coefficients within these regions are calculated. The chemical bonding activity values ​​at each stress-prone region location are then calculated. Elastic modulus of aluminum substrate Perform a product operation to obtain the interface stiffness matching coefficient at that location. ,Right now Interface stiffness matching coefficient The physical significance of this value lies in its comprehensive reflection of the synergistic level between the chemical bonding ability of the interface and the mechanical stiffness of the matrix. A higher value indicates that the interface at that location possesses a stronger ability to coordinate deformation and transfer loads. Below the coordination critical coefficient The location is further marked as the stiffness mismatch zone. Compatibility critical coefficient. The discrimination criteria are determined in advance based on the design requirements and material combination characteristics of the multi-layer interface system. The location in the stiffness mismatch zone is more prone to interface damage under external excitation such as temperature cycling due to insufficient chemical bonding ability or excessive difference between the stiffness of the matrix and the coating.

[0044] Establishing a time-domain evolution sequence of interfacial stress within the stiffness mismatch region is a core step in predicting the risk of interfacial delamination. For each location of the stiffness mismatch region, the interfacial normal stress at that location is calculated progressively according to a set time step, with the temperature cycling process as the time axis. With interface tangential stress Interface normal stress Characterizing tensile or compressive forces perpendicular to the interface direction, interfacial tangential stress The shear force, parallel to the interface direction, and the shear force together determine the stress state of the interface at a specific moment. In the calculation of temperature cycling, the coefficient of thermal expansion... With elastic modulus As a key input parameter, it participates in the iterative solution of the stress field, ensuring that the time-domain evolution sequence can accurately reflect the stress accumulation and release process caused by thermal mismatch between the aluminum substrate and each coating. The time-domain evolution sequence fully records... and The data from the entire process of temperature cycling provides complete time history information for subsequent extraction of stress peaks.

[0045] When extracting the time and spatial location of stress peak occurrence from the time-domain evolution sequence, the location of each stiffness mismatch zone is... Extremum search is performed on the time series to identify the moment when the interface normal stress reaches a local maximum. And the corresponding spatial coordinates. The occurrence of stress peaks usually corresponds to the heating peak or cooling trough in the temperature cycle. At these moments, the difference in thermal expansion between the aluminum substrate and the coating is most significant, and the interface bears the greatest tensile or shear load, thus most easily triggering the initiation of interface damage. Recording the timing and location information of these peaks constitutes the input dataset for subsequent calculation of the bond failure susceptibility factor.

[0046] At the location of the stress peak, the peak value of the interface normal stress will be... Chemical bonding activity at this position The ratio is defined as the bond-breaking tendency factor. ,Right now . The physical meaning of is to measure the degree of overload of the normal tensile stress on the interface relative to the chemical bonding capacity at that location. A larger value indicates that the interface is closer to the critical state of chemical bonding failure at that location. Exceeding the preset destructive initiation factor At that time, it was considered that the temperature cycle triggered a potential bond breakage event at that location. Breakage initiation factor. The failure criteria for chemical bonding between aluminum substrate and coating material system are determined based on the standard peel test and finite element inversion analysis.

[0047] Statistics on each location during the complete temperature cycle Exceed cumulative number of times This cumulative number of occurrences directly reflects the probability risk of interfacial delamination at that location under repeated thermal loads. The higher the value, the more stress shocks that location experienced during service, approaching or exceeding the critical state of bond failure, and the higher the risk of interface delamination. [The text then abruptly shifts to a seemingly unrelated topic:] ...to each location... The values ​​are normalized in a three-dimensional coordinate system. The normalization process will... Mapped to The interval allows for comparison and visualization of peeling risks at different locations under a unified dimension. The normalized risk values ​​form a continuous distribution field in three-dimensional space, i.e., an interface peeling risk distribution map. High-risk areas are clearly presented spatially, providing precise spatial positioning basis for the subsequent generation of adaptive layer construction strategies. Through this complete calculation process, it is possible to pre-identify weak points in the interface peeling of various regions on the aluminum substrate surface under temperature cycling service conditions before implementing multi-layer interface enhancement treatment, thereby providing data support for differentiated interface design.

[0048] Figure 2 This is a flowchart illustrating the method for generating an adaptive layer construction strategy according to an embodiment of the present invention. An adaptive layer construction strategy is generated based on the interface stripping risk distribution. This adaptive layer construction strategy dynamically optimizes the thickness distribution of each functional interface layer and the component gradient design of the interface transition layer to minimize the stress concentration of the multi-layer interface system, including: The spatial distribution of cumulative occurrences is extracted from the risk distribution of the interface peeling, and the spatial gradient of the cumulative occurrences is calculated to obtain the risk propagation direction field. The risk propagation direction field characterizes the propagation path of interface peeling damage in the multi-layer interface system. A stress transmission chain is established along the risk propagation direction field. The stress transmission chain connects the locations of adjacent target accumulation times in space. The stress transmission efficiency at each node on the stress transmission chain is calculated. The stress transmission efficiency is equal to the weighted sum of the mechanical anchoring strength value and the chemical bonding activity value at the node location. The nodes in the stress transmission chain whose stress transmission efficiency is lower than the transmission blocking threshold are identified and marked as stress stagnation points. An interface transition layer is inserted at the stress stagnation point, and the composition gradient of the interface transition layer is determined by the difference in chemical bonding activity values ​​between adjacent nodes. Extract the start and end nodes of the stress transmission chain, establish a thickness distribution optimization model between the start and end nodes, minimize the stress gradient integral value on the stress transmission chain path as the objective function, and solve the thickness distribution optimization model to obtain the thickness distribution scheme of each functional interface layer. The thickness allocation scheme is registered with the component gradient of the interface transition layer in a three-dimensional coordinate system to generate an adaptive layer construction strategy.

[0049] Extracting the cumulative number of occurrences at each spatial location from the risk distribution of interface stripping. After obtaining the spatial distribution data, spatial gradient calculations are performed on the distribution data to obtain the risk propagation direction field. Specifically, for each node on the three-dimensional spatial discrete mesh of the aluminum substrate surface and the multilayer interface system, the risk propagation direction field is calculated. along , , The partial derivatives in the three coordinate directions are combined to form a gradient vector. After normalization, this yields the risk propagation direction vector at that node. The physical meaning of the risk propagation direction field is that the gradient direction points towards... The direction of fastest increase, i.e., the direction along which the interface peeling damage is most likely to extend to adjacent regions, is identified. By interpolating and smoothing the gradient vectors of all nodes, a continuous risk propagation direction field can be obtained, intuitively presenting the potential propagation path of interface peeling damage in a multi-layer interface system, and providing spatial guidance for the subsequent establishment of stress transfer chains.

[0050] Screening along the risk propagation direction field Nodes exceeding a preset target threshold are sequentially linked together, forming a stress transfer chain, by connecting spatially adjacent nodes that have a continuous directional relationship in the risk propagation direction field. Essentially, the stress transfer chain is an ordered sequence of nodes extending along the damage propagation path in three-dimensional space, reflecting the spatial transmission path of interface delamination risk. For each node in the stress transfer chain, the stress transfer efficiency at that node is calculated. Its expression is: , in, This represents the mechanical anchorage strength value at that node location. This represents the chemical bonding activity value at that node location. and These are the weighting coefficients for mechanical anchoring strength and chemical bonding activity on stress transfer efficiency, respectively, and satisfy the following conditions: The weighting coefficient is determined based on the material properties of the aluminum substrate and the design requirements of the multilayer interface system. When the bonding mechanism on the aluminum substrate surface is mainly physical interlocking, the weighting coefficient should be appropriately increased. When the interface is dominated by chemical bonding, appropriately increase... By analyzing all nodes in the stress transfer chain... By performing statistical analysis on the values, weak links in the chain can be identified.

[0051] Identify stress transmission chains Below the transmission blocking threshold The nodes are marked as stress retardation points. The physical meaning of a stress retardation point is that the interfacial bonding capacity at that location is insufficient to effectively transfer and disperse interfacial stress; if left untreated, stress will concentrate at this location and trigger preferential propagation of peeling damage. For each stress retardation point, an interfacial transition layer is inserted at that location. The composition gradient of the interfacial transition layer is determined by the difference in chemical bonding activity between the retardation point and its adjacent nodes: let the chemical bonding activity of the retardation point be... Its precursor node's chemical bonding activity value is The chemical bonding activity value of the successor node is The slope of the component gradient in the thickness direction of the interface transition layer is then... for:

[0052] in, This represents the spatial spacing between adjacent nodes along the stress transfer chain path. Component gradient slope. The larger the absolute value, the more significant the difference in chemical bonding activity between adjacent nodes, requiring a more gradual component transition to avoid additional stress concentration caused by abrupt compositional changes. The components of the interface transition layer follow the thickness direction according to... The described linear or polynomial gradient is distributed to achieve a smooth transition of interfacial bonding ability at hindrance points with low chemical bonding activity, effectively eliminating the blocking effect of local weak interfaces on the overall stress transmission chain.

[0053] After extracting the start and end nodes of the stress transmission chain, a thickness distribution optimization model is established within the path defined by these two endpoints. The objective function is to minimize the integral value of the stress gradient along the stress transmission chain path. Its expression is: , in, Let these be the coordinates of the arc length along the stress transfer chain path. and These correspond to the arc length positions of the starting and ending nodes, respectively. This represents the interface stress distribution at various locations along the path, which is directly related to the thickness allocation scheme of each functional interface layer. The decision variable is the thickness of each functional interface layer. ( ,in (This refers to the total number of functional interface layers). Constraints include: non-negative thickness of each layer, total thickness not exceeding the design upper limit, and each layer thickness meeting the minimum thickness requirement achievable by the manufacturing process. The above optimization model is solved using gradient descent or a genetic algorithm to obtain the stress gradient integral value. Minimum optimal thickness allocation scheme This scheme ensures a smooth transition of interfacial stress along the stress transfer chain path, fundamentally reducing stress concentration.

[0054] Thickness distribution scheme The composition gradient information of the interface transition layer at each stress retardation point is registered in a three-dimensional spatial coordinate system to generate the final adaptive layer construction strategy. During the registration process, the thickness distribution scheme of each functional interface layer is mapped according to its spatial position on the aluminum substrate surface, using the three-dimensional scanning coordinate system of the aluminum substrate surface as a reference. This ensures that the layer thickness design of each region corresponds to the interface delamination risk distribution of that region. The position and composition gradient information of the interface transition layer are also marked in the same coordinate system, forming a complete three-dimensional spatial interface layer design map. The adaptive layer construction strategy is stored in the form of a spatial coordinate index, containing the thickness value of each functional interface layer and the composition gradient parameters of the interface transition layer at each spatial location. This can be directly used as input parameters for subsequent multi-layer interface enhancement processes, guiding the execution of actual coating deposition or surface treatment processes. This ensures that the multi-layer interface system is specifically strengthened in high-risk areas, minimizing stress concentration and maximizing interface bonding reliability overall.

[0055] A thickness distribution optimization model is established between the start and end nodes, with the objective function being to minimize the stress gradient integral value along the stress transmission chain path. Solving the thickness distribution optimization model yields the thickness distribution schemes for each functional interface layer, including: All intermediate nodes between the starting node and the ending node on the stress transmission chain are extracted, and a local stress tensor field is constructed. Tensor decomposition is performed based on the mechanical anchoring strength value and chemical bonding activity value at the node position to obtain the stress tensor distribution sequence on the stress transmission chain path. Gradient calculation is performed on the stress tensor distribution sequence along the path direction, and the spatial rate of change of the principal component of the stress tensor is extracted as the stress gradient field. The position where the gradient magnitude exceeds the gradient jump threshold in the stress gradient field is identified and marked as the stress transmission inflection point. Each stress transmission inflection point is used as the boundary marker of the functional interface layer. The stress gradient field is integrated within the path segment between adjacent stress transmission inflection points. The path integral value represents the degree of stress concentration accumulation within the corresponding functional interface layer. A thickness allocation optimization model is established, with the thickness of each functional interface layer as the decision variable and the objective function being the weighted sum of the path integral values ​​of all functional interface layers. The thickness allocation optimization model is solved by gradient descent iterative method. In each iteration, the thickness allocation is adjusted according to the partial derivative of the objective function with respect to the thickness of each functional interface layer. The iteration is terminated when the change in the objective function value between two adjacent iterations is lower than the convergence criterion, and the thickness allocation scheme of each functional interface layer is output.

[0056] After constructing the stress transmission chain and identifying the bottleneck points, a thickness distribution optimization model needs to be established between the starting and ending nodes to achieve precise thickness allocation for each functional interface layer. All intermediate nodes between the starting and ending nodes in the stress transmission chain are extracted. The spatial coordinates of these nodes, along with their corresponding mechanical anchoring strength and chemical bonding activity values, are used as inputs to construct a local stress tensor field in the local coordinate system. The construction of the local stress tensor field centers on each intermediate node, comprehensively considering the coupling relationship between the normal and tangential stress components within its neighborhood, forming a complete second-order symmetric tensor representation. Based on the mechanical anchoring strength and chemical bonding activity values ​​at each node location, eigenvalue decomposition is performed on the local stress tensor to extract the principal stress directions and corresponding principal stress values, thus obtaining a stress tensor distribution sequence arranged along the stress transmission chain path. This distribution sequence fully characterizes the variation of the internal stress state of the interface with spatial location from the starting to the ending node, providing fundamental data for subsequent gradient calculations.

[0057] When performing gradient calculations on the stress tensor distribution sequence along the path direction, the maximum principal stress component of the stress tensor at each node is extracted as the principal component. The spatial rate of change of the principal components between adjacent nodes is calculated to obtain the stress gradient field. Let the stress tensor distribution sequence along the path be... The maximum principal stress component at each node is Adjacent nodes and The spatial spacing between them is The approximate stress gradient of this segment is... Along the entire path, the stress gradient field is composed of all Composed of, its gradient magnitude is In a stress gradient field, the gradient magnitude is... Exceeding the preset gradient jump threshold The location is marked as the stress transfer inflection point. Gradient jump threshold. Based on the elastic modulus of the aluminum substrate The overall path length is determined in conjunction with the overall path length to ensure that inflection point markers are triggered only at locations where there are significant abrupt changes in stress state, thus avoiding the occurrence of too many redundant boundary points due to local numerical fluctuations.

[0058] Using stress transfer inflection points as boundaries for functional interface layers, the path from the starting node to the ending node is divided into several continuous path segments, each corresponding to a functional interface layer. Within each path segment, the stress gradient field is integrally analyzed; the path integral value characterizes the degree of stress concentration accumulation within that functional interface layer. Let the first... The starting arc length coordinates of the path segment corresponding to each functional interface layer are: The coordinates of the endpoint arc length are Then the path integral value of this layer for ,in A continuous representation of the stress gradient field at each location along the path is obtained by cubic spline interpolation of the discrete gradient sequence. Path integral value. The larger the value, the more severe the stress concentration accumulation within the functional interface layer, requiring the layer thickness to be increased to disperse the stress concentration and reduce the risk of interface peeling.

[0059] When establishing the thickness allocation optimization model, the thickness of each functional interface layer is used as the basis. ( ,in The total number of functional interface layers is the decision variable, and the objective function is... Defined as a weighted sum of the path integrals of all functional interface layers: ,in For the first The weight coefficient corresponding to each functional interface layer reflects the importance of that layer within the overall interface system. Weight coefficient Based on the number of blocking points and the bonding failure tendency factor within this path segment. The cumulative distribution of the comprehensive assignment, the dense blockage points or Layers with higher cumulative values ​​are assigned greater weight. Simultaneously, constraints are imposed on the decision variables: the thickness of each layer. Lower bound constraint must be satisfied This ensures that each layer has a basic physical thickness; total thickness constraint. ,in The target total thickness of the multi-layer interface system is determined by the matrix material properties of the aluminum substrate and the application conditions.

[0060] The thickness allocation optimization model described above is solved using an iterative gradient descent method. During the iterative initialization phase, the thickness of each functional interface layer is uniformly allocated, i.e., let... This serves as the starting point for iteration. In the... In the next iteration, the objective function is calculated. Thickness of each functional interface layer partial derivatives The partial derivative is obtained by integrating the path value. about The numerical derivative is obtained, specifically using a central difference scheme to improve computational accuracy. Based on the partial derivative values, the thickness allocation is adjusted according to the following update rules: ,in The learning rate controls the step size of each iteration. After each thickness update, a projection operation is performed on the updated thickness vector to ensure that the thickness of each layer satisfies the lower bound constraint. At the same time, the total thickness is restored to its original value through proportional normalization. This ensures that the constraints remain true throughout the entire iteration process.

[0061] The convergence criterion has a significant impact on the timing of iteration termination. The change in the objective function value between two adjacent iterations is defined as... ,when Below the preset convergence threshold When convergence is reached, the iteration is considered to have converged and the computation is terminated. Convergence threshold. The objective function is determined based on its dimensions and the required processing precision of the aluminum substrate; typically, 10% of the initial value of the objective function is taken. -4 Scale. After the iteration terminates, output the thickness allocation scheme for each functional interface layer. This is the optimal solution for the thickness allocation optimization model. This optimal thickness allocation scheme ensures that, under the premise of satisfying the total thickness constraint, the weighted sum of the stress concentration accumulation in all functional interface layers is minimized. This provides accurate layer thickness parameter guidance for the implementation of subsequent multi-layer interface enhancement processing, and significantly reduces the risk of interface delamination caused by local stress concentration.

[0062] According to the adaptive layer construction strategy, a multi-layer interface enhancement treatment is performed on the surface of the aluminum substrate, and the interfacial bonding strength of the multi-layer interface system after the treatment is measured, including: Based on the thickness allocation scheme of each functional interface layer and the component gradient of the interface transition layer in the adaptive layer construction strategy, the spatial coordinate positioning information and deposition sequence information of each functional interface layer on the aluminum substrate surface are extracted to generate the process execution sequence of multi-layer interface enhancement treatment. According to the process execution sequence, each functional interface layer is deposited sequentially on the surface of the aluminum substrate, and an interface transition layer is deposited between adjacent functional interface layers. The composition of the interface transition layer is continuously adjusted along the thickness direction according to the composition gradient, so that the composition of the interface transition layer transitions from the main component of the previous functional interface layer to the main component of the next functional interface layer. The interfacial bonding strength of the deposited multi-layer interface system was measured. The interfacial bonding strength was obtained by applying a normal tensile load to the surface of the multi-layer interface system and recording the critical load value when the interface peels off. The critical load value was normalized to the unit interface area to obtain the interfacial bonding strength.

[0063] Based on the thickness allocation scheme of each functional interface layer and the component gradient information of the interface transition layer determined by the adaptive layer construction strategy, these abstract optimization results need to be transformed into an execution sequence that can directly guide the actual process operation. During the extraction process, the spatial coordinate positioning information of each functional interface layer on the aluminum substrate surface is analyzed layer by layer to clarify the coverage area boundary of each layer in the planar direction and the start and end depths in the thickness direction, while recording the deposition sequence of each layer. The spatial coordinate positioning information comes from the interface peeling risk distribution map established in the adaptive layer construction strategy. High-risk areas correspond to thicker functional interface layers or denser interface transition layer arrangements, while low-risk areas are correspondingly thinner. The deposition sequence information is arranged sequentially from the aluminum substrate surface outwards, ensuring that the first functional interface layer in direct contact with the aluminum substrate has optimal chemical bonding activity matching, and the outermost functional interface layer meets the surface mechanical performance requirements of the service environment. After integrating the above spatial coordinate positioning information and deposition sequence information, a complete process execution sequence is generated. This sequence is stored in structured data form, containing the target layer type, target thickness, spatial coverage, and component parameters for each deposition step, which can directly drive the parameter calls of subsequent deposition equipment.

[0064] After the process execution sequence is generated, deposition operations are performed layer by layer on the aluminum substrate surface according to the deposition order specified in the sequence. For the deposition of the first functional interface layer, the process begins immediately after the aluminum substrate surface pretreatment to avoid reduced chemical bonding activity due to natural oxidation in the exposed environment. During deposition, the deposition rate and time are adjusted in real time according to the target thickness of this layer in the process execution sequence to keep the deviation between the actual deposition thickness and the designed thickness within a reasonable range. After the first functional interface layer is deposited, the vacuum or protective atmosphere is not interrupted, and the deposition of the first interface transition layer begins directly. The deposition of the interface transition layer is the most challenging part of the entire multilayer interface enhancement process, requiring continuous control of its composition along the thickness direction, smoothly transitioning from the main components of the previous functional interface layer to the main components of the next functional interface layer, without any abrupt changes in composition.

[0065] The continuous control of the composition of the interface transition layer is achieved by dynamically adjusting the power ratio of each target in the dual-target or multi-target co-deposition device. At the beginning of the interface transition layer deposition, the power of the target corresponding to the main component of the previous functional interface layer dominates. As the deposition thickness increases, the power of this target decreases linearly according to the composition gradient slope determined in the adaptive layer construction strategy, while the power of the target corresponding to the main component of the subsequent functional interface layer increases linearly until the latter power dominates at the end of the interface transition layer deposition. During this dynamic control process, the slope of the composition gradient strictly follows the output result of the adaptive layer construction strategy. Interface transition layers in different spatial regions have different composition gradient slopes; therefore, the deposition equipment needs to have zonal control capabilities, enabling independent adjustment of the target power ratio change rate for different locations on the aluminum substrate surface. After the deposition of the first interface transition layer is completed, subsequent functional interface layers and interface transition layers are deposited sequentially according to the same control logic until all layers in the process execution sequence are deposited, forming a complete multi-layer interface system.

[0066] After the multilayer interface system is deposited, its interfacial bonding strength needs to be quantitatively measured to verify the effectiveness of the adaptive layer construction strategy. The interfacial bonding strength is measured using a normal tensile loading method, which directly reflects the interface's resistance to peeling in the direction perpendicular to the interface plane and has a good correlation with the peeling load mode experienced by the multilayer interface system in actual service. Before measurement, a standard tensile fixture is bonded to the outermost surface of the multilayer interface system. The adhesive bonding strength must be higher than the expected interfacial bonding strength to ensure that failure occurs at the target interface rather than the adhesive layer during measurement. The other end of the aluminum substrate is fixed in the lower fixture of the tensile testing device, ensuring that the tensile load direction is strictly perpendicular to the interface plane, and the angular error is controlled within a reasonable range to avoid introducing tangential components due to skew, which would lead to lower measurement results.

[0067] During the tensile loading process, a normal tensile load is applied at a constant loading rate, and the load-displacement curve is continuously recorded. When the interface peels off, a significant drop in load occurs on the load-displacement curve. The maximum load value at this point is the critical load value at which the interface peels off, denoted as . The critical load value Divide by the interface area of ​​the measurement area The interfacial bonding strength normalized to a unit interfacial area is obtained. ,Right now ,in The unit is Pascal. Normalization eliminates systematic errors introduced by differences in fixture area among different measurement samples, making measurement results from different locations and batches directly comparable.

[0068] To improve the statistical reliability of the interfacial bonding strength measurement results, multiple measurement points were selected at different locations on the surface of the multilayer interface system, corresponding to high-risk, medium-risk, and low-risk areas in the interfacial debonding risk distribution, with at least three repeated measurements performed for each type of area. The mean value of the interfacial bonding strength measurement results for each area was calculated, and the standard deviation was used to characterize the measurement dispersion. The mean value of the measured interfacial bonding strength for each area was compared with the target value of the interfacial bonding strength expected by the adaptive layer construction strategy. If the relative deviation between the measured value and the target value exceeded the preset tolerance range, a feedback correction process for process parameters was triggered. The deposition parameters for the corresponding area in the process execution sequence were adjusted, and deposition and measurement were repeated until the measured interfacial bonding strength met the design requirements. After the measurement was completed, the morphology of the debonding section could be analyzed. By observing the fracture mode (interfacial fracture or cohesive fracture), the location of the weakest interface in the multilayer interface system could be further determined, providing a basis for subsequent process optimization.

[0069] A second aspect of the present invention provides a multilayer interface enhancement treatment system for aluminum-based surfaces, comprising: The data acquisition unit is used to acquire surface morphology scanning data and matrix material property data of aluminum substrate; The spatial distribution modeling unit is used to perform collaborative feature extraction on the surface morphology scanning data through a multi-task learning network, simultaneously identify the surface unevenness features that affect mechanical anchoring and the distribution of active sites that affect chemical bonding, and generate a spatial distribution model that characterizes the interfacial bonding ability. The spatial distribution model includes the mechanical anchoring strength value and chemical bonding activity value at each surface location. The risk distribution prediction unit is used to predict the interfacial stress evolution trend of the multilayer interface system and calculate the interfacial delamination risk distribution based on the spatial distribution model and the matrix material property data. An adaptive layer strategy construction unit is used to generate an adaptive layer construction strategy based on the interface stripping risk distribution. The adaptive layer construction strategy dynamically optimizes the thickness distribution of each functional interface layer and the component gradient design of the interface transition layer to minimize the stress concentration of the multi-layer interface system. An execution unit is used to perform multi-layer interface enhancement treatment on the surface of the aluminum substrate according to the adaptive layer construction strategy, and to measure the interfacial bonding strength of the multi-layer interface system after the treatment is completed.

[0070] A third aspect of the present invention provides an electronic device, comprising: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0071] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0072] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0073] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for multilayer interface enhancement treatment of aluminum-based surfaces, characterized in that, include: Collect surface morphology scanning data and matrix material property data of the aluminum substrate; A multi-task learning network is used to collaboratively extract features from the surface morphology scanning data, simultaneously identifying surface irregularities affecting mechanical anchoring and the distribution of active sites affecting chemical bonding. This generates a spatial distribution model characterizing interfacial bonding ability. The spatial distribution model includes mechanical anchoring strength values ​​and chemical bonding activity values ​​at each surface location, including: A dual-encoder architecture for a multi-task learning network is constructed, comprising a geometric topology encoder and an energy potential field encoder; In the geometric topology encoder, based on the surface topography scanning data, the Gaussian curvature field and the average curvature field of the surface are calculated, the saddle point region and peak-valley region of the surface are identified according to the Gaussian curvature field, and the depression depth gradient of each peak-valley region is calculated according to the average curvature field. The effective containment volume of each peak and valley region is obtained by spatial integration of the depression depth gradient. The curvature direction of the saddle point region is analyzed to obtain the principal curvature direction and the secondary curvature direction. The stress concentration direction of the surface is identified based on the angle distribution between the principal and secondary curvature directions. In the energy potential field encoder, based on the surface topography scanning data, the local curvature variation coefficient at each position on the surface is calculated and geometric abrupt change points are marked. Spatial clustering is performed on the geometric abrupt change points to identify the clustering regions of the geometric abrupt change points. The clustering regions correspond to the grain boundary network distribution of the aluminum substrate. The effective containment volume, the stress concentration direction and the grain boundary network distribution are weighted and fused to generate mechanical anchoring strength and chemical bonding activity value, and then registered in a three-dimensional spatial coordinate system to generate a spatial distribution model. Based on the spatial distribution model and the matrix material property data, the interfacial stress evolution trend of the multilayer interface system is predicted and the distribution of interfacial delamination risk is calculated. An adaptive layer construction strategy is generated based on the interface peeling risk distribution. The adaptive layer construction strategy dynamically optimizes the thickness distribution of each functional interface layer and the component gradient design of the interface transition layer, so as to minimize the stress concentration of the multi-layer interface system. According to the adaptive layer construction strategy, a multi-layer interface enhancement treatment is performed on the surface of the aluminum substrate, and the interfacial bonding strength of the multi-layer interface system after the treatment is measured.

2. The method according to claim 1, characterized in that, Based on the surface morphology scanning data, the local curvature variation coefficient at each location on the surface is calculated and geometric abrupt change points are marked. Spatial clustering is performed on these geometric abrupt change points to identify clustered regions. These clustered regions correspond to the grain boundary network distribution of the aluminum substrate, including: Based on the surface morphology scanning data, the surface normal vector of each sampling point is calculated, and the position where the angle between the normal vectors of adjacent sampling points exceeds the lattice orientation tolerance angle is marked as the orientation abrupt change region. A curvature profile is established along the direction of normal vector change within the orientation abrupt change region, and curvature extrema points on the profile are extracted. These curvature extrema points correspond to surface ridges and grooves at grain boundaries. The distribution density of curvature extrema points in the neighborhood of each sampling point is statistically analyzed, and the location where the density exceeds the grain boundary characteristic density threshold is taken as the peak center of the local curvature variation coefficient and marked as the geometric abrupt change point. Extract the three-dimensional coordinates of the geometric abrupt change points, perform ray tracing along the principal curvature direction of each point, and connect the geometric abrupt change points where the rays intersect to form a spatial chain, which serves as the linear skeleton of the aggregation region; Identify the triangular nodes and closed loops in the linear skeleton. The triangular nodes correspond to the intersection of three grains, and the closed loops correspond to the boundaries of a single grain. Reconstruct the grain boundary network distribution of the aluminum substrate based on the spatial distribution relationship of the triangular nodes and closed loops.

3. The method according to claim 1, characterized in that, Based on the spatial distribution model and the matrix material property data, the evolution trend of interfacial stress in the multilayer interface system is predicted and the distribution of interfacial delamination risk is calculated, including: The mechanical anchoring strength and chemical bonding activity values ​​of each surface location are extracted from the spatial distribution model, and the elastic modulus and thermal expansion coefficient of the aluminum substrate are extracted from the matrix material property data. Calculate the spatial gradient modulus of the mechanical anchoring strength value, mark the position where the spatial gradient modulus exceeds the bearing change threshold as the stress accumulation zone, and in the stress accumulation zone, multiply the chemical bonding activity value with the elastic modulus of the aluminum substrate to obtain the interface stiffness matching coefficient, and mark the position where the interface stiffness matching coefficient is lower than the coordination critical coefficient as the stiffness mismatch zone. A time-domain evolution sequence of interface stress is established within the stiffness mismatch region. The time-domain evolution sequence records the changes of interface normal stress and interface tangential stress with temperature cycling. The time and spatial location of stress peaks in the time-domain evolution sequence are extracted. At the stress peak location, the ratio of the interface normal stress to the chemical bonding activity value is calculated as the bonding failure tendency factor. The cumulative number of times the bonding failure tendency factor exceeds the failure initiation factor during temperature cycling at each location is counted, and a normalized mapping is performed in the three-dimensional spatial coordinate system to generate the interface peeling risk distribution.

4. The method according to claim 1, characterized in that, An adaptive layer construction strategy is generated based on the interface peeling risk distribution. This strategy dynamically optimizes the thickness distribution of each functional interface layer and the component gradient design of the interface transition layer to minimize the stress concentration of the multi-layer interface system. This includes: The spatial distribution of cumulative occurrences is extracted from the risk distribution of the interface peeling, and the spatial gradient of the cumulative occurrences is calculated to obtain the risk propagation direction field. The risk propagation direction field characterizes the propagation path of interface peeling damage in the multi-layer interface system. A stress transmission chain is established along the risk propagation direction field. The stress transmission chain connects the locations of adjacent target accumulation times in space. The stress transmission efficiency at each node on the stress transmission chain is calculated. The stress transmission efficiency is equal to the weighted sum of the mechanical anchoring strength value and the chemical bonding activity value at the node location. The nodes in the stress transmission chain whose stress transmission efficiency is lower than the transmission blocking threshold are identified and marked as stress stagnation points. An interface transition layer is inserted at the stress stagnation point, and the composition gradient of the interface transition layer is determined by the difference in chemical bonding activity values ​​between adjacent nodes. Extract the start and end nodes of the stress transmission chain, establish a thickness distribution optimization model between the start and end nodes, minimize the stress gradient integral value on the stress transmission chain path as the objective function, and solve the thickness distribution optimization model to obtain the thickness distribution scheme of each functional interface layer. The thickness allocation scheme is registered with the component gradient of the interface transition layer in a three-dimensional coordinate system to generate an adaptive layer construction strategy.

5. The method according to claim 4, characterized in that, A thickness distribution optimization model is established between the start and end nodes, with the objective function being to minimize the stress gradient integral value along the stress transmission chain path. Solving the thickness distribution optimization model yields the thickness distribution schemes for each functional interface layer, including: All intermediate nodes between the starting node and the ending node on the stress transmission chain are extracted, and a local stress tensor field is constructed. Tensor decomposition is performed based on the mechanical anchoring strength value and chemical bonding activity value at the node position to obtain the stress tensor distribution sequence on the stress transmission chain path. Gradient calculation is performed on the stress tensor distribution sequence along the path direction, and the spatial rate of change of the principal component of the stress tensor is extracted as the stress gradient field. The position where the gradient magnitude exceeds the gradient jump threshold in the stress gradient field is identified and marked as the stress transmission inflection point. Each stress transmission inflection point is used as the boundary marker of the functional interface layer. The stress gradient field is integrated within the path segment between adjacent stress transmission inflection points. The path integral value represents the degree of stress concentration accumulation within the corresponding functional interface layer. A thickness allocation optimization model is established, with the thickness of each functional interface layer as the decision variable and the objective function being the weighted sum of the path integral values ​​of all functional interface layers. The thickness allocation optimization model is solved by gradient descent iterative method. In each iteration, the thickness allocation is adjusted according to the partial derivative of the objective function with respect to the thickness of each functional interface layer. The iteration is terminated when the change in the objective function value between two adjacent iterations is lower than the convergence criterion, and the thickness allocation scheme of each functional interface layer is output.

6. The method according to claim 1, characterized in that, According to the adaptive layer construction strategy, a multi-layer interface enhancement treatment is performed on the surface of the aluminum substrate, and the interfacial bonding strength of the multi-layer interface system after the treatment is measured, including: Based on the thickness allocation scheme of each functional interface layer and the component gradient of the interface transition layer in the adaptive layer construction strategy, the spatial coordinate positioning information and deposition sequence information of each functional interface layer on the aluminum substrate surface are extracted to generate the process execution sequence of multi-layer interface enhancement treatment. According to the process execution sequence, each functional interface layer is deposited sequentially on the surface of the aluminum substrate, and an interface transition layer is deposited between adjacent functional interface layers. The composition of the interface transition layer is continuously adjusted along the thickness direction according to the composition gradient, so that the composition of the interface transition layer transitions from the main component of the previous functional interface layer to the main component of the next functional interface layer. The interfacial bonding strength of the deposited multi-layer interface system was measured. The interfacial bonding strength was obtained by applying a normal tensile load to the surface of the multi-layer interface system and recording the critical load value when the interface peels off. The critical load value was normalized to the unit interface area to obtain the interfacial bonding strength.

7. A multilayer interface enhancement treatment system for aluminum-based surfaces, used to implement the method as described in any one of claims 1-6, characterized in that, include: The data acquisition unit is used to acquire surface morphology scanning data and matrix material property data of aluminum substrate; The spatial distribution modeling unit is used to perform collaborative feature extraction on the surface morphology scanning data through a multi-task learning network, simultaneously identify the surface unevenness features that affect mechanical anchoring and the distribution of active sites that affect chemical bonding, and generate a spatial distribution model that characterizes the interfacial bonding ability. The spatial distribution model includes the mechanical anchoring strength value and chemical bonding activity value at each surface location. The risk distribution prediction unit is used to predict the interfacial stress evolution trend of the multilayer interface system and calculate the interfacial delamination risk distribution based on the spatial distribution model and the matrix material property data. An adaptive layer strategy construction unit is used to generate an adaptive layer construction strategy based on the interface stripping risk distribution. The adaptive layer construction strategy dynamically optimizes the thickness distribution of each functional interface layer and the component gradient design of the interface transition layer to minimize the stress concentration of the multi-layer interface system. An execution unit is used to perform multi-layer interface enhancement treatment on the surface of the aluminum substrate according to the adaptive layer construction strategy, and to measure the interfacial bonding strength of the multi-layer interface system after the treatment is completed.

8. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, they implement the method described in any one of claims 1 to 6.

Images