Crystal polishing control method and system

Abstract

The invention discloses a crystal polishing control method and system, and belongs to the technical field of crystal processing control, and the method specifically comprises the steps: obtaining the three-dimensional contour data of the surface of a to-be-polished crystal, constructing a regional characteristic spectrum, generating a regional response matrix based on the regional characteristic spectrum, and obtaining the three-dimensional contour data of the surface of the to-be-polished crystal; the region characteristic spectrum is constructed based on a region grading marking result of the surface of the to-be-polished crystal, a polishing path of the to-be-polished crystal is generated based on a region response matrix, a region self-adaptive multi-parameter cooperative control model is constructed, and polishing parameters in the polishing path during polishing are dynamically adjusted; according to the method, the processing precision, the edge consistency and the defect removal efficiency are remarkably improved, the problems that in the traditional polishing process, the edge processing capacity is insufficient, and the response of a defect area is delayed are effectively solved, and the overall processing quality of the crystal surface is remarkably improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of crystal processing control, in particular to a crystal polishing control method and system. Background Art

[0002] Crystal materials are widely used in high-end manufacturing fields such as optoelectronics, microelectronics, and precision optics. During their preparation, materials such as silicon crystals, sapphire, silicon carbide, and gallium nitride typically undergo multiple rounds of ultra-precision polishing to achieve a highly flat, low-roughness, and microcrack-free surface. The stability and control accuracy of crystal polishing technology are directly related to subsequent device performance and service life.

[0003] Existing crystal polishing methods primarily rely on path planning and static parameter setting based on uniform surface distribution across the entire crystal surface. This involves applying a uniform polishing trajectory pattern and fixed process parameters (such as feed rate, polishing pressure, and trajectory frequency) to the entire crystal surface. This approach is effective for crystals with relatively uniform overall surface morphology and a relatively low distribution of defects. However, it has drawbacks when dealing with crystal edges and surface defects such as pits, scratches, and microcracks. Specifically, this approach includes insufficient edge processing capabilities, poor responsiveness to defective areas, and a lack of dynamic control.

[0004] Existing technologies have proposed the introduction of image recognition or auxiliary adjustment methods 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 and other links. It still cannot effectively improve the processing capabilities of crystal edges and defect areas. Summary of the Invention

[0005] In view of the deficiencies in the prior art, the present invention proposes a crystal polishing control method and system.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A crystal polishing control method comprising:

[0008] Acquiring three-dimensional profile data of the crystal surface to be polished;

[0009] Constructing a regional characteristic map, and generating a regional response matrix based on the regional characteristic map, wherein the regional characteristic map is constructed based on the regional graded labeling results of the surface of the crystal to be polished;

[0010] generating a polishing path of the crystal to be polished based on the regional response matrix;

[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 constructing of the regional feature map and generating the regional response matrix based on the regional feature map includes:

[0013] Identifying edge areas and defective areas on the surface of the crystal to be polished, and marking the edge areas and defective areas by grade;

[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 areas and defect areas are numbered according to their spatial distribution and attribute types, and assigned unique area index numbers;

[0018] Extract characteristic parameters of edge and defect areas, including spatial location information, area size and boundary contour, local height, curvature distribution, normal vector change rate, grayscale gradient, and defect morphology parameters;

[0019] The extracted features are combined into a regional feature vector in a preset order, wherein the regional feature vector is a 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 a regional response matrix based on the regional feature map includes:

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

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

[0024] All characteristic response values ​​are weighted and summed to obtain the comprehensive response value of each area;

[0025] Arrange the comprehensive response values ​​of the edge area and defect area according to spatial position and index number to construct the initial regional response matrix;

[0026] The initial regional 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 a final regional response matrix.

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

[0028] Based on the regional 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 areas based on the path weights;

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

[0031] Specifically, based on the regional 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, including:

[0032] Dividing the edge area and defect area of ​​the crystal surface to be polished into equidistant two-dimensional grid cells;

[0033] Map each response value in the regional response matrix to the corresponding two-dimensional grid unit to obtain a two-dimensional response distribution map;

[0034] Generate a continuous spatial response function in the sparse or boundary jump area of ​​the two-dimensional response distribution map;

[0035] Performing spatial gradient calculation on the continuous spatial response function to obtain a response gradient vector field, wherein the gradient vector of each point in the response gradient vector field represents the direction and rate of change of the response intensity at the position of the point;

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

[0037] Specifically, constructing the path weight and selecting the activation area based on the path weight includes:

[0038] Defining weights for any position points on the surface of the crystal to be polished based on the path guidance vector field;

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

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

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

[0042] Dividing the crystal to be polished into a plurality of regional control units along the polishing path, wherein a corresponding relationship is established between the regional control units, the path weights and the regional response matrix;

[0043] Establishing a regional adaptive multi-parameter collaborative control model for each regional control unit, wherein the regional adaptive multi-parameter collaborative control model is used to calculate polishing parameters;

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

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

[0046] A crystal polishing control system, used to implement 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 three-dimensional profile data of the surface of the crystal 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 regional response matrix;

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

[0051] Specifically, the path generation module includes: a grid structure analysis unit, an activation area selection unit and a path generation unit;

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

[0053] The activation area selection unit is used to construct a path weight function and select an activation area;

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

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] This paper proposes a crystal polishing control method and system. Based on multidimensional identification of edge and defect regions on the crystal surface, this system 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. This method significantly improves processing accuracy, edge consistency, and defect removal efficiency compared to existing technologies. It effectively addresses the issues of insufficient edge processing capability and delayed response in defect regions during traditional polishing, significantly improving the overall processing quality of the crystal surface. BRIEF DESCRIPTION OF THE DRAWINGS

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

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

[0059] Figure 3 A flow chart of the surface treatment of the crystal to be polished provided by the present invention;

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

[0061] Figure 5 This is an architectural diagram of a crystal polishing control system provided by the present invention. DETAILED DESCRIPTION

[0062] The present application is described in detail below with reference to specific embodiments. The following embodiments will help those skilled in the art to further understand the present application, but are not intended to limit the present application in any form. It should be noted that those skilled in the art may make several variations and improvements without departing from the scope of the present application. These all fall within the scope of protection of the present application.

[0063] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0064] It should be noted that, if there is no conflict, the various features in the embodiments of the present application can be combined with each other and are all within the scope of protection of the present application. In addition, although the functional modules are divided in the device schematic and the logical order is shown in the flow chart, in some cases, the steps shown or described can be performed in a different order than the module division in the device or the order in the flow chart. In addition, the words "first", "second", "third", etc. used in this application do not limit the data and execution order, but only distinguish between the same items or similar items with basically the same functions and effects.

[0065] Unless otherwise defined, all technical and scientific terms used in this specification have the same meanings as those commonly understood by those skilled in the art to which this application belongs. The terms used in this specification and in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application. The term "and / or" as used in this specification includes any and all combinations of one or more of the relevant listed items.

[0066] Example 1:

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

[0068] Step S1: Acquire three-dimensional profile 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 surface of the crystal to be polished to obtain three-dimensional contour data, i.e., three-dimensional point cloud data of the polished crystal surface; a denoising algorithm is applied to the collected raw point cloud data to remove environmental noise and measurement errors.

[0070] Step S2: constructing a regional feature map, and generating a regional response matrix based on the regional feature map.

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

[0072] Step S201: identifying edge areas and defective areas on the surface of a crystal to be polished, and marking the edge areas and defective areas in a graded manner.

[0073] Specifically, the curvature, normal vector gradient change rate and boundary density of each point in the three-dimensional contour data are calculated to identify the edge area of ​​the crystal surface. Preferably, a curvature mutation extraction algorithm based on the region growing method is used to locate the edge ridges and chamfer areas; the three-dimensional contour data is projected two-dimensionally and converted into a grayscale image, and the surface defect features such as scratches, pits, and microcracks are extracted by combining the image gradient enhancement method and the edge detection algorithm. Furthermore, the defect morphology is automatically classified and located using a convolutional neural network model; the identified various areas are divided into multiple levels, such as mild, moderate, and severe, according to the defect size, shape, depth and position parameters, and are marked with corresponding labels.

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

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

[0076] like Figure 2 , which is a schematic diagram of the structure of the crystal surface to be polished, including an edge area and multiple defect areas, wherein the defect areas include microcrack defect areas, scratch defect areas and pit defect areas.

[0077] The specific steps of step S202 are:

[0078] Step S2021: The identified edge regions and defect regions are numbered according to their spatial distribution and attribute types, and are assigned unique region index numbers.

[0079] In this embodiment, Figure 2 The edge areas and defect areas identified in the

[15] are numbered according to their spatial positions and attribute types. Specifically, the regional index number of the edge area is E001, the regional index number of the microcrack defect area is D001, the regional index number of the scratch defect area is D002, and the regional index number of the pit defect area is D003. Each regional index number is a unique identifier.

[0080] Step S2022: extracting characteristic parameters of the edge area and defect area, including: spatial position information, area size and boundary contour, local height, curvature distribution, normal vector change rate, grayscale gradient and defect morphology parameters.

[0081] It should be noted that the feature parameters of the edge area and the defect area are extracted using the existing technology.

[0082] Step S2023: The extracted features are combined into a regional feature vector in a preset order, where 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 sequence according to a preset order to finally form a regional feature vector.

[0084] Step S2024: Arrange the regional feature vectors of the edge region and the defect region according to the regional 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 structure and property distribution characteristics of various regions on the crystal surface.

[0085] This step organizes multidimensional features such as space, geometry, and optics into a unified structure by constructing a fixed-length floating-point vector. This structure is used for the subsequent numerical calculation of the regional response matrix.

[0086] Step S203: generating 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 for the polishing response of each region through the constructed regional feature map, and finally a regional response matrix is ​​formed.

[0088] The specific steps of step S203 are:

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

[0090] In this embodiment, mapping functions are preset for the influence of various features in the regional feature map. For example, in the response mapping of area attributes, a monotonically increasing function is used to ensure that small-area defects obtain moderate polishing response requirement values, while large-area areas obtain higher polishing response requirement values; in the response mapping of the normal vector disturbance degree, the mapping function uses the normal vector change rate as a direct influencing factor, and the larger the value, the higher the response requirement value.

[0091] All mapping functions are normalized to convert different feature dimensions into a unified 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 level, power device level).

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

[0093] In this embodiment, each region in the regional feature map (such as Figure 2The characteristic response values ​​of the microcracks, scratches, pits and edges shown in the figure are calculated respectively.

[0094] During the actual crystal polishing process, different types of areas (such as microcracks and scratches) have significant differences in processing behavior. Microcracks can lead to the risk of device structure fracture, while scratches may affect the surface finish, and pits 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 convert the area, curvature, normal perturbation, grayscale gradient, depth and other features of each region; each feature is converted into a feature response value between 0 and 1;

[0096] Step S2033: Perform weighted summation on all characteristic response values ​​to obtain a comprehensive response value for each region.

[0097] Step S2034: Arrange the comprehensive response values ​​of the edge area and the defect area according to the spatial position and index number to construct an initial regional response matrix.

[0098] Step S2035: normalizing 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.

[0099] In this embodiment, due to the complex sources of response values, such as different dimensions such as area, curvature, and depth, the values ​​in the initially generated response matrix have problems such as order of magnitude differences and inconsistent discreteness. If used directly for path planning, it will cause path deviation or redundancy. Through 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 areas, regardless of the source of the original eigenvalues. In addition, the edge areas of the crystal are often prone to residual polishing due to problems such as geometric mutations and processing edge effects, but their initial response values ​​may not be high, and the polishing priority is improved through enhancement.

[0100] Step S3: generating a polishing path for the crystal to be polished based on the regional response matrix.

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

[0102] Step S301: constructing a two-dimensional grid structure on the surface of the crystal to be polished based on the regional response matrix, and analyzing the response value corresponding to each cell.

[0103] The specific steps of step S301 are:

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

[0105] like Figure 3 As shown, two-dimensional grid division is performed on the identified edge area and defect area. Figure 3 The two-dimensional mesh element of the microcrack defect area is only an example. The edge area and other defect areas are divided into equidistant two-dimensional mesh elements using the same method.

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

[0107] In this embodiment, each response value in the regional response matrix is ​​bound to a specific defect area or edge area. These areas cover one or more grid cells in the grid. The response value of the area is distributed to all the grid cells it covers. If a grid cell partially overlaps with multiple response areas, mapping is performed using a weighted average method. Ultimately, all grid cells have specific response intensity values, forming a two-dimensional response distribution map of the edge area and defect area.

[0108] Step S3013: generating a continuous spatial response function in a sparse or boundary jump region in the two-dimensional response distribution map.

[0109] In this embodiment, a global traversal is performed on the two-dimensional response distribution map to identify voids or mutation areas, interpolation is used to estimate the response values ​​of missing grids, and a spatial smoothing strategy is used for areas with sharp edge changes. For example, the average or trend changes of multiple surrounding grids are considered, and a response transition is established to output a continuous spatial response function that can give response values ​​at all coordinate points, that is, the response intensity value at each position of the crystal surface edge and defect area.

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

[0111] Specifically, identify Figure 3 The response of the middle pit is high, but the response of its surrounding is zero or blank. The adjacent grids are expanded and assigned medium-high response values ​​according to the proximity weight. For the fault area between the edge band and the inner circle high response, a buffer zone is constructed based on trend difference compensation to form a complete response transition. After processing, a response map with no breakpoints, no jumps, and continuous transition on the edge and defect area of ​​the crystal surface is obtained.

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

[0113] Step S3014: performing spatial gradient calculation on the continuous spatial response function to obtain a response gradient vector field, wherein the gradient vector of each point in the response gradient vector field represents the direction and rate of change of the response intensity at the position of the point.

[0114] In this embodiment, each grid point on the crystal surface is traversed, and its surrounding response values ​​are compared to determine the direction in which the response value rises fastest. A direction arrow or direction vector is obtained to indicate the direction in which the response rises at the point. At the same time, the speed of the 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: constructing a path guidance vector field based on the response gradient vector field and in combination with the preset polishing direction constraint, wherein the path guidance vector field is used to automatically guide the trajectory direction adjustment.

[0116] In this embodiment, allowable polishing direction constraints are set, for example, priority is given to the tangential direction, circular, sharp turns are prohibited, and the feed angle or curvature is limited. The response gradient vector field is fused with the direction constraint, and its direction and strength are adjusted so that the high response area is still guided first, the path complies with the equipment motion rules, the local path is smooth, and the global path is continuous. Finally, each point has a recommended direction, and all points are integrated to obtain the path guidance vector field.

[0117] Step S302: constructing path weights and selecting activation areas based on the path weights.

[0118] The specific steps of step S302 are:

[0119] Step S3021: defining weights for any position points on the surface of the crystal to be polished based on the path guidance vector field.

[0120] Specifically, for each grid point, the direction, strength and local vector change rate of its path guidance vector are read, and the comprehensive path priority weight is calculated based on whether the point is in a high response area and whether it is close to the defect boundary or edge. The comprehensive path priority weight is mapped to between 0 and 1 as the weight of any position point.

[0121] For example, in Figure 3 In the middle, the response of the concave area is high, and the guidance vectors are concentrated towards the center. The weight of this area is defined as 0.9-1. The small area defects (such as slight scratches) have moderate response values, but the guidance vectors vary drastically locally, so a medium-to-high weight of 0.6-0.7 is given. The specific weight setting is obtained by technicians in this field through a large number of experimental simulations or according to actual scene conditions.

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

[0123] Specifically, according to the equipment capabilities and process requirements, an initial weight threshold is set, the entire crystal surface grid is traversed, each weight value is judged, the 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 area.

[0124] For example, Figure 3 As shown, the weight threshold is set to 0.6, and the weight value of the shaded part in the D001 area is 0.65, which exceeds the threshold and forms a narrow and long activation area; the weight of the pit area is 0.85, which exceeds the threshold and forms an elliptical activation area; after the edge response of the wafer edge band is enhanced, its weight is 0.7, which exceeds the threshold and forms a circle of ring-shaped activation area. It should be noted that the pit area and the edge area are not marked in the figure. Referring to the example figure of the D001 area, a schematic diagram of the activation area of ​​the pit area and the edge area can be obtained.

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

[0126] For example, Figure 3 As shown, the four identified activation areas are numbered A, B, C and D, respectively, where A is the edge area with a ring-shaped distribution, B is the pit area with an approximately elliptical distribution, C and D are the scratch and crack areas with a narrow and long distribution, A is classified as the edge area, B is classified as the regular depression type, 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, and those skilled in the art can refer to the description of the A, B, C and D areas to know that Figure 3 The specific location in.

[0128] Step S303: selecting one or more combinations of path modes according to the analysis results of the shape, boundary and response value of the activation area to obtain a polishing path for the crystal to be polished, wherein the path modes include a first path mode, a second path mode and a third path mode.

[0129] In this embodiment, the first path mode can be a spiral or contour line surrounding path, which is suitable for regular areas. The second path mode can be a straight path, which is suitable for long strips and linear crack areas. The third path mode can be a local spiral + boundary fitting mixed path, which is suitable for edges or irregular areas. According to the grouping and classification of the 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 multi-region, multi-type mixed path shown.

[0130] Step S4: constructing a regional 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:

[0132] Step S401: Divide the crystal to be polished into a plurality of regional control units along the polishing path, and establish a corresponding relationship between the regional control units, the path weights, and the regional response matrix.

[0133] Specifically, the generated polishing path is spatially cut, and each segment is defined as a control unit according to its activation area, path weight, and local response value.

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

[0135] Specifically, a separate model is established for each control unit. The inputs include: response value, gradient value, trajectory direction, and boundary type. The model outputs a set of preset parameter recommendations: such as polishing head pressure, speed, polishing disk speed, fluid flow, etc., namely polishing parameters.

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

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

[0138] For example, Figure 3 As shown, in the edge area A, the response value is medium but the boundary risk is high. The model recommends parameters: low pressure, low speed, high rotation speed, and the optimization goal is to minimize the edge error; in the concave area B, the response value is high. The model recommends parameters: medium pressure, low speed, medium rotation speed, and high fluid flow rate. The optimization goal is to have high precision and minimize surface roughness.

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

[0140] Example 2:

[0141] See also Figure 5 , another embodiment provided by the present invention: 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 three-dimensional profile data of the surface of the crystal 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 regional response matrix;

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

[0146] The path generation module includes: a grid structure analysis unit, an activation area selection unit and a path generation unit;

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

[0148] The activation area selection unit is used to construct a path weight function and select an activation area;

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

[0150] In addition, the parts of the above technical solutions provided in the embodiments of the present application that are consistent with the implementation principles of the corresponding technical solutions in the prior art are not described in detail to avoid excessive redundancy.

[0151] The above-described specific embodiments further illustrate the objectives, technical solutions, and beneficial effects of the present invention. It should be understood that the above description is merely a specific embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A method for controlling crystal polishing, characterized in that: include: Acquiring three-dimensional profile data of the crystal surface to be polished; Constructing a regional characteristic map, and generating a regional response matrix based on the regional characteristic map, wherein the regional characteristic map is constructed based on the regional graded labeling results of the surface of the crystal to be polished; generating a polishing path of the crystal to be polished based on the regional response matrix; A regional adaptive multi-parameter collaborative control model is constructed to dynamically adjust the polishing parameters in the polishing path during polishing.

2. A method for controlling crystal polishing according to claim 1, characterized in that: The constructing of the regional feature map and generating the regional response matrix based on the regional feature map includes: Identifying edge areas and defective areas on the surface of the crystal to be polished, and marking the edge areas and defective areas by grade; Based on the hierarchical labeling results, a regional feature map is constructed; A regional response matrix is ​​generated based on the regional feature map.

3. A crystal polishing control method according to claim 2, characterized in that: The step of constructing a regional feature map based on the hierarchical labeling results includes: The identified edge areas and defect areas are numbered according to their spatial distribution and attribute types, and assigned unique area index numbers; Extract characteristic parameters of edge and defect areas, including spatial location information, area size and boundary contour, local height, curvature distribution, normal vector change rate, grayscale gradient, and defect morphology parameters; The extracted features are combined into a regional feature vector in a preset order, wherein the regional feature vector is a fixed-length floating-point data; 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.

4. A method for controlling crystal polishing according to claim 3, characterized in that: Generating a regional response matrix based on the regional feature map includes: Based on various feature dimensions in the regional feature map, a mapping function is preset, wherein the mapping function is used to convert the geometric or image attributes of each region into a polishing response requirement value; For each region in the regional feature map, a preset mapping function is used to calculate the feature response value; All characteristic response values ​​are weighted and summed to obtain the comprehensive response value of each area; Arrange the comprehensive response values ​​of the edge area and defect area according to spatial position and index number to construct the initial regional response matrix; The initial regional 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 a final regional response matrix.

5. A crystal polishing control method according to claim 4, characterized in that: The step of generating a polishing path of the crystal to be polished based on the regional response matrix includes: Based on the regional 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; Construct path weights and select activation areas based on the path weights; According to the analysis results of the shape, boundary and response value of the activation area, one or more combinations of path modes are selected to obtain the polishing path of the crystal to be polished. The path modes include a first path mode, a second path mode and a third path mode.

6. A method for controlling crystal polishing according to claim 5, characterized in that: The method constructs a two-dimensional grid structure on the surface of the crystal to be polished based on the regional response matrix and analyzes the response value corresponding to each cell, including: Dividing the edge area and defect area of ​​the crystal surface to be polished into equidistant two-dimensional grid cells; Map each response value in the regional response matrix to the corresponding two-dimensional grid unit to obtain a two-dimensional response distribution map; Generate a continuous spatial response function in the sparse or boundary jump area of ​​the two-dimensional response distribution map; Performing spatial gradient calculation on the continuous spatial response function to obtain a response gradient vector field, wherein the gradient vector of each point in the response gradient vector field represents the direction and rate of change of the response intensity at the position of the point; Based on the response gradient vector field and combined with the preset polishing direction constraint, a path guidance vector field is constructed, and the path guidance vector field is used to automatically guide the trajectory direction adjustment.

7. A crystal polishing control method according to claim 6, characterized in that: The step of constructing path weights and selecting activation areas based on the path weights includes: Defining weights for any position points on the surface of the crystal to be polished based on the path guidance vector field; Set a weight threshold and select continuous areas with weights greater than the weight threshold as activation areas; All activated regions are numbered and grouped and classified according to the features of the corresponding activated regions in the regional feature map.

8. A crystal polishing control method according to claim 7, characterized in that: The construction of the regional adaptive multi-parameter collaborative control model to dynamically adjust the polishing parameters in the polishing path during polishing includes: Dividing the crystal to be polished into a plurality of regional control units along the polishing path, wherein a corresponding relationship is established between the regional control units, the path weights and the regional response matrix; Establishing a regional adaptive multi-parameter collaborative control model for each regional control unit, wherein the regional adaptive multi-parameter collaborative control model is used to calculate polishing parameters; Establish a multi-objective optimization model of the control relationship between parameters and find the optimal polishing parameters of the multi-objective optimization model; The crystal to be polished is polished based on the optimal polishing parameters and polishing path.

9. A crystal polishing control system, used to implement a crystal polishing control method according to any one of claims 1 to 8, characterized in that: include: Data acquisition module, area response module, path generation module and polishing control module; The data acquisition module is used to acquire three-dimensional profile data of the surface of the crystal to be polished; The regional response module is used to construct a regional feature map and generate a regional response matrix based on the regional feature map; The path generation module is used to generate a polishing path for the crystal to be polished based on the regional response matrix; The polishing control module is used to build a regional adaptive multi-parameter collaborative control model to dynamically adjust the polishing parameters in the polishing path during polishing.

10. A crystal polishing control system according to claim 9, characterized in that: The path generation module includes: a grid structure analysis unit, an activation area selection unit and a path generation unit; The grid structure analysis unit constructs a two-dimensional grid structure on the surface of the crystal to be polished based on the regional response matrix and analyzes the response value corresponding to each cell; The activation area selection unit is used to construct a path weight function and select an activation area; The path generation unit selects a path mode based on the selected activation area and generates a polishing path.

Images