High operating substrate laminate intelligent alignment method and system

Abstract

The present application relates to the technical field of intelligent manufacturing, and more particularly to a high operation base plate laminating intelligent alignment method and system. The method obtains a base plate surface image, material attribute parameters and design coordinate data, calculates interlayer shear stress distribution according to thermal expansion coefficient difference and converts it into a stress potential energy field, identifies local potential extreme points to construct a stress influence domain and calculates deformation sensitivity; divides the base plate surface image into observation windows and extracts the spatial topology structure of the positioning marks to encode them as feature vectors, filters a stable feature set based on deformation sensitivity; matches the stable feature set with the design coordinate data to obtain an initial pose offset and calculates compensation displacement and compensation angle; identifies the continuous region with the maximum potential energy gradient in the stress potential energy field to form a stress transmission path, generates an activation timing of compression contact points in the potential energy decreasing direction; controls the compression module to adjust the posture and drives the compression alignment according to the activation timing. The present application improves the alignment accuracy and efficiency of base plate lamination.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing technology, and in particular to a high-performance intelligent alignment method and system for laminating substrates. Background Technology

[0002] In multilayer lamination processes, alignment accuracy directly affects the electrical performance and mechanical reliability of the product. Current conventional methods typically employ mechanical locating pins in conjunction with an optical alignment system to achieve interlayer positioning. Some advanced solutions utilize temperature sensors to monitor the substrate temperature in real time and perform linear compensation based on the material's coefficient of thermal expansion to reduce deviations caused by thermal effects.

[0003] Conventional methods primarily rely on rigid geometric alignment models, assuming the base plate remains planar and undeformed during lamination. However, in actual lamination, the coefficients of thermal expansion of each layer differ significantly, especially during the heating and pressurization stages, generating interlaminar shear stress that causes non-uniform deformation of the base plate. This deformation is not a simple overall scaling or rotation, but includes local twisting and warping, which can severely deviate the actual position of the positioning markers from the image registration result. Compensation methods based solely on marker matching cannot capture the impact of stress distribution on deformation-sensitive areas, leading to a deterioration in alignment accuracy in stress-concentrated regions.

[0004] Existing pressing sequences typically employ a fixed pattern, proceeding from the center to the edge or from one side to the other, without considering the differences in stress transfer paths. This can lead to premature stress release along the region of least resistance during pressing, exacerbating misalignment in unpressed areas and ultimately affecting the overall alignment consistency of multi-layer structures. Therefore, there is an urgent need for an alignment method that can adapt to stress field distribution and optimize pressing sequence. Summary of the Invention

[0005] The present invention provides a high-performance substrate lamination intelligent alignment method and system, which can solve the problems in the prior art.

[0006] A first aspect of the present invention provides a high-performance substrate lamination intelligent alignment method, comprising: Acquire images of the base plate surface, material property parameters of each layer of the base plate, and design coordinate data of the target overlay layer; The interlayer shear stress distribution is calculated based on the difference in thermal expansion coefficients of each layer in the material property parameters and converted into a stress potential energy field. Local extreme points of potential energy in the stress potential energy field are identified, and a stress influence domain is constructed with the local extreme points of potential energy as the center. The boundary diffusion rate of the stress influence domain is calculated as the deformation sensitivity. The base plate surface image is divided into multiple observation windows. The spatial topology of the positioning marks in each observation window is extracted and encoded into feature vectors. Based on the spatial distribution of deformation sensitivity, the deformation stability region is identified, and the feature vectors corresponding to the observation windows in the deformation stability region are selected to form a stable feature set. The initial pose offset is obtained by matching the stable feature set with the topology in the design coordinate data, and the compensation displacement and compensation angle are calculated based on the initial pose offset. The stress transmission path is formed by identifying the continuous region with the largest potential energy gradient in the stress potential energy field, setting the pressing contact point along the stress transmission path, and generating the activation sequence of the pressing contact point according to the direction of decreasing potential energy. The control pressing module adjusts its position and posture according to the compensation displacement and compensation angle, and drives the pressing contact points to complete the lamination alignment according to the activation sequence.

[0007] Based on the differences in the thermal expansion coefficients of each layer in the material properties parameters, the interlayer shear stress distribution is calculated and converted into a stress potential energy field. Local extrema of the potential energy in the stress potential energy field are identified, and a stress influence domain is constructed with the local extrema of the potential energy as the center. The boundary diffusion rate of the stress influence domain is calculated as the deformation sensitivity, including: The shear strain distribution at the interface of each layer is calculated based on the difference in thermal expansion coefficient between adjacent layers and the amount of temperature change in the material property parameters. The interlayer shear stress distribution is then calculated based on the shear strain distribution and the shear modulus of each layer. The interlaminar shear stress distribution is used as a load source to construct a three-dimensional stress potential energy function, and the stress potential energy field is obtained by spatial integration of the three-dimensional stress potential energy function. In the stress potential energy field, calculate the spatial gradient distribution of potential energy values, and identify the spatial locations where the potential energy gradient is zero and the potential energy values ​​satisfy the local maximum condition as the local extreme points of potential energy. Starting from the local extreme point of potential energy, the decay curve of potential energy value is traced along multiple radial directions. Based on the characteristics of potential energy value change in the decay curve, the stress influence boundary points in each radial direction are determined. Multiple stress influence boundary points are connected to form a closed boundary as the boundary of the stress influence domain. Calculate the radial projection components of the potential energy gradient vector at each point on the boundary of the stress influence domain, sum the radial projection components to obtain the boundary diffusion rate vector, and calculate the magnitude of the boundary diffusion rate vector as the boundary diffusion rate. The boundary diffusion rate is mapped to the corresponding position on the base plate surface using the spatial coordinates of the stress influence domain, and this is used as the deformation sensitivity.

[0008] Using the interlaminar shear stress distribution as a load source, a three-dimensional stress potential energy function is constructed. Spatial integration of this function yields the stress potential energy field, which includes: The interlayer shear stress distributed at each spatial location on the base plate surface is decomposed into two mutually orthogonal shear stress components, and a shear stress vector field is constructed based on the two shear stress components at each spatial location. Extract the elastic modulus and Poisson's ratio of each layer from the material property parameters, and calculate the flexibility tensor of each layer based on the elastic modulus and Poisson's ratio; Calculate the stress jump between adjacent layers in the thickness direction of the base plate shear stress vector field, identify the interface coupling location of stress transmission based on the stress jump, and introduce an interface strain energy correction term at the interface coupling location. The shear stress vector field is accumulated layer by layer along the thickness direction of the base plate to obtain the cumulative shear stress vector. The cumulative shear stress vector is combined with the interface strain energy correction term to form a load source. A three-dimensional stress potential energy function is constructed based on the load source and the compliance tensor of each layer. The three-dimensional stress potential energy function is spatially integrated along two orthogonal directions on the surface of the base plate and along the thickness direction of the base plate. The potential energy contribution at each spatial location is accumulated to obtain the stress potential energy field.

[0009] The base plate surface image is divided into multiple observation windows. The spatial topology of the positioning markers within each observation window is extracted, and the spatial topology is encoded into feature vectors. Based on the spatial distribution of deformation sensitivity, deformation-stable regions are identified. The feature vectors corresponding to the observation windows within the deformation-stable regions are selected to form a stable feature set, including: Multiple observation windows are obtained by dividing the image of the base plate surface into a grid, and the center coordinates of the observation windows are recorded. Detect the positioning markers within the observation window and extract their position coordinates. Calculate spatial relationship parameters and construct a spatial topology based on the position coordinates. Extract topological feature parameters from the spatial topology and encode them into feature vectors. Calculate the spatial distance between observation windows based on the center coordinates of the observation windows, identify adjacent observation windows based on the spatial distance, and calculate the gradient value of the feature vectors by calculating the difference in feature vectors between the observation window and its adjacent observation windows. Read the spatial distribution data of deformation sensitivity and extract the corresponding deformation sensitivity value based on the center coordinates of the observation window; Calculate the correlation coefficient between the gradient value of the feature vector and the value of the deformation sensitivity, identify the observation window with a negative correlation coefficient, determine the spatial region where the identified observation window is located as the deformation stable region, and extract the feature vectors corresponding to the observation window in the deformation stable region to form a stable feature set.

[0010] The initial pose offset is obtained by matching the stable feature set with the topology in the design coordinate data. The compensation displacement and compensation angle are then calculated based on the initial pose offset, including: Extract the design location coordinates of the positioning markers from the design coordinate data, and extract the center coordinates and deformation sensitivity values ​​of the observation window corresponding to each feature vector in the stable feature set. Calculate the topological similarity between the center coordinates of the observation window and the coordinates of the design location. Then, weight the topological similarity according to the deformation sensitivity value to obtain the deformation-corrected similarity. Finally, identify the matching coordinate correspondence based on the deformation-corrected similarity. Extract the center coordinates of the observation window and the design position coordinates based on the matched coordinate correspondence. Calculate the positional difference between the center coordinates of the observation window and the design position coordinates to obtain the position offset vector. Calculate the offset confidence level based on the deformation sensitivity value of the observation window corresponding to the position offset vector. The position offset vector is weighted and averaged according to the offset confidence to obtain the weighted average offset vector. The weighted average offset vector is then decomposed into principal component to obtain translation and rotation components. The translation and rotation components are then determined as the initial pose offset. The position compensation amount of the base plate is calculated as the compensation displacement based on the translation component in the initial pose offset, and the angle compensation amount of the base plate is calculated as the compensation angle based on the rotation component in the initial pose offset.

[0011] The stress transmission path is constructed by identifying the continuous region with the largest potential energy gradient in the stress potential energy field. Pressing contact points are then set along the stress transmission path. The activation sequence of the pressing contact points, generated according to the direction of decreasing potential energy, includes: In the stress potential energy field, the potential energy gradient vector is calculated, and the spatial position with the largest potential energy gradient vector magnitude is extracted as the stress transmission source point. From the stress transmission source point, the spatial positions where the potential energy gradient vector magnitude is continuously greater than the preset gradient threshold are traced along the direction of the potential energy gradient vector and connected to form a stress transmission path. Set up pressing contact points along the stress transmission path, and extract the spatial position and corresponding potential energy value of the pressing contact points in the stress potential energy field; The initial activation sequence is obtained by sorting the pressing contact points according to the decreasing potential energy based on the potential energy value. The spatial position of the first pressing contact point in the initial activation sequence is extracted and marked as the pressing constraint region. Pressing boundary conditions are applied to the pressing constraint region and the stress potential energy field is recalculated to obtain the updated stress potential energy field. In the updated stress potential energy field, the potential energy values ​​of the remaining pressing contact points are extracted again and the initial activation sequence is updated. This process is repeated iteratively until all pressing contact points are sorted and the activation sequence is obtained.

[0012] The control pressing module adjusts its position and attitude according to the compensation displacement and compensation angle, and drives the pressing contact points to complete the lamination alignment according to the activation sequence, including: The translational motion trajectory of the pressing module is established based on the compensation displacement, and the rotational motion trajectory of the pressing module is established based on the compensation angle. The translational motion trajectory and the rotational motion trajectory are combined to obtain the six-degree-of-freedom motion trajectory of the pressing module, and the pressing module is controlled to adjust its position and attitude according to the six-degree-of-freedom motion trajectory. Based on the activation timing, the spatial location and activation order of the pressing contact points are extracted. The pressing contact points that are ranked earlier in the activation timing are marked as priority pressing points, and the pressing contact points that are ranked later in the activation timing are marked as delayed pressing points. The control module moves to the spatial position corresponding to the priority pressing point, drives the pressing contact point to apply pressing force at the priority pressing point and maintain the pressing state, and establishes the pressing constraint of the priority pressing point; Under the pressure constraint of the priority pressing point, the control pressing module moves sequentially to the spatial position corresponding to the delayed pressing point, and drives the pressing contact point to apply pressing force sequentially at the delayed pressing point until all pressing contact points in the activation sequence are pressed, thereby achieving lamination alignment.

[0013] A second aspect of the present invention provides a high-performance substrate lamination intelligent alignment system, comprising: The data acquisition unit is used to acquire images of the base plate surface, material property parameters of each layer of the base plate, and design coordinate data of the target overlay layer; The stress calculation unit is used to calculate the interlayer shear stress distribution based on the difference in thermal expansion coefficients of each layer in the material property parameters and convert it into a stress potential energy field. It identifies the local extreme points of potential energy in the stress potential energy field and constructs a stress influence domain with the local extreme points of potential energy as the center. It calculates the boundary diffusion rate of the stress influence domain as the deformation sensitivity. The feature extraction unit is used to divide the base plate surface image into multiple observation windows, extract the spatial topology of the positioning marks in each observation window, encode the spatial topology into feature vectors, identify the deformation stable region based on the spatial distribution of deformation sensitivity, and select the feature vectors corresponding to the observation windows in the deformation stable region to form a stable feature set. The offset calculation unit is used to match the stable feature set with the topology in the design coordinate data to obtain the initial pose offset, and calculate the compensation displacement and compensation angle based on the initial pose offset. The path planning unit is used to identify the continuous region with the largest potential energy gradient from the stress potential energy field to form a stress transmission path, set the pressing contact point along the stress transmission path, and generate the activation sequence of the pressing contact point according to the direction of decreasing potential energy. The pressing execution unit is used to control the pressing module to adjust its position and posture according to the compensation displacement and compensation angle, and to drive the pressing contact points to complete the lamination alignment according to the activation sequence.

[0014] 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.

[0015] 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.

[0016] This invention effectively quantifies the deformation trend caused by the thermal expansion differences among the layers of the substrate by constructing a stress potential energy field and identifying local extreme points of potential energy. By selecting a stable feature set based on deformation sensitivity, it avoids the distortion of positioning markers caused by severe local deformation within the observation window, significantly improving the accuracy of initial pose offset calculation. Based on the dynamic adjustment of compensation displacement and compensation angle, real-time compensation for the thermal deformation error of the substrate is achieved, enabling alignment accuracy to reach the sub-pixel level, far exceeding traditional fixed-reference alignment methods. By setting pressing contact points along the stress transmission path and generating activation sequences according to the direction of decreasing potential energy, the physical process of stress release is precisely matched. This guides the pressing pressure to be gradually transmitted along the stress gradient direction, avoiding local warping or delamination caused by stress concentration. The pressing module, based on the synergistic effect of the attitude adjusted by the pose offset and the activation sequence, ensures that each layer of material maintains uniform contact during lamination, significantly reducing the peak value of interlayer shear stress and improving the overall stability of the laminated structure. Closed-loop optimization from sensor data to drive control is achieved. Deformation stability zone screening reduces computational complexity while ensuring robustness of feature matching. Stress transfer path identification provides a path planning basis for the pressing process, reducing invalid pressing actions. The overall alignment process improves accuracy while shortening adjustment time, making it suitable for high-reliability manufacturing scenarios involving multi-layer, high-computation base plates. Attached Figure Description

[0017] Figure 1 A flowchart illustrating the intelligent alignment method for high-performance substrate lamination; Figure 2 Flowchart for extracting positioning marker coordinates and compensating for base plate pose. Detailed Implementation

[0018] 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.

[0019] 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.

[0020] Figure 1 This is a flowchart illustrating the intelligent alignment method for high-performance base plate lamination according to an embodiment of the present invention.

[0021] The high-computation-capacity substrate lamination intelligent alignment method includes: Acquire images of the base plate surface, material property parameters of each layer of the base plate, and design coordinate data of the target overlay layer; The interlayer shear stress distribution is calculated based on the difference in thermal expansion coefficients of each layer in the material property parameters and converted into a stress potential energy field. Local extreme points of potential energy in the stress potential energy field are identified, and a stress influence domain is constructed with the local extreme points of potential energy as the center. The boundary diffusion rate of the stress influence domain is calculated as the deformation sensitivity. The base plate surface image is divided into multiple observation windows. The spatial topology of the positioning marks in each observation window is extracted and encoded into feature vectors. Based on the spatial distribution of deformation sensitivity, the deformation stability region is identified, and the feature vectors corresponding to the observation windows in the deformation stability region are selected to form a stable feature set. The initial pose offset is obtained by matching the stable feature set with the topology in the design coordinate data, and the compensation displacement and compensation angle are calculated based on the initial pose offset. The stress transmission path is formed by identifying the continuous region with the largest potential energy gradient in the stress potential energy field, setting the pressing contact point along the stress transmission path, and generating the activation sequence of the pressing contact point according to the direction of decreasing potential energy. The control pressing module adjusts its position and posture according to the compensation displacement and compensation angle, and drives the pressing contact points to complete the lamination alignment according to the activation sequence.

[0022] Based on the differences in the thermal expansion coefficients of each layer in the material properties parameters, the interlayer shear stress distribution is calculated and converted into a stress potential energy field. Local extrema of the potential energy in the stress potential energy field are identified, and a stress influence domain is constructed with the local extrema of the potential energy as the center. The boundary diffusion rate of the stress influence domain is calculated as the deformation sensitivity, including: The shear strain distribution at the interface of each layer is calculated based on the difference in thermal expansion coefficient between adjacent layers and the amount of temperature change in the material property parameters. The interlayer shear stress distribution is then calculated based on the shear strain distribution and the shear modulus of each layer. The interlaminar shear stress distribution is used as a load source to construct a three-dimensional stress potential energy function, and the stress potential energy field is obtained by spatial integration of the three-dimensional stress potential energy function. In the stress potential energy field, calculate the spatial gradient distribution of potential energy values, and identify the spatial locations where the potential energy gradient is zero and the potential energy values ​​satisfy the local maximum condition as the local extreme points of potential energy. Starting from the local extreme point of potential energy, the decay curve of potential energy value is traced along multiple radial directions. Based on the characteristics of potential energy value change in the decay curve, the stress influence boundary points in each radial direction are determined. Multiple stress influence boundary points are connected to form a closed boundary as the boundary of the stress influence domain. Calculate the radial projection components of the potential energy gradient vector at each point on the boundary of the stress influence domain, sum the radial projection components to obtain the boundary diffusion rate vector, and calculate the magnitude of the boundary diffusion rate vector as the boundary diffusion rate. The boundary diffusion rate is mapped to the corresponding position on the base plate surface using the spatial coordinates of the stress influence domain, and this is used as the deformation sensitivity.

[0023] After obtaining the material property parameters of each layer of the base plate, the shear strain distribution at the interface of each layer is calculated based on the difference in thermal expansion coefficients and temperature changes between adjacent layers. Specifically, for the first... Layer and First The interface between the layers has thermal expansion coefficients of the two materials as follows: and When the temperature change is Under these conditions, the thermal mismatch strain at the interface is The thermal mismatch strain is distributed in a shear manner within the interface plane, forming a shear strain field. Because the temperature distribution in different areas of the base plate is not completely uniform, It has spatial distribution characteristics within the plane of the base plate, therefore the shear strain distribution Similarly, it exhibits a non-uniform distribution within the plane of the base plate. After performing the above calculations on all adjacent layer interfaces of the base plate, based on the shear modulus of each layer... ,according to Calculate the interlaminar shear stress distribution at each interface. By comprehensively superimposing the shear stress distributions of all interfaces, a complete dataset of interlayer shear stress distributions is obtained, providing load input for the subsequent construction of the stress potential energy field.

[0024] Using interlaminar shear stress distribution as the load source, a three-dimensional stress potential energy function is constructed. This function uses the three-dimensional spatial coordinates of the base plate. Using the independent variable, the shear stress distribution at each interface is mapped to the corresponding spatial depth, and the degree of local stress accumulation is expressed in the form of elastic potential energy density. For the three-dimensional stress potential energy function... Spatial integration is performed over the entire volume domain of the base plate to obtain the stress potential energy field. The stress potential energy field reflects the energy accumulation state of the substrate at various spatial locations during the lamination process. Regions with higher potential energy values ​​correspond to locations with greater material deformation risk, while regions with lower potential energy values ​​are relatively stable. Through a comprehensive analysis of the stress potential energy field, the distribution law of deformation sensitivity of each region of the substrate under thermo-mechanical coupling conditions can be intuitively understood.

[0025] In the stress potential energy field In the calculation of the spatial gradient distribution of potential energy values The gradient distribution describes the rate of change of the potential energy field in various spatial directions. A location with a gradient of zero means that the potential energy value at that point does not increase or decrease with changes in spatial coordinates, and is considered a candidate point for potential energy extrema. Among the candidate points that satisfy the zero gradient condition, it is further determined whether their potential energy value satisfies the local maximum condition, that is, within a neighborhood centered on that point, the potential energy value at that point is not lower than the potential energy value of any other point in the neighborhood. Spatial locations that satisfy both of these conditions are identified as potential energy local extrema points. subscript The extreme points are numbered. The local extreme points of potential energy represent the core locations where the base plate is most prone to deformation concentration under the action of lamination thermal stress, and are the basis for the subsequent construction of the stress influence domain.

[0026] For each local extremum of potential energy Starting from the radial direction, along multiple radial directions (in (Numbered in the radial direction) Tracking potential energy values The decay curve shows that along each radial direction, the potential energy value gradually decreases as the distance from the extreme point increases. Based on the potential energy value change characteristics of the decay curve, when the potential energy value decreases to a set threshold or the decay rate shows a significant inflection point, that location is determined as the stress influence boundary point in that radial direction. For extreme points After determining the boundary points for all radial directions, all boundary points... Connecting them sequentially forms a closed boundary curve of the stress influence domain; the spatial region enclosed by this closed boundary is the stress influence domain. The stress influence domains corresponding to different extreme points may have different shapes, which reflects the anisotropic deformation characteristics of different regions of the base plate under different material combinations and geometric constraints.

[0027] After determining the boundary of the stress influence domain, calculate the potential energy gradient vector at each point on the boundary curve. In the corresponding radial direction Projected components This projected component characterizes the magnitude of the radial driving force for the diffusion of the potential energy field at the boundary. For extreme points... Projection components in all radial directions Perform vector summation to obtain the boundary diffusion rate vector. Its direction reflects the dominant direction of the overall diffusion of stress influence domain, and its modulus This is the boundary diffusion rate. The physical meaning of the boundary diffusion rate is to describe the speed at which the stress-affected domain expands outward under continuous thermal load. The larger the diffusion rate, the stronger the deformation dynamics of the area around the extreme point during the lamination process, and the more significant the threat to the lamination alignment accuracy.

[0028] Each extreme point Corresponding boundary diffusion rate Through the stress influence domain The spatial coordinates are mapped to the corresponding positions on the substrate surface, completing the assignment of deformation sensitivity values ​​from the three-dimensional stress potential energy field to the two-dimensional surface of the substrate. Specifically, the mapping method is as follows: the projection area of ​​the stress influence domain onto the normal direction of the substrate surface is used as the mapping target, and the boundary diffusion rate is used as the deformation sensitivity value of each point within this projection area, forming a deformation sensitivity distribution map of the substrate surface. For locations on the substrate surface simultaneously covered by multiple stress influence domains, the maximum value of the boundary diffusion rate corresponding to each influence domain is taken as the final deformation sensitivity at that location, ensuring sufficient identification of high-risk areas. The deformation sensitivity distribution map directly reflects the probability and magnitude of deformation at various locations on the substrate surface under lamination thermal stress, providing a quantitative basis for subsequent division of deformation-stable zones and selection of stable observation windows, ensuring that the final lamination alignment operation establishes a reliable pose reference benchmark within the area with minimal deformation influence.

[0029] Using the interlaminar shear stress distribution as a load source, a three-dimensional stress potential energy function is constructed. Spatial integration of this function yields the stress potential energy field, which includes: The interlayer shear stress distributed at each spatial location on the base plate surface is decomposed into two mutually orthogonal shear stress components, and a shear stress vector field is constructed based on the two shear stress components at each spatial location. Extract the elastic modulus and Poisson's ratio of each layer from the material property parameters, and calculate the flexibility tensor of each layer based on the elastic modulus and Poisson's ratio; Calculate the stress jump between adjacent layers in the thickness direction of the base plate shear stress vector field, identify the interface coupling location of stress transmission based on the stress jump, and introduce an interface strain energy correction term at the interface coupling location. The shear stress vector field is accumulated layer by layer along the thickness direction of the base plate to obtain the cumulative shear stress vector. The cumulative shear stress vector is combined with the interface strain energy correction term to form a load source. A three-dimensional stress potential energy function is constructed based on the load source and the compliance tensor of each layer. The three-dimensional stress potential energy function is spatially integrated along two orthogonal directions on the surface of the base plate and along the thickness direction of the base plate. The potential energy contribution at each spatial location is accumulated to obtain the stress potential energy field.

[0030] After obtaining the interlaminar shear stress distribution at each interface, it is necessary to use it as a load source to construct a potential energy function that can describe the three-dimensional stress state of the base plate, and then obtain the complete stress potential energy field through spatial integration. The specific process begins with the decomposition of shear stress components. At each spatial location on the base plate surface, the interlaminar shear stress at that location is decomposed along two mutually orthogonal horizontal directions, yielding the stress along... shear stress components in the direction Along shear stress components in the direction The two components together constitute the shear stress vector at that location. After traversing all spatial locations on the base plate surface, the complete shear stress vector field can be obtained. The significance of this step is to transform the scalar form of interlaminar shear stress into a vector field with directional information, providing a basis for subsequent cumulative calculations along the thickness direction.

[0031] Extract the elastic modulus of each layer from the material property parameters. Compared with Poisson subscript The layer number is indicated. For isotropic materials, the compliance tensor is... The compliance tensor, which can be directly derived from the elastic modulus and Poisson's ratio, describes the material's ability to produce strain under unit stress and is a key material parameter for constructing the potential energy function. For anisotropic material layers (such as fiber-reinforced layers or copper foil layers), each component of the compliance tensor needs to be calculated separately based on the principal elastic constants of the material to ensure accurate characterization of the differences in mechanical response between different material layers. The introduction of the compliance tensor allows the potential energy function to distinguish the contribution weights of different material layers to the overall stress state, avoiding the errors caused by treating all layers as homogeneous materials.

[0032] After obtaining the shear stress vector field, it is necessary to calculate the stress jump between adjacent layers. Along the thickness direction of the base plate, adjacent layers... and At the interface between the two materials, the shear stress vector field often exhibits discontinuity along the thickness direction. This discontinuity reflects the abrupt change in stress transmission at the interface. (The last sentence appears to be incomplete and possibly contains errors.) The shear stress vector at the layer interface and the first The stress jump is obtained by subtracting the shear stress vectors at the layer interface. .when When the threshold is exceeded, the interface is determined to be a stress-transfer interface coupling location. Identifying interface coupling locations is of great significance because at these locations, there is a significant mechanical mismatch between adjacent layers. The potential energy function constructed solely based on the continuous medium assumption cannot accurately reflect the actual stress state, and additional correction terms need to be introduced for compensation.

[0033] After identifying the interface coupling locations, interface strain energy correction terms are introduced at these locations. The physical meaning of the interfacial strain energy correction term is the additional elastic strain energy stored at the interface due to stress jumps. Its magnitude is related to the square of the stress jump and the equivalent compliance of the interfacial region. Specifically, Proportional to Equivalent flexibility coefficient of interface The product of, where It is obtained by weighted averaging of the compliance tensors of the materials on both sides of the interface. After introducing the interface strain energy correction term, the potential energy function can capture the local energy concentration effect at the interface coupling position, thus more accurately reflecting the true deformation risk distribution of the substrate during the lamination process.

[0034] The shear stress vector field is accumulated layer by layer along the thickness direction of the base plate, from the first layer at the bottom to the [missing information]. The cumulative shear stress vector is obtained by sequentially superimposing all the shear stress vectors of the layers. During the accumulation process, at each interface coupling point, the corresponding interface strain energy correction term will be... This is superimposed on the current accumulated load to ensure that the energy abrupt change at the interface is included in the construction of the overall load source. Finally, the accumulated shear stress vector after full-thickness accumulation is combined with all interface strain energy correction terms to form a comprehensive load source. The comprehensive load source includes both the overall effect of shear stress in each layer of the base plate and the local energy correction at the interface coupling points, which can comprehensively reflect the stress source of the base plate under thermal load.

[0035] Based on the comprehensive load source With the flexibility tensor of each layer Construct a three-dimensional stress potential energy function At every spatial location place, The value of is jointly determined by the compliance tensor of the layer at that location and the combined load source at that location, reflecting the elastic potential energy density stored in the base plate at that spatial location. A material layer with a larger compliance tensor stores higher potential energy under the same load, but also carries a greater risk of deformation at that location. The construction of the three-dimensional stress potential energy function unifies material anisotropy, interface coupling effects, and interlayer shear loads into the same energy framework, providing a complete integrand for subsequent potential field integration.

[0036] For three-dimensional stress potential function Along the surface of the base plate respectively direction, Direction and base plate thickness By performing spatial integration along the direction, the potential energy density contributions at each spatial location are accumulated and superimposed to finally obtain the stress potential energy field. The physical meaning of spatial integration is to expand point-like potential energy density information into a spatially continuous potential energy distribution, allowing the overall deformation risk of each region on the substrate to be intuitively represented in the form of a potential energy field. In numerical implementation, the substrate space is discretized using finite difference or finite element meshes, and calculations are performed at each mesh node. The value of the stress potential energy field is obtained by performing numerical integration in three directions sequentially using methods such as trapezoidal integration or Simpson integration. After integration, the stress potential energy field forms a continuous potential energy distribution map on the base plate. Regions with higher potential energy correspond to locations with greater material deformation risk, providing a quantitative basis for subsequent identification of local potential energy extrema, construction of stress influence domains, and determination of deformation stability zones.

[0037] In practical applications, the base plate may contain dozens of layers of different materials, each with a different elastic modulus. Compared with Poisson Significant differences lead to highly uneven distribution of the compliance tensor between layers. By employing the aforementioned layer-by-layer accumulation and interface correction methods, the distortion of the potential energy field caused by neglecting the interface coupling effect can be effectively avoided. This ensures that the subsequent identification of deformation stability zones and the planning of the lamination sequence based on the stress potential energy field have sufficient accuracy, thereby guaranteeing the stability and reliability of the lamination alignment process.

[0038] The base plate surface image is divided into multiple observation windows. The spatial topology of the positioning markers within each observation window is extracted, and the spatial topology is encoded into feature vectors. Based on the spatial distribution of deformation sensitivity, deformation-stable regions are identified. The feature vectors corresponding to the observation windows within the deformation-stable regions are selected to form a stable feature set, including: Multiple observation windows are obtained by dividing the image of the base plate surface into a grid, and the center coordinates of the observation windows are recorded. Detect the positioning markers within the observation window and extract their position coordinates. Calculate spatial relationship parameters and construct a spatial topology based on the position coordinates. Extract topological feature parameters from the spatial topology and encode them into feature vectors. Calculate the spatial distance between observation windows based on the center coordinates of the observation windows, identify adjacent observation windows based on the spatial distance, and calculate the gradient value of the feature vectors by calculating the difference in feature vectors between the observation window and its adjacent observation windows. Read the spatial distribution data of deformation sensitivity and extract the corresponding deformation sensitivity value based on the center coordinates of the observation window; Calculate the correlation coefficient between the gradient value of the feature vector and the value of the deformation sensitivity, identify the observation window with a negative correlation coefficient, determine the spatial region where the identified observation window is located as the deformation stable region, and extract the feature vectors corresponding to the observation window in the deformation stable region to form a stable feature set.

[0039] When dividing the image of the base plate surface into grids, a uniform rectangular grid is used to divide the entire image into several observation windows of the same size. Each observation window has a unique center coordinate in the image coordinate system. The coordinates are recorded synchronously with the grid division results and stored as the spatial index information of the observation window. The granularity of the grid division needs to be determined comprehensively based on the base plate size and the distribution density of the positioning marks. An excessively large window will cause multiple sets of positioning marks to be included in a single window, resulting in topological confusion, while an excessively small window may cause positioning marks to fall near the window edge, resulting in missed detections. In practice, the window side length can be set to 0.8 to 1.2 times the distance between adjacent positioning marks to ensure that each observation window contains at least one complete positioning mark area.

[0040] Within each observation window, the presence of location markers is detected through image binarization and connected component analysis, and the centroid pixel coordinates of each location marker are extracted as its position coordinates. Using the position coordinates of all location markers within the window as nodes, the Euclidean distance and azimuth between any two nodes are calculated, and these distances and azimuths together constitute the spatial relationship parameters of node pairs. Based on these spatial relationship parameters, a spatial topology describing the relative layout of location markers within the window is constructed, with nodes as vertices and node pairs as edges. Topological feature parameters are extracted from the spatial topology, including the mean distance between nodes, the standard deviation of the distance, the entropy value of the azimuth distribution, and the nearest neighbor distance ratio, etc. These topological feature parameters are concatenated in a fixed order and encoded into a fixed-dimensional feature vector. This is used for subsequent matching and comparison. The dimension of the feature vector remains consistent throughout the entire processing flow to ensure that features between different observation windows can be directly compared numerically.

[0041] When identifying adjacent observation windows, the center coordinates of each observation window are used. Using this as a reference, calculate the spatial distance between the centers of any two observation windows. ,in and These are the index numbers of the two observation windows, respectively. Spatial distance... Less than the preset neighborhood radius threshold The windows are determined to be adjacent. Typically, the grid cell side length is 1.5 times the grid cell side length to cover directly adjacent windows in the top, bottom, left, right, and diagonal directions. For each pair of adjacent windows... and Calculate its eigenvectors and The gradient value of the eigenvector is obtained by measuring the Euclidean distance between them. This reflects the magnitude of local changes in the spatial topology of the positioning markers within two adjacent observation windows. A larger value indicates a significant change in the topology between the two windows, implying local deformation in that region. A smaller value indicates that the topology of adjacent windows tends to be consistent, and the deformation in that region is relatively stable. To comprehensively describe the degree of local topological change in a single observation window, the window... The average gradient values ​​of the feature vectors of all its neighboring windows are used to obtain the window. The combined gradient value .

[0042] Read the spatial distribution data of deformation sensitivity calculated in the previous steps. This data records the deformation sensitivity value at each spatial location, indexed by the plane coordinates of the base plate. The larger the value, the stronger the deformation response at that location caused by thermal expansion. This is based on the center coordinates of each observation window. The corresponding deformation sensitivity values ​​are extracted from the deformation sensitivity distribution data using bilinear interpolation. This allows each observation window to obtain a quantified value of deformation sensitivity that matches its spatial location. The interpolation operation can compensate for the spatial discretization error between the deformation sensitivity data and the center coordinates of the observation window, improving the accuracy of sensitivity assignment.

[0043] To obtain the combined gradient value for each observation window And the deformation sensitivity value Then, the Pearson correlation coefficient between the two was calculated across all observation windows. This is used to quantify the degree of linear correlation between the gradient of the feature vector and the deformation sensitivity. The range of values ​​is ,when When the value is negative, it indicates that the eigenvector gradient of regions with high deformation sensitivity is relatively small. This is consistent with the expected physical law—regions with high deformation sensitivity are usually located within the stress influence domain, and local deformation tends to be homogenized, reducing the topological differences between adjacent windows. Conversely, regions with low deformation sensitivity, i.e., deformation-stable regions, also show relatively small changes in the topological structure of their positioning markers between adjacent windows. Both exhibit a negative correlation characteristic. The set of observation windows that satisfy the negative correlation condition, i.e., deformation sensitivity... Below the global mean and combined gradient value Similarly, observation windows that are also below the global mean are used to define the continuous spatial regions covered by these windows on the base plate plane as deformation-stable zones.

[0044] The determination of the deformation-stable region employs spatial connectivity constraints, requiring adjacent observation windows satisfying the aforementioned double-low conditions to form a spatially connected set. Isolated single windows do not constitute a deformation-stable region independently, thus eliminating misjudgments caused by occasional noise. A minimum threshold is set for the area of ​​the connected set, eliminating candidate regions with excessively small areas and retaining stable regions with sufficient coverage as the final deformation-stable region. Feature vectors corresponding to all observation windows within the deformation-stable region are extracted. These features are aggregated to form a stable feature set. The feature vectors in the stable feature set are derived from the spatial region on the substrate where the deformation behavior is most stable. Its topology is least affected by thermal expansion stress and can most realistically reflect the design layout relationship of the positioning marks. This provides a highly reliable input for subsequent topology matching with the design coordinate data, thereby effectively suppressing alignment errors introduced by interlayer stress deformation and improving the accuracy and stability of lamination alignment.

[0045] like Figure 2 As shown, Figure 2 This is a flowchart illustrating the operation of extracting positioning marker coordinates and compensating for base plate pose in an embodiment of the present invention.

[0046] The initial pose offset is obtained by matching the stable feature set with the topology in the design coordinate data. The compensation displacement and compensation angle are then calculated based on the initial pose offset, including: Extract the design location coordinates of the positioning markers from the design coordinate data, and extract the center coordinates and deformation sensitivity values ​​of the observation window corresponding to each feature vector in the stable feature set. Calculate the topological similarity between the center coordinates of the observation window and the coordinates of the design location. Then, weight the topological similarity according to the deformation sensitivity value to obtain the deformation-corrected similarity. Finally, identify the matching coordinate correspondence based on the deformation-corrected similarity. Extract the center coordinates of the observation window and the design position coordinates based on the matched coordinate correspondence. Calculate the positional difference between the center coordinates of the observation window and the design position coordinates to obtain the position offset vector. Calculate the offset confidence level based on the deformation sensitivity value of the observation window corresponding to the position offset vector. The position offset vector is weighted and averaged according to the offset confidence to obtain the weighted average offset vector. The weighted average offset vector is then decomposed into principal component to obtain translation and rotation components. The translation and rotation components are then determined as the initial pose offset. The position compensation amount of the base plate is calculated as the compensation displacement based on the translation component in the initial pose offset, and the angle compensation amount of the base plate is calculated as the compensation angle based on the rotation component in the initial pose offset.

[0047] The design position coordinates of each positioning marker are extracted from the design coordinate data. Simultaneously, the center coordinates and deformation sensitivity values ​​of the observation window corresponding to each feature vector in the stable feature set are extracted. The design position coordinates are derived from the overlay layout data, recording the precise spatial position of each positioning marker under ideal, deformation-free conditions. The center coordinates of the observation window in the stable feature set reflect the detection position of each positioning marker in the actual scanned image of the substrate. Because the substrate may experience local deformation due to factors such as thermal and mechanical stress during processing and storage, there is a systematic deviation between the two sets of coordinates. Therefore, a correspondence between the two sets of coordinates needs to be established through topological matching.

[0048] When calculating the topological similarity between the center coordinates of the observation window and the design location coordinates, the relative spatial distribution of each positioning marker in its respective coordinate system is used as the matching basis, rather than directly comparing absolute coordinate values. Specifically, taking each positioning marker as the center, the distance ratio, angle distribution, and connectivity between it and its neighboring positioning markers are statistically analyzed to form a local topological descriptor. Then, the similarity score between the local topological descriptor of each observation window in the stable feature set and the corresponding descriptor in the design coordinate data is calculated, denoted as... ,in For the index of the observation window of the stable feature set, To design the index of the positioning markers in the coordinate data.

[0049] Obtaining the initial topological similarity Then, a deformation sensitivity value is introduced for weighted correction. Deformation sensitivity value This reflects the degree of local deformation at the location of the observation window. Higher deformation sensitivity indicates stronger stress disturbance at that location, potentially leading to a larger deviation between the detected and true coordinates, and consequently, lower reliability of the topological similarity. Therefore, a deformation correction weight is introduced. , defined as a monotonically decreasing function of deformation sensitivity, meaning that the lower the deformation sensitivity, the higher the weight. and Multiplication yields deformation-corrected similarity. ,by The maximum value corresponding to The combination serves as a matching coordinate correspondence, thereby completing a one-to-one correspondence between the observation window in the stable feature set and the positioning marker in the design coordinates.

[0050] Based on the matching coordinate correspondence, extract the center coordinates of the observation window in each pair of matching points. Corresponding design location coordinates Calculate the difference between the two to obtain the position offset vector of the matching pair. subscript This is the index number for the matching pair. The position offset vector reflects the actual displacement deviation of the base plate in this local area. The offset vectors at different positions may differ in direction and magnitude. This difference stems from both the rigid body motion (translation and rotation) of the base plate as a whole and may also include contributions from local elastic deformation.

[0051] To reasonably distinguish rigid body motion components from deformation noise in the offset vectors of multiple matched pairs, different confidence weights need to be assigned to the offset vectors of each matched pair. For each matched pair, the confidence weight is determined based on the deformation sensitivity value at the center of its corresponding observation window. Calculate the offset confidence level It is defined as a normalized inverse proportional mapping of deformation sensitivity. The lower the deformation sensitivity of the matching pair, the more likely its offset vector is to accurately reflect the rigid body motion of the base plate, and the higher the confidence level. The higher the confidence level, the better; conversely, areas with higher deformation sensitivity are more susceptible to local deformation interference, resulting in lower confidence levels. The weighting is lower. In this way, matching pairs in the deformation-stable region receive a higher weight contribution in subsequent calculations, while the influence of matching pairs in the deformation-active region is effectively suppressed.

[0052] Based on the offset confidence of each matching pair For position offset vector Perform a weighted average to obtain the weighted average offset vector. The calculation method is as follows The weighted average offset vector integrates the main trends of the overall offset of the base plate, eliminates random noise introduced by local deformation, and provides a more reliable input for subsequent pose decomposition.

[0053] Principal component decomposition is performed on the weighted average offset vector, decomposing it into translation and rotation components. The translation component corresponds to the overall linear displacement of the base plate in the plane, expressed as a two-dimensional vector. It means that, among them and These represent the translations along the horizontal and vertical directions, respectively. The rotation component corresponds to the overall angular deflection of the base plate around its center point, expressed as a rotation angle. The specific implementation of principal component decomposition is as follows: A covariance matrix is ​​constructed using the coordinate difference vectors of each matching pair. Then, through singular value decomposition or least squares rotation estimation, the overall offset is decomposed into two independent components: rigid body translation and rigid body rotation, thus obtaining the initial pose offset, which is determined by the translation component. With rotational components Together they constitute.

[0054] Based on the translation component in the initial pose offset Calculate the positional compensation of the base plate at the pressing station, and... The inverse vector is used as the compensation displacement, that is, the pressing module needs to be along... direction and The directions are shifted by corresponding distances to compensate for the linear deviation between the detected position and the designed position of the base plate. This is based on the rotational component in the initial pose offset. Calculate the angle compensation amount of the base plate, and As a compensation angle, the pressing module needs to rotate in the opposite direction around the center point of the base plate. Angle adjustment is used to eliminate angular deflection errors of the base plate. Compensation displacement and compensation angle together constitute the posture adjustment command for the lamination module to perform alignment actions, ensuring that the cover layer can accurately fall on the design coordinate position of the base plate during lamination, meeting the process requirements of high-precision lamination alignment.

[0055] In practical applications, when there are many layers in the base plate or the materials are complex, the distribution of deformation sensitivity in different regions may exhibit strong spatial non-uniformity, leading to significant differences in the confidence levels of each matching pair in the stable feature set. By introducing a deformation sensitivity weighting mechanism, the contribution of high-confidence matching pairs can be adaptively highlighted, effectively improving the estimation accuracy of the initial pose offset, thereby ensuring the accuracy of the compensation displacement and compensation angle, and laying a reliable pose foundation for the subsequent execution of the pressing contact point activation sequence.

[0056] The stress transmission path is constructed by identifying the continuous region with the largest potential energy gradient in the stress potential energy field. Pressing contact points are then set along the stress transmission path. The activation sequence of the pressing contact points, generated according to the direction of decreasing potential energy, includes: In the stress potential energy field, the potential energy gradient vector is calculated, and the spatial position with the largest potential energy gradient vector magnitude is extracted as the stress transmission source point. From the stress transmission source point, the spatial positions where the potential energy gradient vector magnitude is continuously greater than the preset gradient threshold are traced along the direction of the potential energy gradient vector and connected to form a stress transmission path. Set up pressing contact points along the stress transmission path, and extract the spatial position and corresponding potential energy value of the pressing contact points in the stress potential energy field; The initial activation sequence is obtained by sorting the pressing contact points according to the decreasing potential energy based on the potential energy value. The spatial position of the first pressing contact point in the initial activation sequence is extracted and marked as the pressing constraint region. Pressing boundary conditions are applied to the pressing constraint region and the stress potential energy field is recalculated to obtain the updated stress potential energy field. In the updated stress potential energy field, the potential energy values ​​of the remaining pressing contact points are extracted again and the initial activation sequence is updated. This process is repeated iteratively until all pressing contact points are sorted and the activation sequence is obtained.

[0057] In the stress potential energy field Based on the established foundation, calculate the potential energy gradient vector at each spatial location within the field. and extract The spatial location with the largest modulus is taken as the stress transfer source point. This source point represents the location where stress accumulation is most concentrated and potential energy changes most drastically within the field, and is the initiating region most prone to interlayer misalignment or warping during the lamination process. Starting from, along The direction of the pointing is tracked step by step, requiring that the gradient magnitude at every consecutive spatial location along the tracking path continuously exceed a preset gradient threshold. When the gradient magnitude at a certain location first falls below... When the tracking terminates, it will be from All spatial locations leading to the termination point are connected sequentially to form a stress transfer path. This path, in a physical sense, corresponds to the main channel through which stress energy propagates along the direction of maximum gradient during the lamination process of the base plate, and serves as the spatial basis for the subsequent arrangement of lamination contact points.

[0058] Along the stress transfer path A set of pressing contact points are set according to the arc length parameter of the path, using equal or adaptive spacing. ,in Index number of the press contact point , This represents the total number of pressing contact points. The adaptive spacing is set as follows: On longer path segments, appropriately reduce the spacing between adjacent contact points to improve the pressing accuracy in that area; Increase the spacing appropriately on shorter path segments to reduce the number of unnecessary contact points. Each pressing contact point... In the stress potential energy field, there is a corresponding spatial coordinate. and its corresponding potential energy value Extract all The potential energy values ​​of each pressing contact point constitute a set of potential energy values. This provides a numerical basis for subsequent sorting and activation sequence generation.

[0059] According to the set of potential energy values All pressing contact points are sorted according to the direction of decreasing potential energy, that is, the contact point with the higher potential energy value is placed at the beginning of the activation sequence, thus obtaining the initial activation sequence. ,satisfy The contact point with the highest potential energy. Located in the region where stress accumulation is most concentrated, priority activation can release or constrain the stress in that region, thereby reducing the risk of cumulative deformation during subsequent pressing processes.

[0060] Extract the first and second compression contact points from the initial activation sequence. The spatial location of the area is used to mark the spatial region within a certain radius around it as the compression constraint region. .exist An internal compression boundary condition is applied, specifically by constraining the displacement degrees of freedom of all nodes within the region to zero. This means that after the first contact point completes compression, the interlayer relative displacement in this local region is considered fixed. Under this boundary condition, the stress potential energy field is numerically solved again to obtain an updated stress potential energy field. The updated potential energy field reflects the actual redistribution of stress within the base plate after the first contact point has completed compression: the stress originally concentrated in... The potential energy in the vicinity is partially released, while the potential energy distribution in the remaining area may change due to the introduction of boundary conditions, and the potential energy value at some contact points may increase or decrease compared to the initial state.

[0061] In updating the stress potential energy field In the middle, re-extract the remaining Each pressing contact point The potential energy value at that point is used to obtain the updated potential energy value set. The remaining contact points are then reordered according to the direction of decreasing potential energy, and the activation sequence is updated accordingly. The spatial region corresponding to the first contact point in the updated sequence is marked as the new compression constraint region. Apply the pressure boundary conditions again and solve for the result. This process is repeated iteratively, with each iteration including the following operations: extracting the spatial location of the first contact point in the current active sequence, marking it as the new compression constraint region and applying boundary conditions, recalculating the stress potential energy field, re-extracting the potential energy values ​​of the remaining contact points in the updated field and reordering them to generate a new active sequence. The iteration terminates when all... All pressing contact points have been sorted, meaning that each position in the activation sequence has been uniquely determined.

[0062] After the iterative process is completed, the final activation sequence is obtained. The physical meaning of this activation sequence is that each pressing operation selects the contact point with the highest potential energy under the current stress potential energy field state for priority activation, ensuring that each pressing action intervenes in the current area of ​​maximum stress accumulation, thereby achieving gradual control over the stress propagation path throughout the lamination process and avoiding stress superposition and interlayer slippage caused by improper pressing sequence.

[0063] Activation sequence Finally, the data is passed to the pressing module, which sequentially drives each pressing contact point to complete the contact and pressure application actions according to a time sequence. A time interval is set between adjacent activation steps. The interval is set based on the relaxation time required for the stress field to reach quasi-static equilibrium again after two adjacent contact points are activated, which can be estimated using the viscoelastic parameters of the material. In practical engineering applications, The typical range of values ​​is closely related to the interlayer coupling stiffness and pressing temperature of the base plate material. For multi-layer base plate structures with large differences in thermal expansion coefficients, it is appropriate to extend the range. This helps reduce the accumulation of residual stress. The entire activation timing-driven process is related to the compensation displacement. Compensation angle The pose adjustment is executed in a coordinated manner to ensure that the lamination alignment reaches the optimal state simultaneously in both spatial position and stress release dimensions.

[0064] The control pressing module adjusts its position and attitude according to the compensation displacement and compensation angle, and drives the pressing contact points to complete the lamination alignment according to the activation sequence, including: The translational motion trajectory of the pressing module is established based on the compensation displacement, and the rotational motion trajectory of the pressing module is established based on the compensation angle. The translational motion trajectory and the rotational motion trajectory are combined to obtain the six-degree-of-freedom motion trajectory of the pressing module, and the pressing module is controlled to adjust its position and attitude according to the six-degree-of-freedom motion trajectory. Based on the activation timing, the spatial location and activation order of the pressing contact points are extracted. The pressing contact points that are ranked earlier in the activation timing are marked as priority pressing points, and the pressing contact points that are ranked later in the activation timing are marked as delayed pressing points. The control module moves to the spatial position corresponding to the priority pressing point, drives the pressing contact point to apply pressing force at the priority pressing point and maintain the pressing state, and establishes the pressing constraint of the priority pressing point; Under the pressure constraint of the priority pressing point, the control pressing module moves sequentially to the spatial position corresponding to the delayed pressing point, and drives the pressing contact point to apply pressing force sequentially at the delayed pressing point until all pressing contact points in the activation sequence are pressed, thereby achieving lamination alignment.

[0065] When adjusting the position and attitude of the pressing module based on compensated displacement and compensated angle, the two types of motion commands need to be constructed as independent motion trajectories before being synthesized. The compensated displacement includes translational components in the horizontal and vertical directions. A translational motion trajectory is constructed in a Cartesian coordinate system based on these two components. The trajectory starts at the current spatial position of the pressing module and ends at the target alignment position. A smooth translational path is generated using linear interpolation or spline interpolation to ensure no abrupt acceleration occurs during the motion. The compensated angle corresponds to a rotational component about the vertical axis. A rotational motion trajectory is constructed in rotational space based on this angle. The rotational trajectory starts at the current attitude quaternion and ends at the target attitude quaternion. A continuous rotational path is generated using spherical linear interpolation to avoid lamination position jitter caused by direct angle jumps. The translational and rotational motion trajectories are synthesized into a six-degree-of-freedom motion trajectory. At each time step, the six-degree-of-freedom motion trajectory simultaneously contains the position coordinates of three translational degrees of freedom and the attitude parameters of three rotational degrees of freedom. When the pressing module moves along this trajectory, translation and rotation are executed synchronously, resulting in a continuous and smooth overall attitude adjustment process. Control commands are sent to the drive unit of the pressing module in a time-step sequence. The drive unit provides real-time feedback on the current position error based on the target pose at each time step and performs closed-loop correction until the pressing module reaches the target position and completes the initial alignment.

[0066] After the position and attitude adjustment is completed, the spatial coordinates of each pressing contact point and its corresponding activation sequence number are extracted from the activation sequence. The activation sequence is arranged in the direction of decreasing potential energy. The pressing contact points that are ranked earlier are located in areas with higher stress potential energy, corresponding to locations with concentrated interlayer shear stress and greater risk of deformation in the base plate; the pressing contact points that are ranked later are located in areas with lower potential energy, corresponding to locations where the interlayer stress in the base plate is relatively dispersed. The pressing contact points ranked earlier in the activation sequence are marked as priority pressing points. The role of priority pressing points is to apply pressing constraints first in high-stress areas to suppress relative slippage between layers. The pressing contact points ranked later in the activation sequence are marked as delayed pressing points. Delayed pressing points are activated sequentially after the priority pressing points have established stable constraints, so that the order of applying pressing force matches the distribution law of interlayer stress, reducing the risk of local warping or misalignment caused by improper pressing sequence.

[0067] The control pressing module first moves to the spatial position corresponding to the priority pressing point. During the movement, the pressing module moves along a pre-planned path. After reaching the pre-pressing position above the priority pressing point, it drives the pressing contact point to contact the base plate surface at a controlled rate. The contact force is collected in real time by a force sensor. When the contact force reaches the preset pressing force threshold, the pressing contact point stops pressing and maintains the current pressing state, establishing the pressing constraint of the priority pressing point. The establishment of the pressing constraint means that the relative displacement between the cover layer and the base plate at this position is fixed, and subsequent pressing operations at other positions will not cause interlayer slippage in this area. During the pressing holding phase, the force sensor continuously monitors the contact force value. If the contact force is lower than the threshold, the pressing amount is compensated; if the contact force exceeds the upper limit, the pressing force is reduced. A stable pressing constraint state is maintained through closed-loop control.

[0068] During the holding period of the pressing constraint at the priority pressing point, the pressing module is controlled to move sequentially to the spatial position corresponding to each delayed pressing point. The activation of the delayed pressing points is executed one by one in the order of the activation sequence. The activation process of each delayed pressing point is the same as that of the priority pressing point. After the pressing module reaches the target position, it drives the pressing contact point to contact the surface of the base plate. After the contact force reaches the preset threshold, the pressing state is maintained. Since the pressing constraint has been established at the priority pressing point, the position of the base plate in the high-stress area is fixed. When the pressing force is gradually applied to the low-stress area at the delayed pressing point, the interlayer stress is gradually released outward along the stress transmission path, and will not accumulate in the unpressed area, causing local deformation. After each delayed pressing point completes pressing, the pressing constraint of that point is superimposed on the existing constraint system. The overall constraint coverage gradually expands as the activation sequence progresses until all pressing contact points in the activation sequence have completed pressing.

[0069] After all pressing contact points are pressed, the lamination alignment between the cover layer and the base plate is completed. The pressing force at each pressing contact point remains stable during the holding phase, and the interlayer interface is solidified and bonded under full-area constraints. After pressing, the pressing module lifts each pressing contact point sequentially in the opposite direction to the pressing sequence, starting with the delayed pressing point and ending with the priority pressing point to avoid interlayer rebound caused by sudden release of constraints. The entire lamination alignment process achieves precise pose adjustment through a six-degree-of-freedom motion trajectory. By activating timing control to control the pressing sequence, pose compensation and pressing constraints are organically combined, enabling the cover layer to achieve stable lamination bonding with the base plate at the alignment position, meeting the lamination alignment accuracy requirements of multi-layer structures in high-computation base plates.

[0070] In practical implementation, the time step of the six-degree-of-freedom motion trajectory can be adjusted according to the motion speed and positioning accuracy requirements of the lamination module. A smaller time step results in higher discretization accuracy of the motion trajectory, but also places higher demands on the real-time control capabilities of the drive unit. The number and location of priority lamination points are determined by the activation sequence. Typically, one or more of the earliest lamination contact points in the activation sequence are selected as priority lamination points, with the remainder designated as delayed lamination points. The lamination force threshold is pre-calibrated based on the substrate material properties and the cover layer thickness to ensure sufficient constraint force without damaging the substrate surface structure. The activation time interval between delayed lamination points is set with reference to the time interval between adjacent activation steps, allowing sufficient time for interlayer stress to redistribute between each lamination operation, ensuring the stability of the overall lamination process.

[0071] A second aspect of the present invention provides a high-performance substrate lamination intelligent alignment system, comprising: The data acquisition unit is used to acquire images of the base plate surface, material property parameters of each layer of the base plate, and design coordinate data of the target overlay layer; The stress calculation unit is used to calculate the interlayer shear stress distribution based on the difference in thermal expansion coefficients of each layer in the material property parameters and convert it into a stress potential energy field. It identifies the local extreme points of potential energy in the stress potential energy field and constructs a stress influence domain with the local extreme points of potential energy as the center. It calculates the boundary diffusion rate of the stress influence domain as the deformation sensitivity. The feature extraction unit is used to divide the base plate surface image into multiple observation windows, extract the spatial topology of the positioning marks in each observation window, encode the spatial topology into feature vectors, identify the deformation stable region based on the spatial distribution of deformation sensitivity, and select the feature vectors corresponding to the observation windows in the deformation stable region to form a stable feature set. The offset calculation unit is used to match the stable feature set with the topology in the design coordinate data to obtain the initial pose offset, and calculate the compensation displacement and compensation angle based on the initial pose offset. The path planning unit is used to identify the continuous region with the largest potential energy gradient from the stress potential energy field to form a stress transmission path, set the pressing contact point along the stress transmission path, and generate the activation sequence of the pressing contact point according to the direction of decreasing potential energy. The pressing execution unit is used to control the pressing module to adjust its position and posture according to the compensation displacement and compensation angle, and to drive the pressing contact points to complete the lamination alignment according to the activation sequence.

[0072] 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.

[0073] 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.

[0074] 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.

[0075] 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 high-performance base plate lamination intelligent alignment method, characterized in that, include: Acquire images of the base plate surface, material property parameters of each layer of the base plate, and design coordinate data of the target overlay layer; The interlayer shear stress distribution is calculated based on the difference in thermal expansion coefficients of each layer in the material property parameters and converted into a stress potential energy field. Local extreme points of potential energy in the stress potential energy field are identified and stress influence domains are constructed with the local extreme points of potential energy as the center. The boundary diffusion rate of the stress influence domain is calculated as the deformation sensitivity. This includes: calculating the shear strain distribution at the interface of each layer based on the difference in thermal expansion coefficients and temperature change between adjacent layers in the material property parameters, and calculating the interlayer shear stress distribution based on the shear strain distribution and the shear modulus of each layer. The interlaminar shear stress distribution is used as a load source to construct a three-dimensional stress potential energy function, and the stress potential energy field is obtained by spatial integration of the three-dimensional stress potential energy function. The base plate surface image is divided into multiple observation windows. The spatial topology of the positioning marks in each observation window is extracted and encoded into feature vectors. Based on the spatial distribution of deformation sensitivity, the deformation stability region is identified, and the feature vectors corresponding to the observation windows in the deformation stability region are selected to form a stable feature set. The initial pose offset is obtained by matching the stable feature set with the topology in the design coordinate data, and the compensation displacement and compensation angle are calculated based on the initial pose offset. The stress transmission path is constructed by identifying the continuous region with the largest potential energy gradient in the stress potential energy field, setting up pressing contact points along the stress transmission path, and generating the activation sequence of the pressing contact points according to the direction of decreasing potential energy. This includes: calculating the potential energy gradient vector in the stress potential energy field, extracting the spatial position with the largest potential energy gradient vector magnitude as the stress transmission source point, tracing the spatial positions with potential energy gradient vector magnitude continuously greater than a preset gradient threshold from the stress transmission source point along the direction of the potential energy gradient vector, and connecting them to form the stress transmission path. Set up pressing contact points along the stress transmission path, and extract the spatial position and corresponding potential energy value of the pressing contact points in the stress potential energy field; The initial activation sequence is obtained by sorting the pressing contact points according to the decreasing potential energy based on the potential energy value. The spatial position of the first pressing contact point in the initial activation sequence is extracted and marked as the pressing constraint region. Pressing boundary conditions are applied to the pressing constraint region and the stress potential energy field is recalculated to obtain the updated stress potential energy field. In the updated stress potential energy field, the potential energy values ​​of the remaining pressing contact points are extracted again and the initial activation sequence is updated. This process is repeated iteratively until all pressing contact points are sorted to obtain the activation sequence. The control pressing module adjusts its position and posture according to the compensation displacement and compensation angle, and drives the pressing contact points to complete the lamination alignment according to the activation sequence.

2. The method according to claim 1, characterized in that, Based on the differences in the thermal expansion coefficients of each layer in the material properties parameters, the interlayer shear stress distribution is calculated and converted into a stress potential energy field. Local extrema of the potential energy in the stress potential energy field are identified, and a stress influence domain is constructed with the local extrema of the potential energy as the center. The boundary diffusion rate of the stress influence domain is calculated as the deformation sensitivity, including: In the stress potential energy field, calculate the spatial gradient distribution of potential energy values, and identify the spatial locations where the potential energy gradient is zero and the potential energy values ​​satisfy the local maximum condition as the local extreme points of potential energy. Starting from the local extreme point of potential energy, the decay curve of potential energy value is traced along multiple radial directions. Based on the characteristics of potential energy value change in the decay curve, the stress influence boundary points in each radial direction are determined. Multiple stress influence boundary points are connected to form a closed boundary as the boundary of the stress influence domain. Calculate the radial projection components of the potential energy gradient vector at each point on the boundary of the stress influence domain, sum the radial projection components to obtain the boundary diffusion rate vector, and calculate the magnitude of the boundary diffusion rate vector as the boundary diffusion rate. The boundary diffusion rate is mapped to the corresponding position on the base plate surface using the spatial coordinates of the stress influence domain, and this is used as the deformation sensitivity.

3. The method according to claim 2, characterized in that, Using the interlaminar shear stress distribution as a load source, a three-dimensional stress potential energy function is constructed. Spatial integration of this function yields the stress potential energy field, which includes: The interlayer shear stress distributed at each spatial location on the base plate surface is decomposed into two mutually orthogonal shear stress components, and a shear stress vector field is constructed based on the two shear stress components at each spatial location. Extract the elastic modulus and Poisson's ratio of each layer from the material property parameters, and calculate the flexibility tensor of each layer based on the elastic modulus and Poisson's ratio; Calculate the stress jump between adjacent layers in the thickness direction of the base plate shear stress vector field, identify the interface coupling position of stress transmission based on the stress jump, and introduce an interface strain energy correction term at the interface coupling position. The shear stress vector field is accumulated layer by layer along the thickness direction of the base plate to obtain the cumulative shear stress vector. The cumulative shear stress vector is combined with the interface strain energy correction term to form a load source. A three-dimensional stress potential energy function is constructed based on the load source and the compliance tensor of each layer. The three-dimensional stress potential energy function is spatially integrated along two orthogonal directions on the surface of the base plate and along the thickness direction of the base plate. The potential energy contribution at each spatial location is accumulated to obtain the stress potential energy field.

4. The method according to claim 1, characterized in that, The base plate surface image is divided into multiple observation windows. The spatial topology of the positioning markers within each observation window is extracted, and the spatial topology is encoded into feature vectors. Based on the spatial distribution of deformation sensitivity, deformation-stable regions are identified. The feature vectors corresponding to the observation windows within the deformation-stable regions are selected to form a stable feature set, including: Multiple observation windows are obtained by dividing the image of the base plate surface into a grid, and the center coordinates of the observation windows are recorded. Detect the positioning markers within the observation window and extract their position coordinates. Calculate spatial relationship parameters and construct a spatial topology based on the position coordinates. Extract topological feature parameters from the spatial topology and encode them into feature vectors. Calculate the spatial distance between observation windows based on the center coordinates of the observation windows, identify adjacent observation windows based on the spatial distance, and calculate the gradient value of the feature vectors by calculating the difference in feature vectors between the observation window and its adjacent observation windows. Read the spatial distribution data of deformation sensitivity and extract the corresponding deformation sensitivity value based on the center coordinates of the observation window; Calculate the correlation coefficient between the gradient value of the feature vector and the value of the deformation sensitivity, identify the observation window with a negative correlation coefficient, determine the spatial region where the identified observation window is located as the deformation stable region, and extract the feature vectors corresponding to the observation window in the deformation stable region to form a stable feature set.

5. The method according to claim 1, characterized in that, The initial pose offset is obtained by matching the stable feature set with the topology in the design coordinate data. The compensation displacement and compensation angle are then calculated based on the initial pose offset, including: Extract the design location coordinates of the positioning markers from the design coordinate data, and extract the center coordinates and deformation sensitivity values ​​of the observation window corresponding to each feature vector in the stable feature set. Calculate the topological similarity between the center coordinates of the observation window and the coordinates of the design location. Then, weight the topological similarity according to the deformation sensitivity value to obtain the deformation-corrected similarity. Finally, identify the matching coordinate correspondence based on the deformation-corrected similarity. Extract the center coordinates of the observation window and the design position coordinates based on the matched coordinate correspondence. Calculate the positional difference between the center coordinates of the observation window and the design position coordinates to obtain the position offset vector. Calculate the offset confidence level based on the deformation sensitivity value of the observation window corresponding to the position offset vector. The position offset vector is weighted and averaged according to the offset confidence to obtain the weighted average offset vector. The weighted average offset vector is then decomposed into principal component to obtain translation and rotation components. The translation and rotation components are then determined as the initial pose offset. The position compensation amount of the base plate is calculated as the compensation displacement based on the translation component in the initial pose offset, and the angle compensation amount of the base plate is calculated as the compensation angle based on the rotation component in the initial pose offset.

6. The method according to claim 1, characterized in that, The control pressing module adjusts its position and attitude according to the compensation displacement and compensation angle, and drives the pressing contact points to complete the lamination alignment according to the activation sequence, including: The translational motion trajectory of the pressing module is established based on the compensation displacement, and the rotational motion trajectory of the pressing module is established based on the compensation angle. The translational motion trajectory and the rotational motion trajectory are combined to obtain the six-degree-of-freedom motion trajectory of the pressing module, and the pressing module is controlled to adjust its position and attitude according to the six-degree-of-freedom motion trajectory. Based on the activation timing, the spatial location and activation order of the pressing contact points are extracted. The pressing contact points that are ranked earlier in the activation timing are marked as priority pressing points, and the pressing contact points that are ranked later in the activation timing are marked as delayed pressing points. The control module moves to the spatial position corresponding to the priority pressing point, drives the pressing contact point to apply pressing force at the priority pressing point and maintain the pressing state, and establishes the pressing constraint of the priority pressing point; Under the pressure constraint of the priority pressing point, the control pressing module moves sequentially to the spatial position corresponding to the delayed pressing point, and drives the pressing contact point to apply pressing force sequentially at the delayed pressing point until all pressing contact points in the activation sequence are pressed, thereby achieving lamination alignment.

7. A high-performance substrate lamination intelligent alignment system, 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 images of the base plate surface, material property parameters of each layer of the base plate, and design coordinate data of the target overlay layer; The stress calculation unit is used to calculate the interlayer shear stress distribution based on the difference in thermal expansion coefficients of each layer in the material property parameters and convert it into a stress potential energy field. It identifies the local extreme points of potential energy in the stress potential energy field and constructs a stress influence domain with the local extreme points of potential energy as the center. It calculates the boundary diffusion rate of the stress influence domain as the deformation sensitivity. The feature extraction unit is used to divide the base plate surface image into multiple observation windows, extract the spatial topology of the positioning marks in each observation window, encode the spatial topology into feature vectors, identify the deformation stable region based on the spatial distribution of deformation sensitivity, and select the feature vectors corresponding to the observation windows in the deformation stable region to form a stable feature set. The offset calculation unit is used to match the stable feature set with the topology in the design coordinate data to obtain the initial pose offset, and calculate the compensation displacement and compensation angle based on the initial pose offset. The path planning unit is used to identify the continuous region with the largest potential energy gradient from the stress potential energy field to form a stress transmission path, set the pressing contact point along the stress transmission path, and generate the activation sequence of the pressing contact point according to the direction of decreasing potential energy. The pressing execution unit is used to control the pressing module to adjust its position and posture according to the compensation displacement and compensation angle, and to drive the pressing contact points to complete the lamination alignment according to the activation sequence.

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