A method and system for controlling the polishing of crystals

Abstract

This invention discloses a polishing control method and system for crystals, belonging to the field of crystal processing control technology. The method specifically includes: acquiring three-dimensional contour data of the crystal surface to be polished; constructing a regional feature map; generating a regional response matrix based on the regional feature map, wherein the regional feature map is constructed based on the regional hierarchical labeling results of the crystal surface to be polished; generating a polishing path for the crystal based on the regional response matrix; constructing a regional adaptive multi-parameter collaborative control model; and dynamically adjusting the polishing parameters in the polishing path during polishing. This application significantly improves processing accuracy, edge consistency, and defect removal efficiency, effectively solving the problems of insufficient edge processing capability and delayed response in defect areas during traditional polishing processes, and significantly improving the overall processing quality of the crystal surface.

Description

Technical Field

[0001] This invention belongs to the field of crystal processing control technology, specifically a crystal polishing control method and system. Background Technology

[0002] Crystal materials are widely used in high-end manufacturing fields such as optoelectronics, microelectronics, and precision optics. Materials such as silicon crystals, sapphire, silicon carbide, and gallium nitride typically require multiple rounds of ultra-precision polishing during their fabrication to achieve a surface with high flatness, low roughness, and no microcracks. The stability and control precision of crystal polishing technology directly affect the performance and service life of subsequent devices.

[0003] Existing crystal polishing methods mainly rely on path planning and static parameter settings based on uniform distribution across the entire surface. This involves using a uniform polishing trajectory pattern and fixed process parameters (such as feed rate, polishing pressure, and trajectory frequency) across the entire crystal surface. While this method is effective for crystals with relatively uniform surface morphology and few defects, it falls short when dealing with crystal edge regions and surface defect areas such as pits, scratches, and microcracks. Specifically, it suffers from insufficient edge processing capability, poor response capability in defect areas, and a lack of dynamic control methods.

[0004] Existing technologies have proposed auxiliary adjustment methods that introduce image recognition or are based on surface roughness feedback, but their control accuracy is limited, the defect recognition dimension is single, and it is difficult to achieve deep collaborative control in path planning, parameter scheduling, feedback correction, etc., so they still cannot effectively improve the processing capability of crystal edges and defect areas. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention proposes a method and system for controlling the polishing of crystals.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A method for controlling the polishing of a crystal, comprising:

[0008] Obtain the three-dimensional contour data of the crystal surface to be polished;

[0009] Construct a regional feature map, and generate a regional response matrix based on the regional feature map. The regional feature map is constructed based on the regional hierarchical labeling results of the crystal surface to be polished.

[0010] Based on the region response matrix, a polishing path is generated for the crystal to be polished.

[0011] A regional adaptive multi-parameter collaborative control model is constructed to dynamically adjust the polishing parameters in the polishing path during polishing.

[0012] Specifically, the construction of the regional feature map and the generation of the regional response matrix based on the regional feature map include:

[0013] Identify the edge regions and defect regions on the surface of the crystal to be polished, and classify and mark the edge regions and defect regions accordingly;

[0014] Based on the hierarchical labeling results, a regional feature map is constructed;

[0015] A regional response matrix is ​​generated based on the regional feature map.

[0016] Specifically, the construction of a regional feature map based on the hierarchical labeling results includes:

[0017] The identified edge regions and defect regions are numbered according to their spatial distribution and attribute type, and assigned a unique region index number;

[0018] Extract feature parameters of edge and defect regions, including: spatial location information, area size and boundary contour, local height, curvature distribution, normal vector change rate, gray-level gradient and defect morphology parameters;

[0019] The extracted features are arranged in a preset order to form a region feature vector, and the region feature vector is fixed-length floating-point data.

[0020] The regional feature vectors of the edge region and the defect region are arranged according to the region index number to construct a regional feature map.

[0021] Specifically, generating the regional response matrix based on the regional feature map includes:

[0022] Based on the various feature dimensions in the regional feature map, a preset mapping function is used to convert the geometric or image attributes of each region into polishing response requirement values.

[0023] For each region in the regional feature map, the feature response value is calculated using a preset mapping function;

[0024] The weighted sum of all characteristic response values ​​is used to obtain the comprehensive response value for each region.

[0025] The combined response values ​​of the edge region and the defect region are arranged according to spatial location and index number to construct an initial region response matrix;

[0026] The initial region response matrix is ​​normalized, and a response weight gain coefficient is applied to the edge region of the crystal surface to be polished to obtain the final region response matrix.

[0027] Specifically, generating the polishing path for the crystal to be polished based on the region response matrix includes:

[0028] Based on the region response matrix, a two-dimensional grid structure is constructed on the surface of the crystal to be polished, and the response value corresponding to each cell is analyzed.

[0029] Construct path weights and select activation regions based on the path weights;

[0030] Based on the analysis results of the shape, boundary and response value of the activated region, one or more combinations of path modes are selected to obtain the polishing path of the crystal to be polished. The path modes include the first path mode, the second path mode and the third path mode.

[0031] Specifically, the process of constructing a two-dimensional grid structure on the surface of the crystal to be polished based on the region response matrix, and analyzing the response value corresponding to each cell, includes:

[0032] The edge and defect regions of the crystal surface to be polished are divided into equidistant two-dimensional grid units.

[0033] Each response value in the region response matrix is ​​mapped to the corresponding two-dimensional grid cell to obtain a two-dimensional response distribution map;

[0034] In the sparse or boundary transition regions of the two-dimensional response distribution map, a continuous spatial response function is generated;

[0035] Spatial gradient calculation is performed on the continuous spatial response function to obtain the response gradient vector field. The gradient vector of each point in the response gradient vector field represents the direction and rate of change of the response intensity at that point.

[0036] Based on the response gradient vector field and combined with the preset polishing direction constraint, a path guidance vector field is constructed, which is used to automatically guide the trajectory direction adjustment.

[0037] Specifically, the construction of path weights and the selection of activation regions based on path weights include:

[0038] Weights are defined for arbitrary points on the surface of the crystal to be polished based on the path-guided vector field.

[0039] Set a weight threshold and select a continuous region with a weight greater than the weight threshold as the activation region;

[0040] All activated regions are numbered and grouped and classified according to the characteristics of the corresponding activated regions in the region feature map.

[0041] Specifically, the construction of the region-adaptive multi-parameter collaborative control model to dynamically adjust the polishing parameters in the polishing path during polishing includes:

[0042] The crystal to be polished is divided into multiple region control units along the polishing path, and the region control units are associated with the path weights and the region response matrix.

[0043] A regional adaptive multi-parameter collaborative control model is established for each regional control unit, and the regional adaptive multi-parameter collaborative control model is used to calculate polishing parameters;

[0044] Establish a multi-objective optimization model for the control relationship between parameters, and find the optimal polishing parameters of the multi-objective optimization model;

[0045] Polishing is performed on the crystal to be polished based on the optimal polishing parameters and polishing path.

[0046] A crystal polishing control system for implementing the crystal polishing control method, comprising: a data acquisition module, a region response module, a path generation module, and a polishing control module;

[0047] The data acquisition module is used to acquire the three-dimensional contour data of the crystal surface to be polished;

[0048] The regional response module is used to construct a regional feature map and generate a regional response matrix based on the regional feature map.

[0049] The path generation module is used to generate a polishing path for the crystal to be polished based on the region response matrix.

[0050] The polishing control module is used to construct a region-adaptive multi-parameter collaborative control model and dynamically adjust the polishing parameters in the polishing path during polishing.

[0051] Specifically, the path generation module includes: a mesh structure analysis unit, an active region selection unit, and a path generation unit;

[0052] The mesh structure analysis unit constructs a two-dimensional mesh structure on the surface of the crystal to be polished based on the region response matrix, and analyzes the response value corresponding to each cell.

[0053] The activation region selection unit is used to construct the path weight function and select the activation region;

[0054] The path generation unit selects a path mode based on the selected activation region and generates a polishing path.

[0055] Compared with the prior art, the beneficial effects of the present invention are:

[0056] This invention proposes a polishing control method and system for crystals. Based on multi-dimensional identification of edge and defect regions on the crystal surface, it constructs regional feature maps, generates a response matrix, automatically plans the polishing path, and introduces a multi-parameter collaborative control and real-time feedback mechanism to achieve differentiated processing of edge and defect regions. Compared with existing technologies, this method significantly improves processing accuracy, edge consistency, and defect removal efficiency, effectively solving the problems of insufficient edge processing capability and delayed response in defect regions during traditional polishing processes, and significantly improving the overall processing quality of the crystal surface. Attached Figure Description

[0057] Figure 1 A flowchart of a crystal polishing control method provided by the present invention;

[0058] Figure 2 This is a schematic diagram of the crystal surface to be polished provided by the present invention;

[0059] Figure 3 This invention provides a flowchart of the surface treatment process for a crystal to be polished.

[0060] Figure 4 This is a schematic diagram of the polishing path for the crystal to be polished provided by the present invention;

[0061] Figure 5 This invention provides an architecture diagram of a crystal polishing control system. Detailed Implementation

[0062] The present application will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present application, but do not limit the present application in any way. It should be noted that those skilled in the art can make several modifications and improvements without departing from the concept of the present application. These all fall within the protection scope of the present application.

[0063] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0064] It should be noted that, unless there is a conflict, the various features in the embodiments of this application can be combined with each other, all of which are within the protection scope of this application. Furthermore, although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than the module division in the device or the order in the flowchart. In addition, the terms "first," "second," and "third" used in this application do not limit the data or execution order, but only distinguish identical or similar items with essentially the same function and effect.

[0065] Unless otherwise defined, all technical and scientific terms used in this specification have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of this application. The term "and / or" as used in this specification includes any and all combinations of one or more of the associated listed items.

[0066] Example 1:

[0067] Please see Figures 1-4 The present invention provides an embodiment of a crystal polishing control method, comprising the following specific steps:

[0068] Step S1: Obtain the three-dimensional contour data of the crystal surface to be polished.

[0069] In this embodiment, a laser confocal scanner, a white light interferometer, or a structured light scanning system is used to collect point cloud data of the crystal surface to be polished, thereby obtaining three-dimensional contour data, i.e., three-dimensional point cloud data of the polished crystal surface; for the collected raw point cloud data, a denoising algorithm is applied to remove environmental noise and measurement errors.

[0070] Step S2: Construct a regional feature map and generate a regional response matrix based on the regional feature map.

[0071] like Figure 2 As shown, the specific steps of step S2 are as follows:

[0072] Step S201: Identify the edge regions and defect regions on the surface of the crystal to be polished, and classify and mark the edge regions and defect regions.

[0073] Specifically, the curvature, normal vector gradient change rate, and boundary density of each point in the 3D contour data are calculated to identify the edge regions of the crystal surface. Preferably, a curvature mutation extraction algorithm based on region growing is used to locate edge lines and chamfer regions. The 3D contour data is then projected into a 2D image and converted into a grayscale image. Surface defect features such as scratches, pits, and microcracks are extracted by combining image gradient enhancement methods and edge detection algorithms. Furthermore, a convolutional neural network model is used to automatically classify and locate the defect morphology. The identified regions are classified into multiple levels, such as slight, moderate, and severe, based on defect size, shape, depth, and location parameters, and are labeled accordingly.

[0074] like Figure 2 As shown in the figure, the labels are edge areas and defect areas. The defect area labels include microcracks, pits, and scratches.

[0075] Step S202: Construct a regional feature map based on the hierarchical labeling results.

[0076] like Figure 2 The diagram shows the structure of the crystal surface to be polished, including the edge region and multiple defect regions, including microcrack defect regions, scratch defect regions and pit defect regions.

[0077] The specific steps of step S202 are as follows:

[0078] Step S2021: Number the identified edge regions and defect regions according to their spatial distribution and attribute type, and assign them unique region index numbers.

[0079] In this embodiment, for Figure 2 The edge regions and defect regions identified in the data are numbered according to their spatial location and attribute type. Specifically, the region index number of the edge region is E001, the region index number of the microcrack defect region is D001, the region index number of the scratch defect region is D002, and the region index number of the pit defect region is D003. Each region index number is a unique identifier.

[0080] Step S2022: Extract feature parameters of edge regions and defect regions, including: spatial location information, area size and boundary contour, local height, curvature distribution, normal vector change rate, gray-level gradient and defect morphology parameters.

[0081] It should be noted that existing technologies are used to extract feature parameters for edge and defect regions.

[0082] Step S2023: The extracted features are arranged in a preset order to form a regional feature vector, wherein the regional feature vector is fixed-length floating-point data.

[0083] In this embodiment, a fixed-length floating-point vector structure is adopted, that is, the feature information of each region is represented by a fixed-length floating-point array, and the parameters are arranged in a preset order to form the region feature vector.

[0084] Step S2024: Arrange the regional feature vectors of the edge region and the defect region according to the region index number to construct a regional feature map. The regional feature map is in the form of a two-dimensional array or a sparse matrix, which is used to represent the structural and property distribution characteristics of various regions on the crystal surface.

[0085] This step constructs a fixed-length floating-point vector to organize multi-dimensional features such as spatial, geometric, and optical dimensions into a unified structure, which is used for the subsequent numerical calculation of the regional response matrix.

[0086] Step S203: Generate a regional response matrix based on the regional feature map.

[0087] In this embodiment, reference Figure 2 The regional structure shown on the surface of the crystal to be polished is further modeled by the polishing response of each region through the already constructed regional feature map, and finally a regional response matrix is ​​formed.

[0088] The specific steps of step S203 are as follows:

[0089] Step S2031: Based on the various feature dimensions in the region feature map, a preset mapping function is established. The mapping function is used to convert the geometric or image attributes of each region into polishing response requirement values.

[0090] In this embodiment, a pre-defined mapping function is used to map the influence of various features in the regional feature map. For example, the response mapping of the area attribute uses a monotonically increasing function so that small area defects obtain a moderate polishing response requirement value, while large area regions obtain a higher polishing response requirement value. The response mapping of the normal vector perturbation degree uses the normal vector change rate as a direct influencing factor. The larger the value, the higher the response requirement value.

[0091] All mapping functions are normalized to convert different feature dimensions to a uniform response value range, such as [0,1]. In addition, it should be noted that the specific gain parameters of the mapping function need to be dynamically adjusted according to the crystal type (such as silicon, sapphire, gallium nitride, etc.) and the target application scenario (such as optical grade, power device grade).

[0092] Step S2032: For each region in the regional feature map, calculate the feature response value using a preset mapping function.

[0093] In this embodiment, by using the various mapping functions set in step S2031, each region in the regional feature map (such as...) is mapped... Figure 2The characteristic response values ​​of the microcracks, scratches, pits, and edges shown in the figure are calculated respectively.

[0094] In the actual process of crystal polishing, different types of regions (such as microcracks and scratches) exhibit significant differences in processing behavior. Microcracks can lead to the risk of device structural breakage, while scratches may affect surface finish, and pits may affect the uniformity of the film layer. These differences in structural properties are reflected in different characteristic response values.

[0095] Specifically, the mapping function is called to transform the features of each region, such as area, curvature, normal perturbation, grayscale gradient, and depth; each feature is converted into a feature response value between 0 and 1.

[0096] Step S2033: Sum all the characteristic response values ​​by weight to obtain the comprehensive response value for each region.

[0097] Step S2034: Arrange the combined response values ​​of the edge region and the defect region according to their spatial location and index number to construct the initial region response matrix.

[0098] Step S2035: Normalize the initial region response matrix and apply a response weight gain coefficient to the edge region of the crystal surface to be polished to obtain the final region response matrix.

[0099] In this embodiment, due to the complex sources of response values, such as different dimensions like area, curvature, and depth, the values ​​in the initially generated response matrix have differences in magnitude and inconsistent dispersion. If directly used for path planning, it will lead to path deviation or redundancy. By normalization and edge enhancement, the response values ​​of all regions are in the same numerical range, so that high response values ​​always represent high priority regions, regardless of the source of the original feature values. Furthermore, due to geometric abrupt changes and processing edge effects, crystal edge regions are often prone to residual polishing, but their initial response values ​​may not be high. Enhancement is used to improve the polishing priority.

[0100] Step S3: Based on the region response matrix, generate the polishing path for the crystal to be polished.

[0101] like Figure 3 and Figure 4 As shown, the specific steps of step S3 are as follows:

[0102] Step S301: Based on the region response matrix, construct a two-dimensional grid structure on the surface of the crystal to be polished, and analyze the response value corresponding to each cell.

[0103] The specific steps of step S301 are as follows:

[0104] Step S3011: Divide the edge region and defect region of the crystal surface to be polished into equidistant two-dimensional grid units.

[0105] like Figure 3 As shown, a two-dimensional mesh is created in the identified edge and defect regions. Figure 3 The two-dimensional mesh element for the microcrack defect region in the image is just an example; the edge region and other defect regions are divided into equidistant two-dimensional mesh elements using the same method.

[0106] Step S3012: Map each response value in the region response matrix to the corresponding two-dimensional grid cell to obtain a two-dimensional response distribution map.

[0107] In this embodiment, each response value in the region response matrix is ​​bound to a specific defect region or edge region. These regions cover one or more grid cells in the grid. The response value of the region is assigned to all the grid cells it covers. If a grid cell partially overlaps with multiple response regions, a weighted average method is used for mapping. Finally, all grid cells have specific response intensity values, forming a two-dimensional response distribution map on the edge region and defect region.

[0108] Step S3013: Generate a continuous spatial response function in the sparse or boundary transition regions of the two-dimensional response distribution map.

[0109] In this embodiment, the two-dimensional response distribution map is traversed globally to identify voids or abrupt changes in regions. Interpolation is used to estimate the response values ​​of missing grids. For regions with abrupt changes in edges, a spatial smoothing strategy is used, such as considering the average or trend changes of multiple surrounding grids, to establish a response transition. A continuous spatial response function that can provide response values ​​at all coordinate points is output, namely the response intensity value at each location of the crystal surface edge and defect region.

[0110] For example, such as Figure 3 As shown, a regular grid structure has been established between the crystal surface edge and the defect region, and there are curved paths. These paths are mainly concentrated in: the pit region in the middle, with high response; the crack region above, with medium response; and the edge ring region, with enhanced response after weighting. Since these response values ​​were initially assigned only to individual grid cells or their directly adjacent grids, some regions in the figure show obvious response fractures, which are the places in the grid where there are no defects.

[0111] Specifically, it was identified Figure 3 The central pit has a high response, but the response around it is zero or blank. Value expansion is performed on adjacent grids, and medium to high response values ​​are assigned to them according to proximity weight. For the discontinuity region between the edge zone and the inner circle with high response, a buffer zone is constructed based on trend difference to form a complete response transition. After processing, a response map with no discontinuities, no jumps, and a continuous transition on the crystal surface edge and defect region is obtained.

[0112] It should be noted that the sparsity here is similar to the concept of sparsity in sparse matrices.

[0113] Step S3014: Perform spatial gradient calculation on the continuous spatial response function to obtain the response gradient vector field. The gradient vector of each point in the response gradient vector field represents the direction and rate of change of the response intensity at that point.

[0114] In this embodiment, each grid point on the crystal surface is traversed, and its surrounding response values ​​are compared to determine which direction to move in, where the response value rises the fastest. A directional arrow or directional vector is then generated to represent the direction of the response rise at that point. At the same time, the rate of rise is recorded as the vector length. All points are used to form a response gradient vector field to guide path streamline planning.

[0115] Step S3015: Based on the response gradient vector field and combined with the preset polishing direction constraint, a path guidance vector field is constructed, which is used to automatically guide the trajectory direction adjustment.

[0116] In this embodiment, allowable polishing direction constraints are set, such as prioritizing tangential directions, circular paths, prohibiting sharp angle turns, and limiting feed angles or excessive curvature. The response gradient vector field is fused with the direction constraints, and its direction and intensity are adjusted so that high response areas are still preferentially guided, the path conforms to the equipment motion rules, local paths are smooth, and the global path is continuous. Finally, each point has a recommended path, and all points are integrated to obtain the path guidance vector field.

[0117] Step S302: Construct path weights and select activation regions based on path weights.

[0118] The specific steps of step S302 are as follows:

[0119] Step S3021: Define weights for arbitrary points on the surface of the crystal to be polished based on the path-guided vector field.

[0120] Specifically, for each grid point, the direction, intensity, and local vector change rate of its path guidance vector are read. Combined with whether the point is in a high response zone or close to a defect boundary or edge, the comprehensive path priority weight is calculated and mapped to a value between 0 and 1 as the weight of any point.

[0121] For example, in Figure 3 In the middle, the concave area has a high response, and the guiding vectors converge towards the center. The weight of this area is defined as 0.9-1. Small area defects (such as slight scratches) have moderate response values, but the guiding vectors change drastically locally. They are given a medium to high weight of 0.6-0.7. The specific weight settings are based on the results obtained by those skilled in the art through a large number of experimental simulations or based on the actual scenario.

[0122] Step S3022: Set a weight threshold and select a continuous region with a weight greater than the weight threshold as the activation region.

[0123] Specifically, based on equipment capabilities and process requirements, an initial weight threshold is set, the entire crystal surface grid is traversed, each weight value is judged, grid cells with weights greater than the weight threshold are marked as valid points, the valid points are clustered, and adjacent grid blocks are extracted to form an activation region.

[0124] For example, such as Figure 3 As shown, with a weight threshold of 0.6, the shaded area in region D001 has a weight of 0.65, exceeding the threshold and forming a narrow, elongated activation region; the pit area has a weight of 0.85, exceeding the threshold and forming an elliptical activation region; the wafer edge band, due to edge response enhancement, has a weight of 0.7, exceeding the threshold and forming a ring-shaped activation region. It should be noted that the pit area and edge area are not labeled in the figure. Refer to the example figure of region D001 to obtain a schematic diagram of the activation regions of the pit area and edge area.

[0125] Step S3023: Number all active regions and group and classify them according to the features of the corresponding active regions in the regional feature map.

[0126] For example, such as Figure 3 As shown, the four identified activation regions are numbered A, B, C, and D respectively. A is an edge region with a ring-shaped distribution, B is a pit region with an approximately elliptical distribution, and C and D are scratch and crack regions with a narrow distribution. A is classified as an edge region, B is classified as a regular depression region, and C and D are classified as linear defects.

[0127] It should be noted that, Figure 3 The specific numbers A, B, C, and D are not marked. Those skilled in the art can refer to the descriptions of areas A, B, C, and D to determine the meaning of the text. Figure 3 The specific location within.

[0128] Step S303: Based on the analysis results of the shape, boundary and response value of the activated region, select one or more combinations of path modes to obtain the polishing path of the crystal to be polished. The path modes include the first path mode, the second path mode and the third path mode.

[0129] In this embodiment, the first path mode can be a spiral or contour-line-style winding path, suitable for regular areas; the second path mode can be a straight path, suitable for long, narrow, or linear crack areas; and the third path mode can be a hybrid path of local spiral and boundary adhesion, suitable for edge or irregular areas. Based on the grouping and classification of activated areas in the previous step, area A selects the third path mode, area B selects the first path mode, and areas C and D select the second path mode. The three path modes are automatically combined to finally generate the following... Figure 4 The example shows a multi-regional, multi-type mixed path.

[0130] Step S4: Construct a region-adaptive multi-parameter collaborative control model to dynamically adjust the polishing parameters in the polishing path during polishing.

[0131] The specific steps of step S4 are as follows:

[0132] Step S401: Divide the crystal to be polished into multiple region control units along the polishing path, and establish a correspondence between the region control units and the path weights and region response matrices.

[0133] Specifically, the generated polishing path is spatially divided, and each segment is defined as a control unit based on its active region, path weight, and local response value.

[0134] Step S402: Establish a regional adaptive multi-parameter collaborative control model for each regional control unit. The regional adaptive multi-parameter collaborative control model is used to calculate polishing parameters.

[0135] Specifically, a separate model is built for each control unit. The inputs include response value, gradient value, trajectory direction, and boundary type. The model outputs a set of preset parameter suggestions, such as polishing head pressure, speed, polishing disc rotation speed, and fluid flow rate, i.e., polishing parameters.

[0136] Step S403: Establish a multi-objective optimization model for the control relationship between parameters, and find the optimal polishing parameters of the multi-objective optimization model.

[0137] Specifically, multiple optimization objectives are set, such as minimizing surface roughness, minimizing processing time, and minimizing edge error. Based on the region model, a multi-objective function is constructed, and a multi-objective optimization algorithm is applied to automatically search for the optimal parameter combination that satisfies different objectives.

[0138] For example, such as Figure 3 As shown, region A is the edge region, with a moderate response value but high boundary risk. The recommended parameters for the model are: low pressure, low speed, and high rotation speed. The optimization objective is to minimize the edge error. Region B is the concave region, with a high response value. The recommended parameters for the model are: medium pressure, low speed, medium rotation speed, and high fluid velocity. The optimization objective is to achieve high accuracy and minimum surface roughness.

[0139] Step S404: Polish the crystal to be polished based on the optimal polishing parameters and polishing path.

[0140] Example 2:

[0141] Please see Figure 5 Another embodiment of the present invention provides a crystal polishing control system, comprising: a data acquisition module, a region response module, a path generation module, and a polishing control module;

[0142] The data acquisition module is used to acquire the three-dimensional contour data of the crystal surface to be polished;

[0143] The regional response module is used to construct a regional feature map and generate a regional response matrix based on the regional feature map.

[0144] The path generation module is used to generate a polishing path for the crystal to be polished based on the region response matrix.

[0145] The polishing control module is used to construct a region-adaptive multi-parameter collaborative control model and dynamically adjust the polishing parameters in the polishing path during polishing.

[0146] The path generation module includes: a mesh structure analysis unit, an active region selection unit, and a path generation unit;

[0147] The mesh structure analysis unit constructs a two-dimensional mesh structure on the surface of the crystal to be polished based on the region response matrix, and analyzes the response value corresponding to each cell.

[0148] The activation region selection unit is used to construct the path weight function and select the activation region;

[0149] The path generation unit selects a path mode based on the selected activation region and generates a polishing path.

[0150] In addition, the parts of the technical solutions provided in the embodiments of this application that are consistent with the implementation principles of the corresponding technical solutions in the prior art have not been described in detail, so as to avoid excessive elaboration.

[0151] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the invention. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method of polishing control of a crystal, characterized by, The method comprises the following steps: acquiring three-dimensional profile data of a crystal surface to be polished; constructing a regional feature map, and generating a regional response matrix based on the regional feature map, wherein the regional feature map is constructed based on a regional grading marking result of the crystal surface to be polished; generating a polishing path of the crystal to be polished based on the regional response matrix; constructing a regional adaptive multi-parameter collaborative control model, and dynamically adjusting polishing parameters in the polishing path during polishing; the constructing of the regional feature map and the generating of the regional response matrix based on the regional feature map comprise the following steps: identifying an edge region and a defect region of the crystal surface to be polished, and grading marking the edge region and the defect region; constructing a regional feature map based on the grading marking result; generating a regional response matrix based on the regional feature map, wherein the regional response matrix applies a response weight gain coefficient to the edge region of the crystal surface to be polished; the constructing of the regional feature map based on the grading marking result comprises the following steps: numbering the identified edge region and defect region according to spatial distribution and attribute type, and assigning a unique regional index number; extracting feature parameters of the edge region and the defect region, including spatial position information, area size and boundary profile, local height, curvature distribution, normal vector change rate, gray gradient and defect morphology parameters; arranging the extracted features in a preset order to form a regional feature vector, wherein the regional feature vector is a fixed-length floating-point data; arranging the regional feature vectors of the edge region and the defect region according to the regional index number to construct a regional feature map.

2. A method of polishing control of a crystal as claimed in claim 1, wherein, the generating of the regional response matrix based on the regional feature map comprises the following steps: based on each feature dimension in the regional feature map, presetting a mapping function, wherein the mapping function is used to convert the geometric or image attributes of each region into a polishing response requirement value; using the preset mapping function to calculate a feature response value for each region in the regional feature map; performing weighted summation on all feature response values to obtain a comprehensive response value of each region; arranging the comprehensive response values of the edge region and the defect region according to spatial position and index number to construct an initial regional response matrix; performing normalization processing on the initial regional response matrix, and applying a response weight gain coefficient to the edge region of the crystal surface to be polished to obtain a final regional response matrix.

3. A method of polishing control of a crystal as claimed in claim 2, wherein, the generating of the polishing path of the crystal to be polished based on the regional response matrix comprises the following steps: based on the regional response matrix, constructing a two-dimensional grid structure on the surface of the crystal to be polished, and analyzing the response value corresponding to each unit cell; constructing a path weight, and selecting an active region based on the path weight; selecting one or more combinations of path modes according to the shape, boundary and response value analysis result of the active region to obtain the polishing path of the crystal to be polished, wherein the path modes include a first path mode, a second path mode and a third path mode.

4. A method of polishing control of a crystal as claimed in claim 3, wherein, the constructing of the two-dimensional grid structure on the surface of the crystal to be polished based on the regional response matrix, and the analyzing of the response value corresponding to each unit cell comprise the following steps: dividing the edge region and the defect region of the crystal surface to be polished into equidistant two-dimensional grid cells; mapping each response value in the area response matrix to a corresponding two-dimensional grid cell to obtain a two-dimensional response distribution map; generating a continuous spatial response function in a sparse or boundary jump area in the two-dimensional response distribution map; performing spatial gradient calculation on the continuous spatial response function to obtain a response gradient vector field, wherein a gradient vector of each point in the response gradient vector field represents a direction and rate of change of response intensity at the point position; based on the response gradient vector field, combining a preset polishing direction constraint to construct a path guide vector field, the path guide vector field being used to automatically guide trajectory direction adjustment.

5. A method of polishing control of a crystal as claimed in claim 4, wherein, The constructing path weight and selecting an active area based on the path weight include: defining a weight for an arbitrary position point on the crystal surface to be polished based on the path guide vector field; setting a weight threshold value, and selecting a continuous area with a weight greater than the weight threshold value as an active area; numbering all the active areas, and grouping and classifying the active areas according to the characteristics of the corresponding active areas in the area characteristic map.

6. A method of polishing control of a crystal as described in claim 5, wherein, The constructing area adaptive multi-parameter collaborative control model dynamically adjusts the polishing parameters in the polishing path during polishing, including: dividing the crystal to be polished into a plurality of area control units along the polishing path, the area control units having a corresponding relationship with the path weight and the area response matrix; establishing an area adaptive multi-parameter collaborative control model for each area control unit, the area adaptive multi-parameter collaborative control model being used to calculate polishing parameters; establishing a multi-objective optimization model of the control relationship between parameters, and solving optimal polishing parameters of the multi-objective optimization model; polishing the crystal to be polished based on the optimal polishing parameters and the polishing path.

7. A system for implementing the method of any one of claims 1-6, wherein the system comprises: a polishing machine; a controller; and a computer readable medium having stored therein a computer program comprising instructions for causing the controller to perform the method of any one of claims 1-6. including: a data acquisition module, an area response module, a path generation module, and a polishing control module; the data acquisition module is configured to acquire three-dimensional profile data of the crystal surface to be polished; the area response module is configured to construct an area characteristic map and generate an area response matrix based on the area characteristic map; the path generation module is configured to generate a polishing path of the crystal to be polished based on the area response matrix; the polishing control module is configured to construct an area adaptive multi-parameter collaborative control model and dynamically adjust polishing parameters in the polishing path during polishing.

8. A crystal polishing control system as in claim 7, wherein, The path generation module includes: a grid structure analysis unit, an active area selection unit, and a path generation unit; the grid structure analysis unit constructs a two-dimensional grid structure on the crystal surface to be polished based on the area response matrix, and analyzes the response value corresponding to each cell; the active area selection unit is configured to construct a path weight function and select an active area; the path generation unit selects a path mode based on the selected active area, and generates a polishing path.

Images