Micro-nano array deformation self-adaptive adjustment method for femtosecond laser processing

Abstract

The invention belongs to the technical field of femtosecond laser processing, and relates to a femtosecond laser processing micro-nano array deformation self-adaptive adjustment method, which comprises the following steps of: 1) acquiring a theoretical three-dimensional model and an entity scanning model of a workpiece to be processed; the method comprises the following steps of (1) obtaining a theoretical three-dimensional model, (2) aligning the theoretical three-dimensional model obtained in the step (1) with an entity scanning model, and (3) reconstructing an array three-dimensional model to complete micro-nano array deformation self-adaptive adjusting.The invention provides a high-precision micro-nano array deformation self-adaptive adjustment method for femtosecond laser processing.

Description

Technical Field

[0001] This invention belongs to the field of femtosecond laser processing technology, and relates to a method for adaptive adjustment of micro-nano array deformation, and more particularly to a method for adaptive adjustment of micro-nano array deformation in femtosecond laser processing. Background Technology

[0002] In the field of femtosecond laser processing, complex functional surfaces composed of a large number of micro- and nano-units have become key carriers for advanced optical devices and electronic systems. The final performance of such surfaces highly depends on the accurate reproduction of the theoretically designed configuration by thousands of micro- and nano-arrays. However, due to the coupled influence of multiple factors such as differences in material physical properties, process fluctuations, and assembly stress, a significant manufacturing deformation inevitably occurs between the actual geometry of the finished product and the original digital model. This deviation between digital and physical forms leads to a core contradiction: even with the extremely high processing precision of femtosecond laser processing, the actual arrangement of these elements on the deformed substrate deviates from the theoretically optimal layout, thus severely limiting the final functional realization and performance ceiling of the device.

[0003] Currently, the common practice in the industry is to rely on engineers to adjust the overall arrangement of micro-nano arrays for each part and to manually adjust problematic arrays. This method is inefficient, inaccurate, and inconsistent, and it is difficult to find the optimal solution under multiple constraints such as "constant quantity, no deformation, equal spacing, and high precision," thus failing to meet the needs of mass production. Therefore, developing an automatic and high-precision adaptive adjustment method for the deformation deviation of micro-nano arrays in femtosecond laser processing to replace inefficient and unstable manual operation has become an urgent technical requirement to drive the digitalization and automation of this type of femtosecond laser processing. Summary of the Invention

[0004] To address the aforementioned technical problems in the background art, this invention provides a high-precision femtosecond laser processing method for adaptive adjustment of micro / nano array deformation.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: A method for adaptive adjustment of the deformation of a micro / nano array processed by femtosecond laser, characterized in that the method includes the following steps: 1) Obtain the theoretical 3D model and solid scan model of the workpiece to be processed; 2) Align the theoretical 3D model and the solid scan model obtained in step 1); 3) Reconstruct the three-dimensional model of the array to complete the adaptive adjustment of the micro-nano array deformation.

[0006] The theoretical 3D model mentioned in step 1) above is a model obtained by theoretical modeling based on the 3D drawing of the workpiece to be processed; the solid scanning model is a scanning model of the actual femtosecond laser processing of the workpiece to be processed.

[0007] The specific implementation method of step 2) above is as follows: 2.1) Align the theoretical 3D model and the solid scan model obtained in step 1) with a reference to obtain the output point set of the reference alignment; 2.2) Based on the completion of step 2.1), iterative optimization is performed using the nearest neighbor method to obtain the aligned transformation matrix, thus completing the internal point alignment of the theoretical 3D model and the solid scan model; 2.3) The alignment matrix obtained in step 2.2) is used to unify the entity scan model to the coordinate system of the theoretical 3D model, thus completing the alignment of the theoretical 3D model and the entity scan model.

[0008] The specific implementation method of step 2.1) above is: to perform benchmark alignment on the theoretical 3D model and the solid scan model obtained in step 1) using an automatic alignment method based on feature points or an interactive automatic alignment method, and obtain the benchmark aligned output point set.

[0009] The specific implementation method of the above-mentioned automatic alignment based on feature points is as follows: 2.a.1) Randomly select 3 pairs of non-collinear reference feature points from the dataset of the theoretical 3D model; 2.a.2) Calculate a candidate rigid body transformation matrix (R, t) based on the three pairs of reference feature points obtained in step 2.a.1); where R is the rigid body rotation matrix and t is the rigid body translation vector; 2.a.3) Apply the candidate rigid body transformation matrix to all feature points of the solid scan model and calculate the Euclidean distance between the rigid body transformation matrix and the feature points of the solid scan model; feature points whose Euclidean distance is less than the threshold are considered interior points of the transformation. 2.a.4) Repeat steps 2.a.1) to 2.a.3) until a rigid body transformation matrix with the most interior points is found. Based on the rigid body transformation matrix with the most interior points, the theoretical 3D model and the solid scan model are aligned to obtain the output point set of the alignment. The specific implementation method of the interactive automatic alignment is as follows: 2.b.1) Manually select at least three pairs of non-collinear corresponding points on both the theoretical 3D model and the solid scan model; 2.b.2) Calculate the rigid body transformation matrix of the corresponding points obtained in step 2.b.1) using the least squares method, and find the optimal rigid body rotation matrix R and rigid body translation vector t that minimize the average distance between corresponding points of the theoretical 3D model and the solid scan model; the mathematical expression for finding this is: in: p i It is the point set data points of the entity scanning model; q i It is the point set data points of the theoretical three-dimensional model; n is the total number of points in the corresponding point set. i is a point that corresponds to the current theoretical 3D model point set and the solid scan model point set; R is the rigid body rotation matrix; t is the rigid body translation vector; The rigid body rotation matrix R is calculated as follows: R=VU T in: p i 'and q i These are, respectively, decentralized data points of the entity scanning model and decentralized data points of the theoretical 3D model; V and U are 3×3 orthogonal matrices; T is the matrix transpose; The rigid body translation vector t is calculated as follows: t=μ q -R·μ p in: μ q With μ p They are p i With q i The point set of particles; 2.b.3) Based on the optimal rigid body rotation matrix R and rigid body translation vector t obtained in step 2.b.2), the theoretical 3D model and the solid scanning model are aligned to obtain the output point set of the alignment.

[0010] The specific implementation method of step 2.2) above is as follows: 2.2.1) Perform nearest point index matching between the benchmark-aligned output point set obtained in step 2.1) and the entity scan model point set. For all points in the benchmark-aligned output point set, search for the points in the entity scan model point set with the nearest Euclidean distance to establish a correspondence. Remove points in the benchmark-aligned output point set that do not have a corresponding relationship in the entity scan model point set. Finally, obtain the valid point pairs (p). i q i ), where p i It is the point set data points of the entity scanning model; q i It is the point set data points of the theoretical three-dimensional model; 2.2.2) Based on the valid point pairs (p) obtained in step 2.2.1), i q i The alignment transformation matrix is ​​calculated to achieve internal point alignment between the theoretical 3D model and the solid scan model.

[0011] The specific implementation method of step 2.2.2) above is as follows: 2.2.2.1) For the valid point pairs (p) obtained in step 2.2.1), i q i Perform centroid difference biasing to obtain the point set (p) i ', q i '), the p i 'and q i The calculation method for ' is: p i '= p i -μ p q i '= q i -μ q in: μ p With μ q These are the centroids of the output point set aligned to the baseline and the centroids of the point set of the entity scan model, respectively. 2.2.2.2) Based on the point set (p) obtained in step 2.2.2.1) i ', q i ') Calculate the covariance matrix H, the expression for which is: H=∑p i '×(q i ') T in: T is the matrix transpose; 2.2.2.3) Perform singular value decomposition on the covariance matrix H obtained in step 2.2.2.2) to obtain the optimal rotation matrix R. p With the optimal translation vector t p : H=U∑V T R p =VU T t p =μ q -R·μ p in: R p It is the optimal rotation matrix; t p It is the optimal translation vector; 2.2.2.4) Based on the optimal rotation matrix R p With the optimal translation vector t p The aligned transformation matrix is ​​obtained, enabling the alignment of interior points between the theoretical 3D model and the solid scan model.

[0012] The specific implementation method of step 3) above is as follows: 3.1) Separate each element from the theoretical 3D model and calculate the theoretical center point of each element using the graphical contour coordinates to obtain the set of theoretical center points of the elements; 3.2) Based on the array alignment algorithm, the theoretical array center point set is driven as a whole to a new position that matches the surface of the solid scan model to obtain the aligned new array center points; 3.3) Project the newly aligned center points of the new array onto the surface of the scanned model vertically; 3.4) Adjust the coordinates of each point of the original theoretical array of the theoretical 3D model according to the transformation relationship of its center point, and attach it to the surface of the solid scanning model to generate the final adjusted array 3D model, thus completing the adaptive adjustment of micro-nano array deformation.

[0013] The specific implementation method of step 3.2) above is as follows: 3.2.1) Obtain the theoretical center point set Y0 for internal point alignment and the actual workpiece surface point set X0 for actual femtosecond laser scanning, and preprocess X0, including downsampling, noise reduction and key point extraction; 3.2.2) Initialize the velocity weight matrix W of the discrete parameterized velocity field v, and initialize the noise variance as δ. 2 ; 3.2.3) Calculate the relationship γ(p, q) between the theoretical point set and the actual point set based on the simple relationship of the nearest neighbor, and update γ(p, q) in the following way: γ(p, q) ∝ π q ·e-‖y p -x q || 2 in: π q The weighting parameter is based on a uniform distribution: π q =1 / Q; Q is the number of points in the theoretical concentration of the phasor; y p and x q These are the coordinates of point p and point q, respectively. 3.2.4) Update the velocity field v discrete parameterized velocity weight matrix W and noise variance δ 2 The update method is as follows: (G+(δ)2 / λ)I)W=P·XG·Y in: G is an N×N Gaussian kernel matrix; λ is the regularization parameter; P is an N×M weight matrix; X is an M×3 scanning point coordinate matrix; X is the result of matrixing the actual workpiece surface point set X0 from the actual femtosecond laser scanning. Y is an N×3 theoretical point coordinate matrix; I is an N×M identity matrix; 3.2.5) Repeat steps 3.2.3) and 3.2.4) until the noise variance δ 2 Once the set threshold or the maximum number of iterations is reached, the center point of the newly aligned subarray is finally obtained.

[0014] The above-mentioned femtosecond laser processing micro / nano array deformation adaptive adjustment method further includes, after step 3): 4) Evaluate the results of the adaptive adjustment of the micro-nano array deformation. The evaluation includes the statistical changes in the number of arrays, the changes in the total side length of each array, and the changes in the minimum spacing between arrays.

[0015] The advantages of this invention are: This invention provides a method for adaptive adjustment of micro / nano array deformation in femtosecond laser processing, comprising: 1) acquiring a theoretical 3D model and a physical scanning model of the workpiece; 2) aligning the theoretical 3D model and the physical scanning model obtained in step 1); and 3) reconstructing the array 3D model to complete the adaptive adjustment of micro / nano array deformation. This invention, through a two-stage adjustment strategy of model adjustment and array adjustment, can adaptively handle complex workpiece deformation with micro / nano-level precision. Simultaneously, by defining the priority of the adjustment strategy, this invention cleverly resolves the conflict between multiple constraints, making the adjustment results more consistent with actual business needs. Furthermore, this invention achieves full automation from model adjustment to array adjustment, replacing inefficient manual adjustment and greatly improving production efficiency, especially suitable for batch production. This invention, by selecting undeformed feature areas during array alignment and leveraging its inherent robustness to noise and missing data, jointly ensures the stability and reliability of the method in real industrial scenarios. Obviously, this invention overcomes the shortcomings of existing femtosecond laser processing micro-nano array adjustment technology, and provides an automated, high-precision micro-nano array adaptive adjustment method and system. It can automatically adjust the position and shape of the theoretical micro-nano array model according to the actual workpiece model obtained by scanning. Under the premise of ensuring that the number of arrays remains unchanged, it meets the business requirements of unchanged number, no deformation, equal spacing, and high precision as much as possible, thereby significantly improving production efficiency and product quality. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the overall process of the micro / nano array deformation adaptive adjustment method for femtosecond laser processing provided by the present invention; Figure 2 This is a detailed flowchart of the model adjustment process used in this invention; Figure 3 This is a flowchart of the array adjustment process used in this invention; Figure 4 This is the adjustment and evaluation flowchart used in this invention; Figure 5 It is a tile set map after automatic alignment; Figure 6 This is a schematic diagram showing the deformed area marked by the automatically aligned tile point set; Figure 7 This is a cut-out diagram of the pattern (the pattern is green, and the red dots are deformation points); Figure 8 It is the point set graph before matching the template and deformed tile point sets; Figure 9 It is a point set graph after matching the point sets of templates and deformed tiles; Figure 10 This is a schematic diagram of the new array generated after automatic alignment and deformation. Detailed Implementation

[0017] See Figure 1 This invention provides a method for adaptive adjustment of the deformation of micro / nano arrays processed by femtosecond lasers, comprising the following steps: 1) Obtain the theoretical 3D model and the solid scanning model of the workpiece to be processed; the theoretical 3D model is a model obtained by theoretical modeling based on the 3D drawings of the workpiece to be processed; the solid scanning model is the scanning model of the workpiece to be processed by actual femtosecond laser processing.

[0018] 2) Align the theoretical 3D model and the solid scan model obtained in step 1), see [link to relevant documentation]. Figure 2 Specifically, this includes the following methods: 2.1) Align the theoretical 3D model and the entity scan model obtained in step 1) with a reference to obtain the output point set of the reference alignment; for example, the reference alignment can be performed by an automatic alignment method based on feature points or an interactive automatic alignment method to align the theoretical 3D model and the entity scan model obtained in step 1).

[0019] The specific implementation method for automatic alignment based on feature points is as follows: 2.a.1) Randomly select 3 pairs of non-collinear reference feature points from the dataset of the theoretical 3D model; 2.a.2) Calculate a candidate rigid body transformation matrix (R, t) based on the three pairs of reference feature points obtained in step 2.a.1); where R is the rigid body rotation matrix and t is the rigid body translation vector; 2.a.3) Apply the candidate rigid body transformation matrix to all feature points of the solid scan model and calculate the Euclidean distance between the rigid body transformation matrix and the feature points of the solid scan model; feature points whose Euclidean distance is less than the threshold are considered interior points of the transformation; feature points are geometric feature points, such as corner points, centroids, hariss points, etc., selected according to the actual situation.

[0020] 2.a.4) Repeat steps 2.a.1) to 2.a.3) until a rigid body transformation matrix with the most interior points is found. Based on the rigid body transformation matrix with the most interior points, the theoretical 3D model and the solid scan model are aligned to obtain the output point set of the alignment. The specific implementation method for interactive automatic alignment is as follows: 2.b.1) Manually select at least three pairs of non-collinear corresponding points on both the theoretical 3D model and the solid scan model; 2.b.2) The rigid body transformation matrix of the corresponding points obtained in step 2.b.1) is calculated using the least squares method. The core of this method is to solve an orthogonal Prouke problem to find the optimal rigid body rotation matrix R and rigid body translation vector t, which minimizes the average distance between corresponding points in the theoretical 3D model and the solid scan model. The resulting mathematical expression is: in: p i It is the point set data points of the entity scanning model; q i It is the point set data points of the theoretical three-dimensional model; n is the total number of points in the corresponding point set. i is a point that corresponds to the current theoretical 3D model point set and the solid scan model point set; R is the rigid body rotation matrix; t is the rigid body translation vector; The rigid body rotation matrix R is calculated as follows: R=VU T in: p i 'and q i These are, respectively, decentralized data points of the entity scanning model and decentralized data points of the theoretical 3D model; V and U are 3×3 orthogonal matrices; T is the matrix transpose; The rigid body translation vector t is calculated as follows: t=μ q -R·μ p in: μ q With μ p They are p i With q i The point set of particles; 2.b.3) Based on the optimal rigid body rotation matrix R and rigid body translation vector t obtained in step 2.b.2), the theoretical 3D model and the solid scanning model are aligned to obtain the output point set of the alignment.

[0021] 2.2) Based on step 2.1), iterative optimization is performed using the nearest neighbor method to obtain the aligned transformation matrix, thus completing the interior point alignment of the theoretical 3D model and the solid scan model. Specifically: 2.2.1) Perform nearest point index matching between the benchmark-aligned output point set obtained in step 2.1) and the entity scan model point set. For all points in the benchmark-aligned output point set, search for the points in the entity scan model point set with the nearest Euclidean distance to establish a correspondence. Remove points in the benchmark-aligned output point set that do not have a corresponding relationship in the entity scan model point set. Finally, obtain the valid point pairs (p). i q i ), where p i It is the point set data points of the entity scanning model; q i It is the point set data points of the theoretical three-dimensional model; 2.2.2) Based on the valid point pairs (p) obtained in step 2.2.1), i q i The alignment transformation matrix is ​​calculated to achieve interior point alignment between the theoretical 3D model and the solid scan model. Step 2.2.2) specifically involves: 2.2.2.1) For the valid point pairs (p) obtained in step 2.2.1), i q i Perform centroid difference biasing to obtain the point set (p) i ', q i '), p i 'and q i The calculation method for ' is: p i '= p i -μ p q i '= q i -μ q in: μ p With μ qThese are the centroids of the output point set aligned to the baseline and the centroids of the point set of the entity scan model, respectively. 2.2.2.2) Based on the point set (p) obtained in step 2.2.2.1) i ', q i ') Calculate the covariance matrix H. The expression for the covariance matrix H is: H=∑p i '×(q i ') T in: T is the matrix transpose; 2.2.2.3) Perform singular value decomposition on the covariance matrix H obtained in step 2.2.2.2) to obtain the optimal rotation matrix R. p With the optimal translation vector t p : H=U∑V T R p =VU T t p =μ q -R·μ p in: R p It is the optimal rotation matrix; t p It is the optimal translation vector; 2.2.2.4) Based on the optimal rotation matrix R p With the optimal translation vector t p The aligned transformation matrix is ​​obtained, enabling the alignment of interior points between the theoretical 3D model and the solid scan model.

[0022] To improve the accuracy and robustness of interior point alignment, a set of multiple orthogonal feature region points in the theoretical 3D model that are not deformed are selected as input (effective points for interior point alignment) to avoid interference from deformed parts on the alignment.

[0023] 2.3) The alignment matrix obtained in step 2.2) is used to unify the entity scan model to the coordinate system of the theoretical 3D model, thus completing the alignment of the theoretical 3D model and the entity scan model.

[0024] 3) Reconstruct the 3D model of the array, complete the adaptive adjustment of the micro / nano array deformation, and clarify the core adjustment strategy of prioritizing maintaining the constant number of arrays, which can be extended to prioritize non-deformation or prioritize equal spacing. See also Figure 3 Specifically:

[0025] 3.1) Separate each element from the theoretical 3D model and calculate the theoretical center point of each element using the graphical contour coordinates to obtain the set of theoretical center points of the elements; 3.2) Based on the qubit alignment algorithm, the theoretical qubit center point set is driven as a whole to a new matching position on the surface of the solid scanned model to obtain the aligned new qubit center points. Specifically: 3.2.1) Obtain the theoretical center point set Y0 for internal point alignment and the actual workpiece surface point set X0 for actual femtosecond laser scanning, and preprocess X0, including downsampling, noise reduction and key point extraction; 3.2.2) Initialize the velocity weight matrix W of the discrete parameterized velocity field v, assuming that the deformation is very small, and the initial noise variance is δ. 2 ; 3.2.3) Calculate the relationship γ(p, q) between the theoretical point set and the actual point set based on the simple relationship of the nearest neighbor. Update γ(p, q) to remove scanning noise, missing point sets, and surface roughness. The update method is as follows: γ(p, q) ∝ π q ·e-‖y p -x q || 2 in: π q The weighting parameter is based on a uniform distribution: π q =1 / Q; Q is the number of points in the theoretical concentration of the phasor; y p and x q These are the coordinates of point p and point q, respectively. 3.2.4) Update the velocity field v discrete parameterized velocity weight matrix W and noise variance δ 2 The update method is: (G+(δ) 2 / λ)I)W=P·XG·Y in: G is an N×N Gaussian kernel matrix; λ is the regularization parameter; P is an N×M weight matrix; X is an M×3 matrix of scan point coordinates; Y is an N×3 theoretical point coordinate matrix; I is an N×M identity matrix; 3.2.5) Repeat steps 3.2.3) and 3.2.4) until the noise variance δ 2 Once the set threshold or the maximum number of iterations is reached, the center point of the newly aligned subarray is finally obtained.

[0026] 3.3) Project the newly aligned center point of the element vertically onto the surface of the scanning model to ensure that the element is positioned on the actual workpiece; 3.4) Adjust the coordinates of each point of the original theoretical array of the theoretical 3D model according to the transformation relationship of its center point, and attach it to the surface of the solid scanning model to generate the final adjusted array 3D model, thus completing the adaptive adjustment of micro-nano array deformation.

[0027] 4) Evaluate the results of adaptive deformation adjustment of micro / nano arrays, see [reference needed]. Figure 4 The evaluation includes statistical analysis of changes in the number of segments, changes in the total side length of each segment, and changes in the minimum spacing between segments. Specifically, the evaluation involves establishing a one-to-one mapping relationship between segments before and after the adjustment. Through basic geometric calculations, the adjustment effect is quantitatively evaluated, including statistical analysis of changes in the number of segments, changes in the total side length of each segment, and changes in the minimum spacing between segments.

[0028] The purpose of this invention is to solve the problem of positional deviation and deformation of micro / nano arrays in femtosecond laser processing, and to provide a high-precision adaptive adjustment method based on micro / nano array deformation. The following detailed description uses array adjustment on curved tiles as an example to illustrate the invention.

[0029] 1. Parameter settings Workpiece type: Curved tile Array type: Regularly arranged planar micro / nano array units (characteristic size 10-50μm) Maximum number of iterations: 150 Convergence error: 1e-6 Deformation region error threshold: 10μm 2. Model Alignment For the actual workpiece model data obtained by femtosecond laser scanning, the theoretical model and the scanning model point set are aligned.

[0030] Algorithm flow: Find the nearest neighbor pair between two point sets.

[0031] Calculate the transformation matrix that minimizes the mean square error between corresponding points.

[0032] Apply the transformation matrix to the theoretical point set.

[0033] The output is a theoretical model that is macroscopically aligned with the scanning model after overall rotation and translation.

[0034] The results of the tile workpiece scanning model are attached. Figure 5 .

[0035] 3. Deformation area and element clipping Based on the common local thermal deformation characteristics in femtosecond laser processing, thermal deformation areas are identified and processed.

[0036] Error calculation: Calculate the nearest neighbor distance from each point in the theoretical point set after benchmark alignment to the scanned point set, and use it as the alignment error.

[0037] Region identification: Mark points with errors exceeding a set threshold as "deformation points".

[0038] Array element trimming: Based on the array element layout information, the array elements located in the deformation point set are trimmed from the theoretical model and used as input for subsequent array element alignment.

[0039] The deformed area is identified as shown in the attached figure. Figure 6 The cutting process is as follows (attached). Figure 7 .

[0040] 4. Periodic adjustments To address the unique non-uniform thermal deformation in femtosecond laser processing, an automatic alignment algorithm is employed to deform the array onto the surface of the scanning model.

[0041] The process is as follows: The system acquires the reference-aligned array center point set and the actual workpiece surface point set scanned by femtosecond laser, and performs preprocessing on the actual workpiece surface point set, including downsampling, noise reduction, and key point extraction.

[0042] Calculate the expected value of the correspondence and the posterior weight of the velocity field.

[0043] Update the posterior velocity field weights and noise variance transformation parameters.

[0044] Repeat steps (2) and (3) until the noise variance change is less than the threshold.

[0045] The effect of point set alignment before and after the algorithm is obtained. Figure 8 , Figure 9 The new array generated after alignment, deformation, and attachment is as follows: Figure 10 In the actual alignment process, in order to keep the array from deforming, the array is translated and rotated as a whole to ensure that the side length of the array does not change and to prevent the array from expanding or contracting.

[0046] 5. Adjusted Model Generation and Evaluation Model generation: The array model after array alignment is merged with the undeformed theoretical part to ensure the integrity of the femtosecond laser processing pattern.

[0047] Effect evaluation: Evaluate the dimensional changes caused by thermal deformation.

[0048] Minimum distance difference between phases: Verifying the resolution retention of femtosecond laser processing.

Claims

1. A method for adaptive adjustment of the deformation of micro / nano arrays processed by femtosecond laser, characterized in that: The micro / nano array deformation adaptive adjustment method for femtosecond laser processing includes the following steps: 1) Obtain the theoretical 3D model and solid scan model of the workpiece to be processed; 2) Align the theoretical 3D model and the solid scan model obtained in step 1); 3) Reconstruct the three-dimensional model of the array to complete the adaptive adjustment of the micro-nano array deformation.

2. The method for adaptive adjustment of micro / nano array deformation by femtosecond laser processing according to claim 1, characterized in that: The theoretical 3D model mentioned in step 1) is a model obtained by theoretical modeling based on the 3D drawing of the workpiece to be processed; the solid scanning model is a scanning model of the workpiece to be processed by actual femtosecond laser processing.

3. The method for adaptive adjustment of micro / nano array deformation by femtosecond laser processing according to claim 2, characterized in that: The specific implementation method of step 2) is as follows: 2.1) Align the theoretical 3D model and the solid scan model obtained in step 1) with a reference to obtain the output point set of the reference alignment; 2.2) Based on the completion of step 2.1), iterative optimization is performed using the nearest neighbor method to obtain the aligned transformation matrix, thus completing the internal point alignment of the theoretical 3D model and the solid scan model; 2.3) The alignment matrix obtained in step 2.2) is used to unify the entity scan model to the coordinate system of the theoretical 3D model, thus completing the alignment of the theoretical 3D model and the entity scan model.

4. The method for adaptive adjustment of micro / nano array deformation by femtosecond laser processing according to claim 3, characterized in that: The specific implementation method of step 2.1) is as follows: the theoretical 3D model and the entity scan model obtained in step 1) are aligned with the reference using an automatic alignment method based on feature points or an interactive automatic alignment method to obtain the output point set of the reference alignment.

5. The method for adaptive adjustment of micro / nano array deformation by femtosecond laser processing according to claim 4, characterized in that: The specific implementation method of the automatic alignment based on feature points is as follows: a.1) Randomly select 3 pairs of non-collinear reference feature points from the dataset of the theoretical 3D model; a.2) Calculate a candidate rigid body transformation matrix (R, t) based on the three pairs of reference feature points obtained in step 2.a.1); where R is the rigid body rotation matrix and t is the rigid body translation vector. a.3) Apply the candidate rigid body transformation matrix to all feature points of the solid scan model, and calculate the Euclidean distance between the rigid body transformation matrix and the feature points of the solid scan model; feature points whose Euclidean distance is less than the threshold are considered interior points of the transformation. a.4) Repeat steps 2.a.1) to 2.a.3) until a rigid body transformation matrix with the most interior points is found. Based on the rigid body transformation matrix with the most interior points, the theoretical 3D model and the solid scan model are aligned to obtain the output point set of the alignment. The specific implementation method of the interactive automatic alignment is as follows: b.1) Manually select at least three pairs of non-collinear corresponding points on both the theoretical 3D model and the solid scan model; b.2) Calculate the rigid body transformation matrix of the corresponding points obtained in step 2.b.1) using the least squares method, and find the optimal rigid body rotation matrix R and rigid body translation vector t that minimize the average distance between corresponding points of the theoretical 3D model and the solid scan model; the mathematical expression for finding this is: in: p i It is the point set data points of the entity scanning model; q i It is the point set data points of the theoretical three-dimensional model; n is the total number of points in the corresponding point set. i is a point that corresponds to the current theoretical 3D model point set and the solid scan model point set; R is the rigid body rotation matrix; t is the rigid body translation vector; The rigid body rotation matrix R is calculated as follows: R=VU T in: p i 'and q i These are, respectively, decentralized data points of the entity scanning model and decentralized data points of the theoretical 3D model; V and U are 3×3 orthogonal matrices; T is the matrix transpose; The rigid body translation vector t is calculated as follows: t=μ q -R·m p in: μ q With μ p They are p i With q i The point set of particles; b.3) Based on the optimal rigid body rotation matrix R and rigid body translation vector t obtained in step 2.b.2), the theoretical 3D model and the solid scanning model are aligned to obtain the output point set of the alignment.

6. The method for adaptive adjustment of micro / nano array deformation by femtosecond laser processing according to claim 5, characterized in that: The specific implementation method of step 2.2) is as follows: 2.2.1) Perform nearest point index matching between the benchmark-aligned output point set obtained in step 2.1) and the entity scan model point set. For all points in the benchmark-aligned output point set, search for the points in the entity scan model point set with the nearest Euclidean distance to establish a correspondence. Remove points in the benchmark-aligned output point set that do not have a corresponding relationship in the entity scan model point set. Finally, obtain the valid point pairs (p). i q i ), where p i It is the point set data points of the entity scanning model; q i It is the point set data points of the theoretical three-dimensional model; 2.2.2) Based on the valid point pairs (p) obtained in step 2.2.1), i q i The alignment transformation matrix is ​​calculated to achieve internal point alignment between the theoretical 3D model and the solid scan model.

7. The method for adaptive adjustment of micro / nano array deformation by femtosecond laser processing according to claim 6, characterized in that: The specific implementation method of step 2.2.2) is as follows: 2.2.2.1) For the valid point pairs (p) obtained in step 2.2.1), i q i Perform centroid difference biasing to obtain the point set (p) i ', q i '), the p i 'and q i The calculation method for ' is: p i '= p i -μ p q i ’= q i -μ q in: μ p With μ q These are the centroids of the output point set aligned to the baseline and the centroids of the point set of the entity scan model, respectively. 2.2.2.2) Based on the point set (p) obtained in step 2.2.2.1) i ', q i ') Calculate the covariance matrix H, the expression for which is: H=∑p i ’×(q i ’) T in: T is the matrix transpose; 2.2.2.3) Perform singular value decomposition on the covariance matrix H obtained in step 2.2.2.2) to obtain the optimal rotation matrix R. p With the optimal translation vector t p : H=U∑V T R p =VU T t p =m q -R·m p in: R p It is the optimal rotation matrix; t p It is the optimal translation vector; 2.2.2.4) Based on the optimal rotation matrix R p With the optimal translation vector t p The aligned transformation matrix is ​​obtained, enabling the alignment of interior points between the theoretical 3D model and the solid scan model.

8. The method for adaptive adjustment of micro / nano array deformation by femtosecond laser processing according to claim 7, characterized in that: The specific implementation method of step 3) is as follows: 3.1) Separate each element from the theoretical 3D model and calculate the theoretical center point of each element using the graphical contour coordinates to obtain the set of theoretical center points of the elements; 3.2) Based on the array alignment algorithm, the theoretical array center point set is driven as a whole to a new position that matches the surface of the solid scan model to obtain the aligned new array center points; 3.3) Project the newly aligned center points of the new array onto the surface of the scanned model vertically; 3.4) Adjust the coordinates of each point of the original theoretical array of the theoretical 3D model according to the transformation relationship of its center point, and attach it to the surface of the solid scanning model to generate the final adjusted array 3D model, thus completing the adaptive adjustment of micro-nano array deformation.

9. The method for adaptive adjustment of micro / nano array deformation by femtosecond laser processing according to claim 8, characterized in that: The specific implementation method of step 3.2) is as follows: 3.2.1) Obtain the theoretical center point set Y0 of the inner point alignment and the actual workpiece surface point set X0 of the actual femtosecond laser scanning, and preprocess X0, including downsampling, noise reduction and key point extraction; 3.2.2) Initialize the velocity weight matrix W of the discrete parameterized velocity field v, and initialize the noise variance as δ. 2 ; 3.2.3) Calculate the relationship γ(p, q) between the theoretical point set and the actual point set based on the simple relationship of the nearest neighbor, and update γ(p, q) in the following way: γ(p,q)∝π q ·e-‖y p -x q ‖ 2 in: π q The weighting parameter is based on a uniform distribution: π q =1 / Q; Q is the number of points of convergence in the phasor theory; y p and x q These are the coordinates of point p and point q, respectively. 3.2.4) Update the velocity field v discrete parameterized velocity weight matrix W and noise variance δ 2 The update method is as follows: (G+(d 2 / λ)I)W=P·XG·Y in: G is an N×N Gaussian kernel matrix; λ is the regularization parameter; P is an N×M weight matrix; X is an M×3 matrix of scan point coordinates; Y is an N×3 theoretical point coordinate matrix; I is an N×M identity matrix; 3.2.5) Repeat steps 3.2.3) and 3.2.4) until the noise variance δ 2 Once the set threshold or the maximum number of iterations is reached, the center point of the newly aligned subarray is finally obtained.

10. The method for adaptive adjustment of micro / nano array deformation by femtosecond laser processing according to any one of claims 1-9, characterized in that: The micro / nano array deformation adaptive adjustment method for femtosecond laser processing further includes, after step 3): 4) Evaluate the results of the adaptive adjustment of the micro-nano array deformation. The evaluation includes the statistical changes in the number of arrays, the changes in the total side length of each array, and the changes in the minimum spacing between arrays.

Images