Double-pulse high-frequency switching power supply control method and system for precise electroplating

Abstract

The invention provides a double-pulse high-frequency switching power supply control method and system for precise electroplating, and relates to the field of electroplating, and the method comprises the steps: collecting area surface data through a three-dimensional scanner, and obtaining a three-dimensional model; extracting a high-curvature area grid according to the three-dimensional model, and calculating to obtain a surface concave-convex degree index; performing current field simulation, and calculating current density vector distribution; when the deviation of the current density vector distribution exceeds a preset deviation threshold value, pulse parameters are optimized through a genetic algorithm; selecting a pulse parameter which is most matched with the surface concave-convex degree index, and carrying out iterative calculation to obtain updated current density vector distribution; based on the difference between the updated current density vector distribution and the initial distribution, predicting a coating thickness value, and determining a preliminary evaluation index; and optimizing pulse parameters to obtain a final control scheme of uniform current distribution. According to the method, pulse parameters can be dynamically optimized according to the geometrical shape of the workpiece, and uniform current distribution and stable plating quality are realized.

Description

Technical Field

[0001] This invention relates to the field of electroplating technology, and in particular to a dual-pulse high-frequency switching power supply control method and system for precision electroplating. Background Technology

[0002] In modern manufacturing, precision electroplating technology serves as a crucial means of enhancing workpiece surface performance, holding irreplaceable value in fields such as aerospace, electronic devices, and medical equipment. The electroplating process directly impacts the corrosion resistance, conductivity, and appearance quality of products; therefore, precisely controlling the electroplating effect has become a core issue for the industry's development.

[0003] In existing technologies, traditional control methods often fall short when dealing with complex workpieces, especially those with complex three-dimensional structures, struggling to adapt to process variations caused by different shapes and sizes. These methods typically cannot flexibly adjust control parameters according to real-time process requirements, leading to problems such as uneven coating thickness or unstable quality. Therefore, how to dynamically optimize pulse parameters based on the workpiece geometry during electroplating to improve the uniformity of current distribution and the stability of coating quality has become a key problem that this research urgently needs to solve.

[0004] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of the present invention, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] In view of this, the present invention provides a dual-pulse high-frequency switching power supply control method and system for precision electroplating, to solve the problem in the prior art that control parameters cannot be flexibly adjusted according to real-time process requirements, and to achieve dynamic optimization of pulse parameters according to the workpiece geometry, thereby improving the uniformity of current distribution and the stability of plating quality. An embodiment of the present invention provides a dual-pulse high-frequency switching power supply control method for precision electroplating, comprising: By acquiring surface data of complex workpieces using a 3D scanner, a 3D model containing local geometric changes is obtained. Based on the three-dimensional model, the high curvature region is extracted, and the high curvature region is meshed by finite element analysis to obtain the high curvature region mesh. Based on the high curvature region mesh, the surface roughness index is calculated using the local radius of curvature. The surface roughness index was simulated using the finite element method, and the current density vector distribution under the initial pulse parameters was calculated by applying a boundary voltage. When the deviation of the current density vector distribution exceeds a preset deviation threshold, the pulse parameters are optimized using a genetic algorithm to obtain the optimized pulse parameters. Select the pulse parameter that best matches the surface roughness index from the optimized pulse parameters. Based on the best-matching pulse parameter, use electromagnetic field simulation software to iteratively calculate the current density distribution to obtain the updated current density vector distribution. Based on the difference between the updated current density vector distribution and the initial distribution, a pre-established linear regression model is used to predict the coating thickness value. Based on the difference between the predicted thickness value and the preset thickness threshold, a preliminary evaluation index for quality stability is determined. Based on the preliminary evaluation indicators, the pulse parameters were optimized using the variability rate adjustment method to obtain the final control scheme with uniform current distribution.

[0006] In some optional embodiments, the step of extracting the high curvature region based on the three-dimensional model and then using finite element analysis software to refine the mesh of the high curvature region to obtain the high curvature region mesh includes: A preliminary set of feature points is obtained by performing a preliminary scan of the surface of the three-dimensional model; Based on the preliminary set of feature points, a curvature analysis algorithm based on point cloud data is used to calculate the local curvature value. The region whose local curvature value is higher than the preset curvature threshold is marked as a high curvature region, thus obtaining the high curvature region; The high curvature region was meshed using the finite element method to obtain the high curvature region mesh.

[0007] In some optional embodiments, the step of simulating the current field of the surface roughness index using finite element analysis, and calculating the current density vector distribution under the initial pulse parameters by applying a boundary voltage, includes: Based on the surface roughness index, a model is established using the finite element analysis method to obtain the surface roughness model; Based on the surface roughness model, the voltage conditions are set using the voltage application method to obtain the electric field distribution under the initial pulse value and determine the input parameters for the current field simulation. Based on the input parameters of the current field simulation, the current density vector distribution is obtained by calculation using the finite element analysis method.

[0008] In some optional embodiments, when the deviation of the current density vector distribution exceeds a preset deviation threshold, the pulse parameters are optimized using a genetic algorithm to obtain optimized pulse parameters, including: Based on the current density vector distribution, the distribution deviation of the current density vector is obtained by comparing local and global values; When the distribution deviation of the current density vector exceeds a preset deviation threshold, a genetic algorithm is used for optimization to obtain optimized pulse parameters.

[0009] In some optional embodiments, the step of selecting the pulse parameter that best matches the surface roughness index from the optimized pulse parameters, and iteratively calculating the current density distribution using electromagnetic field simulation software based on the best-matching pulse parameter to obtain the updated current density vector distribution includes: The matching degree between the optimized pulse parameters and the surface roughness index is calculated using a weighted scoring algorithm, and the pulse parameter with the highest matching degree is taken as the best matching pulse parameter. Based on the best-matching pulse parameters, the current density distribution is iteratively calculated using electromagnetic field simulation software to obtain a preliminary updated current density vector distribution. If the initially updated current density vector distribution deviates from the preset uniformity standard, the pulse parameters are re-optimized using a genetic algorithm, the matching degree is calculated, and the current density distribution is iteratively calculated using electromagnetic field simulation software to obtain the updated current density vector distribution.

[0010] In some optional embodiments, the coating thickness value is predicted using a pre-established linear regression model based on the difference between the updated current density vector distribution and the initial distribution. Based on the difference between the predicted thickness value and a preset thickness threshold, preliminary evaluation indicators for quality stability are determined, including: The initial current density vector distribution data is obtained through numerical simulation software. Based on the initial distribution, the difference matrix between the updated current density vector distribution and the initial distribution is calculated using the matrix difference algorithm. Based on the difference matrix, a pre-established regression model is used to predict the change in coating thickness, thus obtaining the predicted coating thickness. The deviation between the predicted coating thickness and the preset thickness threshold is calculated, and a preliminary evaluation index for quality stability is determined based on the deviation value.

[0011] In some optional embodiments, the step of optimizing the pulse parameters using a variability adjustment method based on the preliminary evaluation index to obtain the final uniform current distribution control scheme includes: Based on the preliminary evaluation indicators, the electric field distribution was recalculated using the finite element analysis method to obtain preliminary current distribution simulation results; Based on the preliminary current distribution simulation results, the pulse parameters are initially corrected using the variability rate adjustment method to obtain the adjusted parameter set. If the adjusted parameter set does not meet the termination condition, the simulation and parameter adjustment are repeated using the finite element analysis method to obtain a new optimized parameter set. If the optimized pulse parameter data meets the termination condition, then the final uniform current distribution control scheme is generated based on the current parameter set.

[0012] In some optional embodiments, after optimizing the pulse parameters using a variability adjustment method based on the preliminary evaluation index to obtain the final uniform current distribution control scheme, the method further includes: Based on the control scheme of the final uniform current distribution, update the surface data in the three-dimensional model to obtain the optimized surface representation model; Based on the surface representation model, the current distribution is calculated using the finite element analysis method to obtain simulated current density data; Based on the simulated current density data, the current density error is obtained by comparing it with the actual measured value; If the current density error is higher than the preset error threshold, the surface representation parameters are adjusted to obtain the adjusted surface representation model. Based on the adjusted surface representation model, the predicted coating thickness is calculated using the electrochemical deposition formula:

[0013] Where h is the coating thickness, η is the electrodeposition efficiency, J is the current density, t is the deposition time, M is the molar mass of the metal, n is the number of electrons transferred, F is the Faraday constant, and ρ is the density of the metal. The stability of the predicted coating thickness and the actual measured thickness are verified by statistical analysis to obtain an accuracy index.

[0014] This invention provides a dual-pulse high-frequency switching power supply control system for precision electroplating, comprising: The model building module is used to acquire surface data of complex workpieces using a 3D scanner to obtain a 3D model that includes local geometric changes. The mesh generation module is used to extract high curvature regions from the three-dimensional model, and to refine the mesh of the high curvature regions using the finite element analysis method to obtain the high curvature region mesh. The index calculation module is used to calculate the surface roughness index based on the high curvature region mesh using the local radius of curvature. The current simulation module is used to simulate the current field of the surface roughness index using the finite element analysis method, and calculates the current density vector distribution under the initial pulse parameters by applying the boundary voltage. The optimization module is used to optimize the pulse parameters using a genetic algorithm when the deviation of the current density vector distribution exceeds a preset deviation threshold, thereby obtaining the optimized pulse parameters. The matching update module is used to select the pulse parameter that best matches the surface roughness index from the optimized pulse parameters, and based on the best matching pulse parameter, iteratively calculate the current density distribution using electromagnetic field simulation software to obtain the updated current density vector distribution. The prediction module is used to predict the coating thickness value based on the difference between the updated current density vector distribution and the initial distribution using a pre-established linear regression model, and to determine the preliminary evaluation index of quality stability based on the difference between the predicted thickness value and the preset thickness threshold. The control module is adjusted to optimize the pulse parameters using a variability adjustment method based on the preliminary evaluation indicators, thereby obtaining a control scheme for a final uniform current distribution.

[0015] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit the invention.

[0016] The dual-pulse high-frequency switching power supply control method and system for precision electroplating of the present invention have the following beneficial effects: This invention acquires surface data of complex workpieces using a 3D scanner to obtain a 3D model containing local geometric changes. High-curvature regions are extracted from the 3D model, and a finite element method (FEM) is used to refine the mesh of these regions, resulting in a high-curvature region mesh. Based on this mesh, a surface roughness index is calculated using the local radius of curvature. A current field simulation is then performed on the surface roughness index using FEM, calculating the current density vector distribution under initial pulse parameters by applying a boundary voltage. If the deviation of the current density vector distribution exceeds a preset deviation threshold, a genetic algorithm is used to optimize the pulse parameters. The process involves: obtaining optimized pulse parameters; selecting the pulse parameter that best matches the surface roughness index from the optimized pulse parameters; using electromagnetic field simulation software to iteratively calculate the current density distribution based on the best-matching pulse parameter to obtain an updated current density vector distribution; using a pre-established linear regression model to predict the coating thickness based on the difference between the updated current density vector distribution and the initial distribution; determining a preliminary evaluation index for quality stability based on the difference between the predicted thickness value and a preset thickness threshold; and optimizing the pulse parameters using a variability rate adjustment method based on the preliminary evaluation index to obtain a final control scheme for uniform current distribution. This invention enables dynamic optimization of pulse parameters based on workpiece geometry, improving the uniformity of current distribution and the stability of coating quality. Attached Figure Description

[0017] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings.

[0018] Figure 1 This is a schematic flowchart of a dual-pulse high-frequency switching power supply control method for precision electroplating disclosed in an embodiment of the present invention. Figure 2This is a schematic diagram of the structure of a dual-pulse high-frequency switching power supply control system for precision electroplating disclosed in an embodiment of the present invention. Detailed Implementation

[0019] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the invention will be more comprehensive and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0020] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0021] The flowchart shown in the attached diagram is merely an illustrative example and does not necessarily include all steps. For example, some steps may be broken down, while others may be combined or partially combined. Therefore, the actual execution order may change depending on the specific circumstances.

[0022] like Figure 1 As shown, this embodiment of the invention provides a dual-pulse high-frequency switching power supply control method for precision electroplating. This method can dynamically optimize pulse parameters according to the workpiece geometry, thereby improving the uniformity of current distribution and the stability of plating quality.

[0023] S10. Model Establishment: First, a high-precision laser scanner is used to acquire data of the complex workpiece. Key areas are scanned at a resolution of 0.01 mm and a density of 100 points per square millimeter to form an initial point cloud. Next, the initial point cloud is denoised, and missing areas are filled in to obtain a complete and clean point cloud. Then, the Delaunay triangulation algorithm is used to construct a triangular mesh, with the unit size controlled at 0.02 mm to ensure accuracy, generating a preliminary 3D digital model. Afterward, the geometric continuity of the model surface is assessed. If discontinuous areas exist, the Laplacian smoothing algorithm is used, with 5 iterations and a smoothing factor of 0.3. This eliminates noise while preserving local geometric features, ultimately resulting in an optimized 3D digital model that closely matches the actual shape of the workpiece and includes local geometric variations.

[0024] S20. Mesh Generation: A preliminary scan of the constructed 3D model surface is performed to form a preliminary set of feature points. Next, based on this preliminary feature point set, a specialized curvature analysis algorithm based on point cloud data is used to calculate local curvature values. The core of calculating local curvature values ​​based on point cloud data is to first fit a quadratic surface using local points, and then derive the curvature based on the geometric properties of the fitted surface. Subsequently, a preset curvature threshold (e.g., 0.05mm) is set. The calculated local curvature values ​​are compared with a threshold, and regions with curvature values ​​exceeding the threshold are marked as high-curvature regions. Finally, the finite element method is used to refine the mesh of these marked high-curvature regions.

[0025] S30. Index Calculation: Based on the high curvature region obtained in S20, the corresponding maximum and minimum curvature radii are calculated according to the inverse relationship between the radius of curvature and the curvature. The average of the two is taken as the local radius of curvature of the mesh unit. On this basis, the surface roughness index is further calculated. The core logic is to quantify the degree of surface roughness by analyzing the difference in curvature radii between adjacent mesh units. Then, the roughness values ​​of all adjacent mesh units within the high curvature region are statistically analyzed. For example, the average value is taken as the overall roughness index of the region, or the maximum value is used to locate the local location with the most drastic roughness changes. The final roughness index transforms the abstract roughness of the workpiece surface into a quantifiable value. The larger the value, the more obvious the surface undulation and the more complex the shape of the region.

[0026] S40. Current Simulation: First, based on the quantified surface roughness index, a surface roughness model is established using finite element analysis tools such as ANSYS. Next, based on the established surface roughness model, the voltage application method is used to set the corresponding voltage conditions to determine the input parameters for the current field simulation. Finally, based on the determined current field simulation input parameters, the current density vector distribution is calculated using finite element analysis.

[0027] S50. Optimization: When it is necessary to determine whether the current density vector distribution needs optimization, the distribution deviation is first calculated by comparing local and global values. When the calculated distribution deviation exceeds a preset threshold (e.g., 20%), a genetic algorithm is initiated to optimize the pulse parameters. First, the initial conditions of the genetic algorithm are determined, based on a population size of 100 initial solutions. Each initial solution contains the core dimensions of the pulse parameters: the pulse amplitude range (initially set to 5-15 V) and the duty cycle step size (initially set to 0.1), forming a set of initial parameters to be optimized. Then, with minimizing the current density distribution deviation as the objective function, iterative optimization is performed through selection, crossover, and mutation operations of the genetic algorithm. After each generation of iteration, the current density distribution deviation is recalculated using the adjusted parameters to determine if it still exceeds the threshold. If it does not, the iteration continues until, after multiple generations of iteration, the parameters converge to a range that makes the deviation below the threshold. The resulting set of pulse parameters is the optimized pulse parameter.

[0028] S60. Matching and Updating: In the process of selecting the parameters that best match the surface roughness index from the optimized pulse parameters, a weighted scoring algorithm is used to quantify the degree of fit between the two. After determining the best-matching pulse parameters, the current density distribution is iteratively calculated using electromagnetic field simulation software based on these parameters. If the initially updated current density vector distribution differs from the preset uniformity standard, such as a deviation of less than 0.01 A / m... A new round of optimization needs to be initiated: First, the pulse parameters are readjusted using a genetic algorithm, based on an initial population of 100 solutions, with the objective function of minimizing the current density distribution deviation. The optimization is then iteratively performed through selection, crossover, and mutation operations using the genetic algorithm. Next, a weighted scoring algorithm is used to calculate the matching degree between the new parameters and the surface roughness index, selecting the optimal parameters for the new round. Finally, the parameters are imported back into the electromagnetic field simulation software to iteratively calculate the current density distribution.

[0029] S70. Prediction: Obtain the updated current density vector distribution from step S60. Using a matrix interpolation algorithm, subtract the current density values ​​of corresponding grid cells in the two distribution matrices one by one, and normalize the difference matrix to obtain a normalized difference matrix. Based on the current density change information in the difference matrix, predict the coating thickness change value through a pre-established linear regression model. This regression model is trained based on a large amount of historical process data. Then, calculate the difference between the predicted thickness value of all grid cells in the region and the preset thickness threshold, and then calculate the standard deviation of the predicted thickness value in the region through statistical calculation. Compare this standard deviation with the preset quality stability threshold (1.0 micrometer). If the standard deviation exceeds the threshold, it is initially assessed as "unstable".

[0030] S80. Adjustment Control: Preliminary evaluation indicators are input as key feedback information into the finite element analysis module to reconstruct the electric field calculation model and generate preliminary current distribution simulation results that better reflect actual deviation characteristics. Based on the preliminary current distribution simulation results, the system uses a variability rate adjustment method to initially correct the pulse parameters, generating a new set of optimized parameters. A control scheme is then generated based on the current parameter set. After the control scheme is imported into the control module, the current density distribution on the workpiece surface is collected in real time by sensors. The measured deviation is compared with the simulation results (e.g., in this case, the measured deviation of 6% is slightly higher than the simulation of 5%). If there are slight differences, the parameters are fine-tuned again through the feedback control algorithm until the uniformity of the actual current distribution reaches the design standard.

[0031] Through the above steps, this embodiment can dynamically optimize pulse parameters according to the workpiece geometry, thereby improving the uniformity of current distribution and the stability of coating quality.

[0032] In some embodiments, based on the above embodiments, the process of acquiring surface data of a complex workpiece using a 3D scanner to obtain a 3D model containing local geometric changes includes: First, during the data acquisition phase, a high-precision laser scanner was used to scan a critical 10cm × 10cm area of ​​the workpiece at a resolution of 0.01mm. Simultaneously, the point cloud density was controlled to 100 points per square millimeter, forming an initial point cloud dataset. Next, the initial point cloud data was processed. First, noise interference generated during the acquisition process was removed using point cloud denoising methods. If missing data areas existed in the cleaned point cloud, interpolation methods were used to fill in the gaps, thus obtaining a complete cleaned point cloud dataset.

[0033] When constructing a 3D model that includes local geometric changes, a surface mesh can be generated using point cloud data through a triangulation algorithm.

[0034] To capture surface undulations in high-curvature regions without causing data redundancy or increased computational load due to excessively small cell sizes, the mesh cell size is controlled at 0.02 mm. This is achieved by preprocessing the complete, clean point cloud before running the Delaunay triangulation algorithm. If the spacing between adjacent points in the original point cloud is less than 0.02 mm, some duplicate or redundant points will be removed without compromising key geometric features, bringing the spacing closer to 0.02 mm. If the spacing in some areas is greater than 0.02 mm, the point cloud will be densified based on the geometric trends of surrounding points, and a small number of transition points will be added to fill the gaps in spacing.

[0035] When processing point clouds, the Delaunay triangulation algorithm follows the principle of maximizing the minimum angle to connect discrete points: it traverses all points, analyzes spatial relationships, avoids connecting narrow or irregular triangles, and prioritizes constructing balanced triangular patches with interior angles close to 60 degrees. These triangular patches are the smallest geometric units constituting the surface of a 3D digital model, and are triangular in shape, serving as the core bridge for transforming discrete point cloud data into a continuous 3D surface. In 3D modeling of complex workpieces, three discrete points from a clean point cloud are used as vertices, connected into a triangle using the Delaunay triangulation algorithm. Numerous such small triangles are then used as building blocks to assemble the complete workpiece surface contour in 3D space. Each triangular patch has a definite spatial location; its vertices correspond to actual sampling points in the point cloud, and its three sides connect adjacent vertices, forming a triangular plane that approximately recreates the local morphology of the workpiece surface between these three points.

[0036] After the algorithm completes the triangulation of points, dense and uniformly sized triangular patches are stitched together in three-dimensional space. The vertices of the patches correspond to discrete points in the point cloud, and the edges connect to each other, forming a continuous triangular mesh structure covering the workpiece surface. This structure is the "digital contour" of the workpiece surface, reproducing the geometric information carried by the point cloud, including local changes in high-curvature areas, thereby generating a preliminary three-dimensional digital model whose accuracy and detail meet the needs of subsequent optimization.

[0037] Next, the geometric continuity of the preliminary model surface is assessed. For example, obvious abrupt angle changes at the joints of triangular facets, sharp creases in certain areas, or surface unevenness caused by minor errors in the point cloud will prevent the model from smoothly reproducing the actual surface morphology of the workpiece; these are considered geometric discontinuities. If geometric discontinuities exist, the Laplacian smoothing algorithm is used to process them. This algorithm eliminates discontinuities by adjusting the spatial position of each triangular facet vertex: it calculates the average position of all adjacent vertices around each vertex and then fine-tunes the current vertex towards this average position based on a preset smoothing factor. Setting 5 iterations and a smoothing factor of 0.3 means that each adjustment only moves the vertex 30% of the distance between the current position and the average position. Setting 5 iterations balances smoothing effect with detail preservation; a single adjustment has limited correction power, but 5 iterations gradually accumulate the smoothing effect, making the connection between discontinuous areas increasingly smooth. Through these parameter settings, noise is eliminated while local geometric features are preserved, ultimately resulting in an optimized 3D digital model that incorporates local geometric changes.

[0038] In some embodiments, based on the above embodiments, the step of extracting the high curvature region according to the three-dimensional model and using the finite element analysis method to refine the mesh of the high curvature region to obtain the high curvature region mesh includes: A preliminary set of feature points is obtained by performing a preliminary scan of the surface of the three-dimensional model; Based on the preliminary set of feature points, a curvature analysis algorithm based on point cloud data is used to calculate the local curvature value. The region whose local curvature value is higher than the preset curvature threshold is marked as a high curvature region, thus obtaining the high curvature region; The high curvature region was meshed using the finite element method to obtain the high curvature region mesh.

[0039] First, a preliminary scan of the constructed 3D model surface is performed. The core of this step is to quickly capture the approximate distribution range of geometric features on the model surface. For example, in the model of an aero-engine blade, areas with obvious geometric changes such as edges and tips are scanned and discrete points in these areas are collected to form a preliminary set of feature points.

[0040] Next, based on this preliminary set of feature points, a specialized curvature analysis algorithm based on point cloud data is used to calculate the local curvature values. The core of calculating local curvature values ​​based on point cloud data is to first fit a quadratic surface using local points, and then derive the curvature based on the geometric properties of the fitted surface. Because point clouds are discrete data, the curvature formula for continuous surfaces cannot be directly used; it is necessary to approximate them into a computable mathematical model through local surfaces.

[0041] Two preprocessing steps are required: First, select local neighborhood points for the target point (the point whose curvature is to be calculated). These are usually points within a radius r centered on the target point, or a fixed number of neighboring points, to ensure coverage of local geometric features. Second, establish a local coordinate system with the target point as the origin. Use principal component analysis (PCA) to determine the local normal vector (set as the z-axis) and tangent plane (xy-plane) of the point cloud. Transform the coordinates of the neighborhood points to this local system to eliminate interference from the global coordinates and simplify subsequent calculations.

[0042] Step 1: Fitting the local quadratic surface In the local coordinate system, the surface around the target point can be approximated as a quadratic surface (because the surface change in the high curvature region, i.e., the region with a small radius of curvature, is gradual, the quadratic surface can accurately fit the shape), and its simplified equation is:

[0043] in, These are the coordinates of the neighboring points in the local coordinate system; , It is the second-order partial derivative of the surface in the x and y directions, reflecting the degree of curvature in the x and y directions; It is the mixed second-order partial derivative of the surface, reflecting the coupled bending characteristics in the x and y directions. By fitting the coordinates of neighboring points using the least squares method, the solution can be obtained. , , The specific value is to minimize the sum of squared errors between the actual z-values ​​of neighboring points and the z-values ​​calculated by the formula. Step 2: Calculate the principal curvature (core curvature index). The local curvature of a point cloud is usually expressed as the principal curvature. This represents the maximum and minimum curvature of a surface in two orthogonal directions at a given point, and is a key indicator for quantifying the "degree of curvature." Its calculation formula originates from the "first fundamental form" and "second fundamental form" of a surface, and can be significantly simplified in a local coordinate system: Define the basic form coefficients: The first fundamental form coefficients are core parameters describing the local metric properties of a surface. The core principle is that the coefficients are defined by the dot product of vectors in two orthogonal directions within the tangent plane of the surface. They are typically expressed using... , , express.

[0044] coefficient The inner product of the tangent vectors in the x-direction; coefficients The inner product of the corresponding tangent vectors in the x and y directions, with coefficients The inner product of the tangent vector in the y-direction.

[0045] In the local coordinate system, the x-axis is a unit vector (of length 1) within the tangent plane, and its inner product is 1. Therefore... .

[0046] Since the x-axis and y-axis are orthogonal in the tangent plane, according to the property of the dot product of vectors, the dot product of orthogonal vectors is 0. .

[0047] and Similarly, the y-axis is also a unit vector in the tangent plane, and its inner product with itself is 1. Therefore .

[0048] The second fundamental form coefficients are core parameters characterizing the local normal curvature of a surface. They primarily describe the direction and degree of curvature of the surface in the direction perpendicular to the tangent plane (i.e., the normal). In the previous point cloud curvature analysis scenario, the x and y planes of the local coordinate system were set as the tangent planes of the surface, and the z-axis as the surface normal vector. At this time, the curvature of the surface is only reflected in the z-axis direction. The second fundamental form coefficients are essentially the second-order normal derivative characteristics of the surface in this local coordinate system, specifically corresponding to three coefficients. Furthermore, since the normal is fixed at the z-axis, they can be directly correlated with the second-order partial derivatives of the quadratic surface obtained from surface fitting. From the second-order partial derivatives, we can directly obtain: The degree of curvature of the surface in the x-direction corresponds to the second partial derivative of the quadratic surface in the x-direction.

[0049] The coupled normal curvature of the corresponding surface in the x and y directions is equal to the mixed second-order partial derivative of the quadratic surface.

[0050] The degree of curvature of the surface in the y-direction corresponds to the second partial derivative of the quadratic surface in the y-direction.

[0051] Solving for the principal curvature: The principal curvatures are the two roots of the following characteristic equation:

[0052] Will and , , Substituting the values ​​and expanding, we get a quadratic equation:

[0053] The principal curvatures, maximum and minimum curvature, can be solved using the quadratic formula:

[0054] After obtaining the maximum and minimum curvatures, the average of the two is taken as the local curvature of the mesh element.

[0055] Then, a preset curvature threshold (e.g., 0.05mm) is set. The calculated local curvature values ​​are compared with a threshold, and areas with curvature values ​​higher than the threshold are marked as high curvature regions. The purpose of this step is to accurately screen out the areas with the most dramatic geometric changes on the model surface, such as blade edges and tips, where the curvature values ​​are significantly higher than other areas, and these will be clearly defined as high curvature regions that require special attention.

[0056] Finally, the finite element method is used to refine the mesh of these marked high curvature regions. Due to the complex surface morphology and rich details of the high curvature regions, the initial conventional mesh cannot meet the subsequent accuracy requirements. Therefore, the software will adaptively adjust and reduce the mesh size of the high curvature regions (e.g., to 0.2 mm), so that the mesh density of the region is increased by at least 5 times, and finally a high-density mesh of the high curvature regions is obtained.

[0057] In some embodiments, based on the above embodiments, the surface roughness index is calculated using the local radius of curvature according to the high curvature region mesh, including: Based on the high curvature region obtained from S20, the corresponding maximum and minimum curvature radii are calculated according to the inverse relationship between the radius of curvature and the curvature. The average of these two values ​​is taken as the local radius of curvature of the mesh unit. Further, the surface roughness index is calculated. The core logic is to quantify the degree of surface roughness by analyzing the difference in curvature radii between adjacent mesh units. Then, the roughness values ​​of all adjacent mesh units within the high curvature region are statistically analyzed. For example, the average value is taken as the overall roughness index of the region, or the maximum value is used to locate the local location with the most drastic roughness changes. The final roughness index transforms the abstract roughness of the workpiece surface into a quantifiable value. The larger the value, the more obvious the surface undulation and the more complex the shape of the region.

[0058] Based on the high curvature region obtained from S20, calculations are performed according to the inverse relationship between the radius of curvature and the curvature.

[0059] in Let be the radius of curvature. For curvature.

[0060] Calculate the corresponding maximum and minimum radii of curvature, and take the average of the two as the local radius of curvature of the mesh unit, which represents the curvature of the workpiece region corresponding to the entire mesh unit. Based on this, the surface roughness index is further calculated. The core logic is to quantify the degree of surface roughness by analyzing the difference in the radius of curvature between adjacent grid cells.

[0061] If the radius of curvature of adjacent grids changes drastically, it indicates that there are significant bumps and dips on the workpiece surface corresponding to these two elements; if the change is gradual, the surface is relatively smooth. The specific calculation formula is as follows:

[0062] in, For concavity and convexity index, This represents the difference in local radii of curvature between two adjacent mesh elements. This represents the area of ​​a single grid cell.

[0063] Then, the concavity and convexity values ​​of all adjacent grid cells in the high curvature region will be statistically analyzed. For example, the average value will be taken as the overall concavity and convexity index of the region, or the maximum value will be used to locate the local position where the concavity and convexity changes are most drastic.

[0064] The resulting surface roughness index can transform the abstract surface roughness of a workpiece into a quantifiable value. The larger the value, the more obvious the surface undulation and the more complex the shape of the area, providing accurate data for subsequent judgment of the processing difficulty of the area and optimization of the processing path.

[0065] In some embodiments, based on the above embodiments, the step of using finite element analysis to simulate the current field of the surface roughness index, and calculating the current density vector distribution under the initial pulse parameters by applying a boundary voltage, includes: Based on the surface roughness index, a model is established using the finite element analysis method to obtain the surface roughness model; Based on the surface roughness model, the voltage conditions are set using the voltage application method to obtain the electric field distribution under the initial pulse value and determine the input parameters for the current field simulation. Based on the input parameters of the current field simulation, the current density vector distribution is obtained by calculation using the finite element analysis method.

[0066] The current field simulation of surface roughness index and the calculation of current density vector distribution under initial pulse parameters using finite element analysis should be carried out step by step with data modeling, parameter setting and calculation solution as the core logic.

[0067] First, based on the quantified surface roughness index, a surface roughness model needs to be established using finite element analysis tools such as ANSYS. This step requires transforming the abstract roughness index into a concrete three-dimensional geometric model. Specifically, the roughness height data of each region is mapped to the surface undulations of the geometric model. To ensure simulation accuracy, triangular elements are used for mesh generation, allowing the mesh to accurately fit the surface roughness of the model. Ultimately, a surface roughness model that can be used for current field simulation is formed, providing the basic geometric basis for subsequent simulations.

[0068] Next, based on the established surface roughness model, the voltage application method is used to set the corresponding voltage conditions to determine the input parameters for the current field simulation. In practice, a preset boundary voltage value (e.g., 5V) is applied to specific boundaries of the model according to the simulation requirements. At the same time, the key physical parameters of the corresponding material are entered. These voltage conditions and material parameters together constitute the core input parameters for the current field simulation. When the voltage is applied to the model boundary, a corresponding electric field is formed inside the model. The electric field distribution under the initial pulse value (e.g., the spatial variation of the electric field intensity) can be obtained in real time using finite element tools. The electric field distribution data further provides a key basis for the subsequent calculation of current density, realizing the connection from parameter setting to electric field analysis.

[0069] Finally, based on the determined input parameters for the current field simulation, the current density vector distribution is calculated using the finite element method. The core of this step is based on the fundamental physical laws of the current field, utilizing the current density formula... Solve the problem.

[0070] in For current density, The conductivity of the material, This is the potential gradient, which is directly related to the previously obtained electric field distribution.

[0071] Finite element tools divide the surface roughness model into a large number of tiny mesh elements and calculate the potential gradient for each element separately. Combined with the known conductivity The magnitude and direction of the current density of each unit are obtained one by one through the formula, and then the data of all units are integrated to form a complete vector space distribution of current density.

[0072] In some embodiments, based on the above embodiments, when the deviation of the current density vector distribution exceeds a preset deviation threshold, the pulse parameters are optimized using a genetic algorithm to obtain optimized pulse parameters, including: Based on the current density vector distribution, the distribution deviation of the current density vector is obtained by comparing local and global values; When the distribution deviation of the current density vector exceeds a preset deviation threshold, a genetic algorithm is used for optimization to obtain optimized pulse parameters.

[0073] When it is necessary to determine whether the current density vector distribution needs optimization, the distribution deviation is first calculated by comparing local and global values. Specifically, key local data in the current density vector distribution are extracted first, and then the global average current density of that region or the entire model is calculated. The deviation rate formula is used to calculate the current density distribution deviation in that local region, thereby clarifying the degree of spatial non-uniformity of the current density.

[0074] When the calculated distribution deviation exceeds a preset threshold (e.g., 20%), the genetic algorithm is activated to optimize the pulse parameters. First, the initial conditions of the genetic algorithm are determined, based on a population size of 100 initial solutions. Each initial solution contains the core dimensions of the pulse parameters: the pulse amplitude range (initially set to 5-15 V) and the duty cycle step size (initially set to 0.1), forming a set of initial parameters to be optimized.

[0075] Next, with minimizing the current density distribution deviation as the objective function, iterative optimization is achieved through selection, crossover, and mutation operations using a genetic algorithm. The selection operation retains the best parameter solutions that minimize the deviation in each generation (e.g., retaining the 50 solutions with the highest fitness in each generation). The crossover operation integrates the parameter characteristics of different best solutions (e.g., combining the amplitude range and duty cycle step size of different solutions). The mutation operation slightly adjusts some parameters to avoid getting trapped in local optima (e.g., fine-tuning the amplitude or duty cycle step size). After each iteration, the current density distribution deviation is recalculated using the adjusted parameters to determine if it still exceeds the threshold. If it does not, iteration continues until, after multiple iterations, the parameters converge to a range that keeps the deviation below the threshold. The resulting set of pulse parameters is the optimized pulse parameter set. These parameters can then be imported into the control module to adjust the output signal to make the current density distribution more uniform, thus solving the problem of uneven distribution in the early stages.

[0076] In some embodiments, based on the above embodiments, the pulse parameter that best matches the surface roughness index is selected from the optimized pulse parameters. Based on the best-matching pulse parameter, the current density distribution is iteratively calculated using electromagnetic field simulation software to obtain an updated current density vector distribution, including: The matching degree between the optimized pulse parameters and the surface roughness index is calculated using a weighted scoring algorithm, and the pulse parameter with the highest matching degree is taken as the best matching pulse parameter. Based on the best-matching pulse parameters, the current density distribution is iteratively calculated using electromagnetic field simulation software to obtain a preliminary updated current density vector distribution. If the initially updated current density vector distribution deviates from the preset uniformity standard, the pulse parameters are re-optimized using a genetic algorithm, the matching degree is calculated, and the current density distribution is iteratively calculated using electromagnetic field simulation software to obtain the updated current density vector distribution.

[0077] In the process of selecting the parameters that best match the surface roughness index from the optimized pulse parameters, the degree of fit between the two is quantified by a weighted scoring algorithm.

[0078] Matching degree = 0.6 × (1 - |target voltage - actual voltage| / target voltage) + 0.4 × (1 - |target frequency - actual frequency| / target frequency) The target voltage and target frequency can be obtained by filtering initial reference values ​​from a historical process database. In industrial settings, long-term accumulation of correlation data on the "surface roughness-pulse parameters-current uniformity" of similar workpieces allows us to find corresponding historical parameters from the surface roughness of the current workpiece as initial reference values. The actual voltage and actual frequency are obtained from the optimized pulse parameters.

[0079] The matching degree value always ranges between 0 and 1, where 1 represents a perfect match between the parameter and the surface roughness index, and 0 represents a complete mismatch. The parameter combination with the highest matching degree is selected by comparing the values, and this combination is the pulse parameter that best matches the surface roughness index.

[0080] After determining the most suitable pulse parameters, the current density distribution is iteratively calculated using electromagnetic field simulation software. The software first needs to recreate the real-world scenario. This includes importing the workpiece's 3D geometric model and assigning physical properties to the workpiece material (e.g., the conductivity of metal is 5.8e7 S / m, determining its current conduction capability); simultaneously replicating the boundary conditions during processing, such as the electrode positions (voltage applied to the upper and lower surfaces), the contact method between the electrodes and the workpiece (whether they are tightly fitted, affecting contact resistance), and even environmental factors (e.g., temperature, which can be ignored at room temperature, but requires additional thermal conductivity parameters under high-temperature conditions). This information collectively constitutes the simulation's "digital twin," ensuring that the calculated scenario matches the actual processing.

[0081] The software relies on finite element analysis and the decomposition of electromagnetic field equations. The finite element method divides complex 3D workpieces into 100×100 (or finer) mesh elements, each treated as a tiny homogeneous body, simplifying the overall computational complexity. Meanwhile, electromagnetic field equations (such as the Laplace equation in steady state) are solved. The calculation basis is provided by the time-varying Maxwell's equations (or the time-varying Maxwell's equations).

[0082] If the pulse frequency in the pulse parameters is extremely low (e.g., 0.1 Hz), the electromagnetic field changes very slowly with time and can be approximated as a steady-state field. In this case, the Laplace equation can be used. The physical meaning of this equation is that "there is no charge accumulation inside the workpiece, and the potential distribution satisfies 'smooth change'". Substituting the boundary voltage (e.g., V=55V for the upper surface and V=0V for the lower surface), the Laplace equation is applied to the mesh element, and numerical methods (e.g., the Gauss-Seidel iteration method) are used to solve for the potential distribution.

[0083] Set initial potential values ​​for all grid points. For example, set them all to 27.5V, which is the average of 55V and 0V. Set the boundary conditions as V=55V for the upper surface and V=0V for the lower surface.

[0084] The Laplace equation in the two-dimensional case is:

[0085] For each internal grid point ( The equation can be discretized using the finite difference method as follows:

[0086] in, The x-coordinate of the internal grid point. y is the ordinate of the internal grid point.

[0087] This formula means that the potential of each grid point is equal to the average of the potentials of its four neighboring points.

[0088] Calculate the new potential using the formula, and immediately update the potential of the current grid point with the new value. Repeat the above process until the potential change of all grid points in two adjacent iterations is less than the preset tolerance (e.g., 1e-6).

[0089] After solving for the potential V of each grid cell, the current density formula is used.

[0090] in, For current density, The conductivity of the material, This represents the potential gradient.

[0091] The current density of each unit is then calculated, which is the calculation logic for the current distribution under steady state.

[0092] If the frequency is not negligible (the field changes periodically with time), then two of the formulas in Maxwell's equations are required.

[0093]

[0094] in, The curl of the electric field. It is the rate of change of the magnetic field with respect to time. J is the curl of the magnetic field, and J is the conduction current density. It is the displacement current density.

[0095] The iterative solution using time-stepping involves calculating the field at the current time step using the field distribution from the previous time step. The specific steps are as follows: At the initial time (t=0), set the initial field distribution (e.g., E=0, B=0), where E is the electric field and B is the magnetic field.

[0096] Transforming pulse voltage into a "time-electric field" excitation function, such as a square wave voltage. The load is applied to the electrode boundary (upper surface), and the result is converted into an electric field at the boundary through integration. .

[0097] Time step iteration (taking t→t+Δt as an example): Step 1: Use calculate

[0098] According to Faraday's law Substitute The spatial distribution of the electric field is calculated using a discretized curl operator to determine the electric curl of each grid cell, and then combined with... and Solving for the given information yields the following results. electric field at time At this point, the protrusion is close to the electrode. The initial value will be higher.

[0099] in, By Faraday's law of electromagnetic induction time derivative The forward difference approximation yields: In a non-magnetic medium, D = εE, therefore ∂D / ∂t = ε∂E / ∂t. The Ampère-Maxwell law is... Therefore Using the forward difference approximation, we obtain .

[0100] Step 2: Use calculate

[0101] According to Ampere's law ,in , .in It is the dielectric constant. Adding the two together gives the right-hand side term, which is then solved using the inverse operation of the curl operator. Magnetic field of time And thus obtain .

[0102] Step 3: Verification and Correction Check whether the field changes at adjacent time steps satisfy energy conservation (e.g., whether the sum of electric field energy and magnetic field energy is reasonable). If the deviation exceeds the threshold, make fine adjustments. Alternatively, recalculate after adjusting the mesh accuracy.

[0103] Full-cycle solution and current density extraction: Repeat the above time-step iterations, covering the entire 12Hz cycle (approximately 83,000 steps), to obtain the E(t) distribution for each time step and each grid cell. Then, through... Calculate the dynamic current density, for example, at the rising edge of a pulse (voltage jumps from 0 to 55V), at the bulge. Instantly increase Rapidly rises to peak value (e.g., 0.6 A / m) During the peak of the pulse, the field tends to stabilize, and J(t) decreases slightly (e.g., to 0.55 A / m). ); at the falling edge of the pulse, Follow Decrease and reduce (e.g., 0.4 A / m) Through this dynamic calculation, the software can fully capture the fluctuation characteristics of current density in each cycle, especially the instantaneous concentration phenomenon in high curvature regions.

[0104] If the initially updated current density vector distribution differs from the preset uniformity standard, the deviation is less than 0.01 A / m. The deviation is calculated using the deviation rate = |local value - global average| / global average × 100%. A new round of optimization needs to be initiated: first, the pulse parameters are readjusted using a genetic algorithm, based on an initial population of 100 solutions, with the objective function being minimizing the current density distribution deviation. Iterative optimization is performed through selection, crossover, and mutation operations of the genetic algorithm. Then, a weighted scoring algorithm is used to calculate the matching degree between the new parameters and the surface roughness index, selecting the optimal parameters for the new round. Subsequently, the parameters are imported back into the electromagnetic field simulation software for iterative calculation of the current density distribution. This process is repeated until the deviation of the obtained current density vector distribution converges to within the uniformity standard. This updated current density vector distribution meets the requirements, adapting to the surface roughness of the workpiece while ensuring current uniformity, providing accurate data support for subsequent process adjustments.

[0105] In some embodiments, based on the above embodiments, a pre-established linear regression model is used to predict the coating thickness value based on the difference between the updated current density vector distribution and the initial distribution. Based on the difference between the predicted thickness value and a preset thickness threshold, preliminary evaluation indicators for quality stability are determined, including: The initial current density vector distribution data is obtained through numerical simulation software. Based on the initial distribution, the difference matrix between the updated current density vector distribution and the initial distribution is calculated using the matrix difference algorithm. Based on the difference matrix, a pre-established regression model is used to predict the change in coating thickness, thus obtaining the predicted coating thickness. The deviation between the predicted coating thickness and the preset thickness threshold is calculated, and a preliminary evaluation index for quality stability is determined based on the deviation value.

[0106] First, the updated current density vector distribution is obtained from step S60. Then, a matrix interpolation algorithm is used to subtract the current density values ​​of corresponding grid cells in the two distribution matrices one by one (e.g., the difference for a planar region is 0.5 A / dm). The difference in high curvature regions is 0.6 A / dm. The difference matrix is ​​then normalized (to eliminate the influence of differences in the initial current density base in different regions) to obtain the normalized difference matrix, which is the core input data for subsequent prediction of coating thickness.

[0107] Next, based on the current density variation information in the difference matrix, a pre-established linear regression model is used to predict the coating thickness variation. This regression model is not constructed out of thin air, but is trained on a large amount of historical process data. In past electroplating experiments, the actual coating thickness increment corresponding to different current density change rates is recorded, and a mathematical relationship between "current density change rate - coating thickness increment" is established through linear fitting. Therefore, when the relative change rate in the difference matrix is ​​input (e.g., 0.1 for planar areas and 0.1875 for high curvature areas), the model can directly calculate the corresponding predicted coating thickness value. For example, the predicted increase is 5.0 micrometers for planar areas and 6.0 micrometers for high curvature areas, forming a coating thickness prediction value distribution covering the entire workpiece. The predicted value for high curvature areas is the focus of subsequent quality assessment.

[0108] Finally, the preliminary evaluation index of quality stability is determined based on the difference between the predicted coating thickness in the high curvature region and the preset thickness threshold. In this process, the key thresholds need to be clarified first: one is the preset thickness threshold used to calculate the deviation, and the other is the quality stability threshold used to judge the stability (such as the standard deviation of the deviation between the predicted thickness value and the preset thickness threshold being less than 1.0 micrometers, which is considered stable).

[0109] There is no fixed standard for preset thickness thresholds. The threshold must be determined comprehensively based on the functional requirements of the coating, the characteristics of the substrate, the application scenario, and the processing capabilities. From a functional perspective, if the coating is used for basic protection (such as galvanizing steel parts for corrosion prevention), the preset thickness threshold is usually set in conjunction with the intensity of the corrosive environment. For example, in a normal atmospheric environment, it can be set to 8-15 micrometers to meet basic rust prevention needs, while in a marine high-salt spray environment, it needs to be increased to 20-30 micrometers to enhance corrosion resistance. If used for precision electronic components, the threshold is often set to 1-5 micrometers to balance conductivity and dimensional accuracy. If used for wear parts, the threshold is often set to 10-50 micrometers to meet wear resistance requirements. Furthermore, industry standards or product specifications must be referenced when setting the threshold (e.g., coatings for automotive parts must comply with ISO 14588, and coatings for electronic components must comply with IPC-6012).

[0110] In practice, the difference between the predicted coating thickness of all grid cells in the high curvature region and the preset thickness threshold is first calculated. Then, the standard deviation of these deviations is calculated statistically, such as 0.8 micrometers. This standard deviation is compared with the preset quality stability threshold (1.0 micrometer). Since 0.8 micrometers is less than 1.0 micrometer, it indicates that the deviation between the predicted coating thickness of the high curvature region and the preset thickness threshold is evenly distributed, with no significant local over-thickness or under-thinness. Based on this, the preliminary assessment index of the quality stability of this region can be determined as "qualified". If the standard deviation exceeds the threshold, the preliminary assessment is "unstable", and further optimization of process parameters is required.

[0111] In some embodiments, based on the above embodiments, the pulse parameters are optimized using a variability rate adjustment method according to the preliminary evaluation index to obtain a control scheme for the final uniform current distribution, including: Based on the preliminary evaluation indicators, the electric field distribution was recalculated using the finite element analysis method to obtain preliminary current distribution simulation results; Based on the preliminary current distribution simulation results, the pulse parameters are initially corrected using the variability rate adjustment method to obtain the adjusted parameter set. If the adjusted parameter set does not meet the termination condition, the simulation and parameter adjustment are repeated using the finite element analysis method to obtain a new optimized parameter set. If the optimized pulse parameter data meets the termination condition, then the final uniform current distribution control scheme is generated based on the current parameter set.

[0112] The process of optimizing pulse parameters based on preliminary evaluation indicators to obtain the final uniform current distribution control scheme is a closed-loop optimization logic of "feedback-simulation-adjustment-iteration". The core lies in guiding the fine-tuning of parameters through preliminary evaluation data until the preset standards are met. First, the preliminary evaluation indicators are input into the finite element analysis module as key feedback information. These indicators essentially reflect the gap between the current distribution and the ideal state. The finite element module reconstructs the electric field calculation model based on these deviation data. This may include adjusting the mesh refinement strategy (refining the mesh in high-deviation regions to capture finer field distributions) and correcting boundary conditions (such as adjusting the electrical connection model between the electrode and the workpiece based on actual measured contact resistance values). This process then recalculates the electric field distribution, generating preliminary current distribution simulation results that more closely match the actual deviation characteristics.

[0113] Based on the preliminary current distribution simulation results, the system employs a mutation rate adjustment method to initially correct the pulse parameters. This method typically relies on a genetic algorithm framework, where the mutation rate controls the magnitude and randomness of parameter mutations. This avoids overly conservative parameter adjustments (making it difficult to escape local optima) while also preventing excessive mutations that could lead to instability. Specifically, an initial population (pulse width in the range of 40-60 microseconds, frequency in the range of 1-5 kHz) is constructed based on the current pulse parameters. The mutation rate is used to randomly adjust some parameters within the population (e.g., randomly adjusting the width of a parameter from 50 microseconds to 48 microseconds, or the frequency from 2 kHz to 3.2 kHz). Simultaneously, a high-quality parameter set is selected based on the objective function of minimizing the current distribution deviation, resulting in an adjusted parameter set. The core of this step is to ensure that parameter adjustments are both targeted at the problems revealed by the simulation results (e.g., shortening the pulse width to address local overheating) and retain a degree of randomness to explore better solutions.

[0114] The iterative verification phase then begins: the adjusted parameter set is input into the finite element module to resimulate the current distribution, calculating the new deviation index and comparing it with preset termination conditions (such as deviation less than 5% or iteration count reaching 20). If the conditions are not met (e.g., the deviation is still 10% after the first adjustment), the new simulation results are used as feedback, and the "variation adjustment-simulation verification" process is repeated. For example, the pulse width range is further reduced to 45-50 microseconds, and the variation rate is increased to 0.15 to enhance the search strength, generating a new optimized parameter set. If the conditions are met (e.g., the deviation drops to 5% after the 10th iteration in this case, with a maximum value of 2.1 A / cm), the process continues. Then, an intermediate control scheme is generated based on the current parameter set, and a genetic algorithm is used for final fine-tuning to ensure that the scheme is stable and effective in practical applications.

[0115] The final uniform current distribution control scheme needs to be verified in practice: After the scheme is imported into the control module, the current density distribution on the workpiece surface is collected in real time by the sensor, and the measured deviation is compared with the simulation results (for example, the measured deviation of 6% in the case is slightly higher than the simulation of 5%). If there are slight differences, the parameters are fine-tuned again through the feedback control algorithm until the uniformity of the actual current distribution reaches the design standard, forming a complete closed loop from evaluation feedback to parameter optimization and then to actual application verification, ensuring that the final scheme can meet the theoretical simulation requirements and adapt to the complex factors of the real processing environment.

[0116] In some embodiments, based on the above embodiments, after optimizing the pulse parameters using the variability rate adjustment method according to the preliminary evaluation index to obtain the final uniform current distribution control scheme, the method further includes: Based on the control scheme of the final uniform current distribution, update the surface data in the three-dimensional model to obtain the optimized surface representation model; Based on the surface representation model, the current distribution is calculated using the finite element analysis method to obtain simulated current density data; Based on the simulated current density data, the current density error is obtained by comparing it with the actual measured value; If the current density error is higher than the preset error threshold, the surface representation parameters are adjusted to obtain the adjusted surface representation model. Based on the adjusted surface representation model, the predicted coating thickness is calculated using the electrochemical deposition formula:

[0117] Where h is the coating thickness, η is the electrodeposition efficiency, J is the current density, t is the deposition time, M is the molar mass of the metal, n is the number of electrons transferred, F is the Faraday constant, and ρ is the density of the metal. The stability of the predicted coating thickness and the actual measured thickness are verified by statistical analysis to obtain an accuracy index.

[0118] The process of updating the 3D model surface data and verifying the accuracy of coating thickness prediction based on the final uniform current distribution control scheme is a closed-loop process of "model optimization - simulation verification - error correction - accuracy confirmation". The core is to improve the reliability of current field simulation and thickness prediction by refining the surface model. First, based on the final uniform current distribution control scheme (such as a pulse parameter combination that reduces the current density standard deviation to 0.05), the surface data of the 3D model needs to be updated. The initial model may contain 10,000 triangular facets, but the mesh accuracy of surface details (especially in high-curvature areas) is insufficient to accurately reflect the impact of the optimized current distribution on the surface morphology. Therefore, local mesh refinement is first performed on high-curvature areas, i.e., areas with small curvature radii (such as areas with a curvature radius less than 5 mm and accounting for 15% of the total surface area), increasing the number of triangular facets in these areas from 1500 to 3000. Simultaneously, based on the current distribution data in the control scheme (such as 0.9-1.1 A / cm² after optimization), the mesh is updated. By adjusting the potential properties of the surface mesh nodes within a uniform range, the geometric features of the model surface are matched with the current field distribution characteristics, ultimately generating an optimized surface representation model. This model can both preserve the overall shape and accurately depict the details of small curvature radius regions, providing a more realistic digital carrier for subsequent simulations.

[0119] Next, a current field simulation environment is constructed using the optimized surface representation model. The current distribution is calculated using finite element analysis methods (such as ANSYS's electromagnetic field module): the model is divided into 1000 (or 1500 after refinement) elements, the pulse parameters from the final control scheme are input, and the potential value of each element is solved (e.g., 0.1-0.8 volts, with an error controlled within 0.01 volts). Then, the simulated current density data is obtained through the relationship between current density and potential (J=σ▽V). Subsequently, these simulated data are compared with the current density values ​​measured by sensors during actual processing (e.g., simulated value 0.5 A / m). Compared with the measured value of 0.48 A / m The relative error (4% in this example) is calculated to determine the reliability of the model.

[0120] If the calculated current density error exceeds a preset threshold (e.g., 5%), it indicates that the surface representation model still lacks detail (e.g., insufficient mesh density leads to the averaging of local current distribution), and the surface representation parameters need to be adjusted. For example, the overall mesh number can be increased from 1000 to 1500, or the mesh in high-curvature regions, i.e., regions with small curvature radii, can be further refined to control the local current density error within ±0.02 A / cm. Within this process, the adjusted surface representation model is regenerated and simulated again until the error drops below the threshold.

[0121] After obtaining simulated current density data that meets the error requirements, the coating thickness is predicted using the electrochemical deposition formula:

[0122] Where h is the coating thickness (unit: meters, converted to micrometers for calculation), η is the electrodeposition efficiency (decimal form, e.g., 90% = 0.9), and J is the current density (A / dm³). ), t is the deposition time (seconds), M is the molar mass of the deposited metal (g / mol), n is the number of electrons transferred, F is the Faraday constant (96485 C / mol), and ρ is the density of the deposited metal (g / cm³). ) The formula needs to include parameters such as current density, electrodeposition efficiency (e.g., 90%), and deposition time (e.g., 3600 seconds), for example, based on a simulated value of 0.5 A / m. The current density was used to calculate the predicted thickness to be 25.6 micrometers.

[0123] The predicted value was then compared with the actual measured coating thickness (e.g., 25.0 micrometers), and the stability of the predicted value was verified using statistical analysis methods (e.g., calculating the root mean square error). In this example, the root mean square error was 0.3 micrometers, indicating that the predicted value had small fluctuations and high stability.

[0124] Finally, the final accuracy index is determined based on the degree of agreement between the predicted thickness and the measured value (e.g., an accuracy of 97.6%). If this index is better than the historical average (e.g., an improvement of 2.6 percentage points from 95%), it proves that the optimized surface representation model is effective and can be used for subsequent process adjustments (e.g., fine-tuning the current density to 0.52 A / m). This provides a reliable basis for further improving uniformity, completing the entire logical chain from model optimization to prediction verification.

[0125] like Figure 2 As shown, this embodiment of the invention provides a dual-pulse high-frequency switching power supply control system for precision electroplating. Through integrated modular design, the system achieves uniform current distribution and stable plating quality.

[0126] Model Building Module 10: As the foundation of the entire system, it is responsible for achieving high-precision digital replication of high-curvature areas of complex workpieces, i.e., areas with small curvature radii. It acquires the initial point cloud of key areas through a high-precision laser scanner with a resolution of 0.01mm and a density of 100 points per square millimeter. After denoising and completion processing to obtain a clean and complete point cloud, the Delaunay triangulation algorithm is used to generate an initial triangular mesh with a unit size of 0.02mm. Then, the geometric discontinuities are processed through a Laplacian smoothing algorithm with 5 iterations and a smoothing factor of 0.3. While preserving local features, noise is eliminated, and finally, an optimized 3D digital model that fits the real shape of the workpiece is output, providing a precise geometric carrier for subsequent modules.

[0127] Mesh Generation Module: 20: Continuing from the output of the model building module, this module focuses on balancing simulation accuracy and computational efficiency. It first fits a quadratic surface based on the model's point cloud data, derives the local curvature value through the surface's geometric properties, and then sets the curvature to 0.05mm. The curvature threshold is used to mark areas with small protrusions, depressions, and edges where the local curvature value is higher than the threshold as high curvature areas. Finally, the finite element analysis method is used to refine the mesh in the high curvature areas, while maintaining the normal mesh density in the flat areas. This generates an adaptive mesh model containing the high curvature area refinement information, providing an adapted mesh basis for subsequent index calculations and current simulations.

[0128] The index calculation module 30 takes the adaptive mesh model as input and transforms the abstract concave and convex shape of the workpiece surface into quantifiable values. First, based on the inverse relationship between curvature and radius of curvature, it calculates the maximum and minimum radii of curvature of each mesh unit and takes the average value as the local radius of curvature of the unit. Then, it quantifies the degree of surface concavity and convexity by analyzing the difference in the radius of curvature of adjacent mesh units. Finally, it statistically analyzes the concavity and convexity values ​​of all units in the high curvature region, takes the average value as the overall concavity and convexity of the region, and locates the local position with the most drastic concavity and convexity changes by the maximum value. The output surface concavity and convexity index provides a basis for judgment for subsequent current simulation and parameter optimization.

[0129] Current simulation module 40: Combining surface roughness index and adaptive mesh model, a surface roughness simulation model is built in finite element analysis tools such as ANSYS. The voltage application method is used to set the current field simulation input parameters such as electrode voltage and electrode position. The electromagnetic field equation is solved by finite element analysis method to obtain the initial current density vector distribution of each mesh element, providing initial current distribution data for pulse parameter optimization.

[0130] Optimization Module 50: Taking the initial current density distribution and surface roughness index as input, and aiming at "uniform current density distribution", it achieves multi-round optimization: First, the genetic algorithm is started, with the objective function of minimizing the deviation of current density distribution. The population size of 100 initial solutions, the pulse amplitude range of 5-15V and the duty cycle step size of 0.1 are set. Iteration is carried out through selection, crossover and mutation operations until the deviation is lower than the 20% threshold.

[0131] Matching and updating module 60: Quantifies the fit between parameters and concavity / convexity using a weighted scoring algorithm (voltage weight 0.6, frequency weight 0.4), and selects the parameter with the highest matching degree; if the current density deviation after substituting this parameter into the electromagnetic field simulation does not reach 0.01 A / m... If the uniformity standard is found, the process of "genetic algorithm adjustment - weighted scoring screening - simulation verification" is repeated to finally output the optimized pulse parameter set that best matches the surface unevenness.

[0132] Prediction Module 70: Receives the optimized current density distribution, calculates the difference matrix between the "optimized" and "initial" current densities using a matrix interpolation algorithm, and normalizes it. The normalized result is input into a linear regression model trained based on historical process data to predict the coating thickness variation. Then, it calculates the difference between the predicted thickness value and the preset thickness threshold within a small curvature radius region, obtains the standard deviation through statistics, and compares it with the preset threshold of 1.0 micrometers. If the standard deviation is ≤ the threshold, it is initially assessed as "qualified"; otherwise, it is assessed as "unstable." The module outputs the predicted coating thickness value and the preliminary assessment result of quality stability.

[0133] Adjusting control module 80: As a closed-loop terminal, it inputs the preliminary evaluation results into the finite element analysis module to reconstruct the electric field calculation model and generate a current distribution simulation result that fits the actual deviation; it uses the variability rate adjustment method to initially correct the pulse parameters, generate a new set of optimized parameters, and construct an intermediate control scheme; after importing the intermediate scheme into the control module, it collects the current density distribution on the workpiece surface in real time through sensors, compares the measured deviation with the simulation results (e.g., the difference between the measured deviation of 6% and the simulation of 5%), and if there is a difference, it fine-tunes the parameters through the feedback control algorithm; it repeats the "correction-verification" process until the uniformity of the actual current distribution meets the standard, and finally outputs a uniform current distribution control scheme that meets the requirements.

[0134] The dual-pulse high-frequency switching power supply control system for precision electroplating provided in this embodiment can improve the uniformity of current distribution and the stability of plating quality.

[0135] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various simple deductions or substitutions can be made without departing from the concept of the present invention, and all such modifications and substitutions should be considered within the scope of protection of the present invention.

Claims

1. A double pulse high frequency switching power supply control method for precision electroplating, characterized in that, Executed by a computer, including: By acquiring surface data of complex workpieces using a 3D scanner, a 3D model containing local geometric changes is obtained. Based on the three-dimensional model, the high curvature region is extracted, and the high curvature region is meshed by finite element analysis to obtain the high curvature region mesh. Based on the high curvature region mesh, the surface roughness index is calculated using the local radius of curvature. The surface roughness index was simulated using the finite element method, and the current density vector distribution under the initial pulse parameters was calculated by applying a boundary voltage. When the deviation of the current density vector distribution exceeds a preset deviation threshold, the pulse parameters are optimized using a genetic algorithm to obtain the optimized pulse parameters. Select the pulse parameter that best matches the surface roughness index from the optimized pulse parameters. Based on the best-matching pulse parameter, use electromagnetic field simulation software to iteratively calculate the current density distribution to obtain the updated current density vector distribution. Based on the difference between the updated current density vector distribution and the initial distribution, a pre-established linear regression model is used to predict the coating thickness value. Based on the difference between the predicted thickness value and the preset thickness threshold, a preliminary evaluation index for quality stability is determined. Based on the preliminary evaluation indicators, the pulse parameters were optimized using the variability rate adjustment method to obtain the final control scheme with uniform current distribution.

2. The double pulse high frequency switching power supply control method for precision electroplating according to claim 1, characterized in that, The high curvature region is extracted based on the three-dimensional model, and the high curvature region is meshed using the finite element analysis method to obtain the high curvature region mesh, including: A preliminary set of feature points is obtained by performing a preliminary scan of the surface of the three-dimensional model; Based on the preliminary set of feature points, a curvature analysis algorithm based on point cloud data is used to calculate the local curvature value. The region whose local curvature value is higher than the preset curvature threshold is marked as a high curvature region, thus obtaining the high curvature region; The high curvature region was meshed using the finite element method to obtain the high curvature region mesh.

3. The double pulse high frequency switching power supply control method for precision electroplating according to claim 1, characterized in that, The process of simulating the current field of the surface roughness index using finite element analysis, and calculating the current density vector distribution under the initial pulse parameters by applying a boundary voltage, includes: Based on the surface roughness index, a model is established using the finite element analysis method to obtain the surface roughness model; Based on the surface roughness model, the voltage conditions are set using the voltage application method to obtain the electric field distribution under the initial pulse value and determine the input parameters for the current field simulation. Based on the input parameters of the current field simulation, the current density vector distribution is obtained by calculation using the finite element analysis method.

4. The double pulse high frequency switching power supply control method for precision electroplating according to claim 1, characterized in that, When the deviation of the current density vector distribution exceeds a preset deviation threshold, the pulse parameters are optimized using a genetic algorithm to obtain optimized pulse parameters, including: Based on the current density vector distribution, the distribution deviation of the current density vector is obtained by comparing local and global values; When the distribution deviation of the current density vector exceeds a preset deviation threshold, a genetic algorithm is used for optimization to obtain optimized pulse parameters.

5. The double pulse high frequency switching power supply control method for precision electroplating according to claim 1, characterized in that, The step involves selecting the pulse parameter that best matches the surface roughness index from the optimized pulse parameters, and then using electromagnetic field simulation software to iteratively calculate the current density distribution based on the best-matching pulse parameter to obtain the updated current density vector distribution, including: The matching degree between the optimized pulse parameters and the surface roughness index is calculated using a weighted scoring algorithm, and the pulse parameter with the highest matching degree is taken as the best matching pulse parameter. Based on the best-matching pulse parameters, the current density distribution is iteratively calculated using electromagnetic field simulation software to obtain a preliminary updated current density vector distribution. If the initially updated current density vector distribution deviates from the preset uniformity standard, the pulse parameters are re-optimized using a genetic algorithm, the matching degree is calculated, and the current density distribution is iteratively calculated using electromagnetic field simulation software to obtain the updated current density vector distribution.

6. The double pulse high frequency switching power supply control method for precision electroplating according to claim 1, characterized in that, Based on the difference between the updated current density vector distribution and the initial distribution, a pre-established linear regression model is used to predict the coating thickness. Based on the difference between the predicted thickness and a preset thickness threshold, preliminary evaluation indicators for quality stability are determined, including: The initial current density vector distribution data is obtained through numerical simulation software. Based on the initial distribution, the difference matrix between the updated current density vector distribution and the initial distribution is calculated using the matrix difference algorithm. Based on the difference matrix, a pre-established regression model is used to predict the change in coating thickness, thus obtaining the predicted coating thickness. The deviation between the predicted coating thickness and the preset thickness threshold is calculated, and a preliminary evaluation index for quality stability is determined based on the deviation value.

7. The dual-pulse high-frequency switching power supply control method for precision electroplating according to claim 1, characterized in that, The step of optimizing the pulse parameters using a variability rate adjustment method based on the preliminary evaluation indicators to obtain a final uniform current distribution control scheme includes: Based on the preliminary evaluation indicators, the electric field distribution was recalculated using the finite element analysis method to obtain preliminary current distribution simulation results; Based on the preliminary current distribution simulation results, the pulse parameters are initially corrected using the variability rate adjustment method to obtain the adjusted parameter set. If the adjusted parameter set does not meet the termination condition, the simulation and parameter adjustment are repeated using the finite element analysis method to obtain a new optimized parameter set. If the optimized pulse parameter data meets the termination condition, then the final uniform current distribution control scheme is generated based on the current parameter set.

8. The dual-pulse high-frequency switching power supply control method for precision electroplating according to claim 1, characterized in that, After optimizing the pulse parameters using the variability rate adjustment method based on the preliminary evaluation indicators to obtain the final uniform current distribution control scheme, the method further includes: Based on the control scheme of the final uniform current distribution, update the surface data in the three-dimensional model to obtain the optimized surface representation model; Based on the surface representation model, the current distribution is calculated using the finite element analysis method to obtain simulated current density data; Based on the simulated current density data, the current density error is obtained by comparing it with the actual measured value; If the current density error is higher than the preset error threshold, the surface representation parameters are adjusted to obtain the adjusted surface representation model. Based on the adjusted surface representation model, the predicted coating thickness is calculated using the electrochemical deposition formula: Where h is the coating thickness, η is the electrodeposition efficiency, J is the current density, t is the deposition time, M is the molar mass of the metal, n is the number of electrons transferred, F is the Faraday constant, and ρ is the density of the metal. The stability of the predicted coating thickness and the actual measured thickness are verified by statistical analysis to obtain an accuracy index.

9. A dual-pulse high-frequency switching power supply control system for precision electroplating, characterized in that, include: The model building module is used to acquire surface data of complex workpieces using a 3D scanner to obtain a 3D model that includes local geometric changes. The mesh generation module is used to extract high curvature regions from the three-dimensional model, and to refine the mesh of the high curvature regions using the finite element analysis method to obtain the high curvature region mesh. The index calculation module is used to calculate the surface roughness index based on the high curvature region mesh using the local radius of curvature. The current simulation module is used to simulate the current field of the surface roughness index using the finite element analysis method, and calculates the current density vector distribution under the initial pulse parameters by applying the boundary voltage. The optimization module is used to optimize the pulse parameters using a genetic algorithm when the deviation of the current density vector distribution exceeds a preset deviation threshold, thereby obtaining the optimized pulse parameters. The matching update module is used to select the pulse parameter that best matches the surface roughness index from the optimized pulse parameters, and based on the best matching pulse parameter, iteratively calculate the current density distribution using electromagnetic field simulation software to obtain the updated current density vector distribution. The prediction module is used to predict the coating thickness value based on the difference between the updated current density vector distribution and the initial distribution using a pre-established linear regression model, and to determine the preliminary evaluation index of quality stability based on the difference between the predicted thickness value and the preset thickness threshold. The control module is adjusted to optimize the pulse parameters using a variability adjustment method based on the preliminary evaluation indicators, thereby obtaining a control scheme for a final uniform current distribution.

Images