Global adaptive filling and boundary splicing mark suppression method

Abstract

The invention belongs to the technical field of femtosecond laser processing, and relates to a global adaptive filling and boundary splicing mark suppression method, which comprises the following steps of: 1) acquiring a breadth to be processed, and dividing the breadth to be processed into a plurality of sub-regions; the method comprises the steps of (1) obtaining a to-be-processed breadth, (2) generating a scanning filling path in each sub-region obtained in the step (1), and (3) optimizing the scanning filling paths of all the sub-regions in the to-be-processed breadth obtained in the step (2) to complete global adaptive filling and boundary splicing mark suppression. The global adaptive filling and boundary splicing mark suppression method is high in precision and efficiency.

Description

Global adaptive filling and boundary stitch suppression methods Technical Field

[0001] This invention belongs to the field of femtosecond laser processing technology, and relates to a global adaptive filling and boundary stitch suppression method, particularly a global adaptive filling and boundary stitch suppression method for large-format femtosecond laser etching. Background Technology

[0002] Large-format surface processing is an advanced manufacturing technology that modifies, texturizes, cleans, roughens, or strengthens the surface of workpieces typically measuring in square meters or larger. This technology, developed to meet the ever-increasing demands of modern industry for the surface performance and appearance of large components, has gradually become an important supplement and upgrade to traditional surface treatment technologies. Currently, traditional methods for large-format surface processing mainly rely on machining and conventional laser processing. While machining and conventional laser processing are widely used, they suffer from severe tool wear, low processing efficiency, and the risk of microcracks and chipping when dealing with materials with extremely high hardness and brittleness. This makes it difficult to meet the requirements of high-precision, damage-free processing, and is particularly unsuitable for the mass production of complex microstructures. Femtosecond pulsed laser processing, with its extremely short pulse width and extremely high peak power, can achieve near-cold processing material removal, offering a potential solution to these problems. However, when the laser beam performs large-area, long-path scanning etching, it is difficult to maintain a constant laser energy density and effect throughout the entire processing area, resulting in uneven etching depth and linewidth, and obvious etching partition splicing marks, directly affecting the planar processing effect. An improper laser scanning path can lead to excessive local heat concentration, increasing the risk of thermal damage. Summary of the Invention

[0003] To address the aforementioned technical problems in the background art, this invention provides a high-precision and high-efficiency global adaptive filling and boundary stitch suppression method.

[0004] To achieve the above objectives, the present invention adopts the following technical solution: a global adaptive filling and boundary stitching suppression method, characterized in that: the global adaptive filling and boundary stitching suppression method includes the following steps: 1) obtaining the surface to be processed and dividing the surface to be processed into multiple sub-regions; 2) generating a scan filling path in each sub-region obtained in step 1); 3) optimizing the scan filling paths of all sub-regions in the surface to be processed obtained in step 2) to complete the global adaptive filling and boundary stitching suppression.

[0005] The specific implementation method of step 1) above is as follows: 1.1) Randomly generate a certain number of seed points within the area to be processed; 1.2) Form an initial sub-region based on the seed points obtained in step 1.1); 1.3) Calculate the centroid for each initial sub-region generated in step 1.2); 1.4) Iteratively optimize the centroid obtained in step 1.3) to obtain the optimized sub-region; 1.5) Obtain the boundary points of the optimized sub-region obtained in step 1.4); 1.6) Perform convex hull calculation on the boundary points of the optimized sub-region obtained in step 1.5) to obtain the sub-region of the convex polygon, thus completing the division of the area to be processed.

[0006] The specific implementation of step 1.2) above is as follows: traverse each point of the surface to be processed, where the coordinates of the point are (X, Y), calculate the distance from each point of the surface to be processed to all seed points obtained in step 1.1), find the closest seed point, and mark this point as the region ID of the seed point; generate a partition map according to the convex hull calculation method, and the partition map is the initial sub-region; the specific implementation of step 1.3) is as follows: for each initial sub-region, sum up all the X coordinates belonging to this initial sub-region and calculate the average to obtain a new X coordinate X_avg; sum up all the Y coordinates and calculate the average to obtain a new Y coordinate Y_avg; this average coordinate (X_avg, Y_avg) is the centroid of the initial sub-region; the specific implementation of step 1.4) is as follows: use the centroids of all the initial sub-regions calculated in step 1.3) as the seed points of the new round, replace the seed points in step 1.1), and repeat steps 1.2) and 1.3) until the area of ​​each sub-region is uniform and the shape of each sub-region is regular, and obtain the optimized sub-region.

[0007] The specific implementation method of step 2) above is as follows: 2.1) Determine the scanning path filling direction of each sub-region based on the shape and position of the sub-region obtained in step 1); 2.2) Generate parallel scanning lines covering the entire sub-region within the sub-region based on the scanning path filling direction determined in step 2.1); 2.3) Cut off the parallel scanning lines at the boundary of two adjacent sub-regions to obtain the scanning filling path of all sub-regions in the area to be processed.

[0008] The row spacing of the parallel scan lines mentioned in step 2.2) above is S. The calculation method of S is: S=d×(1-η); where: d is the effective spot diameter of the femtosecond laser; η is the overlap rate required by the process.

[0009] The optimization in step 3) above includes path optimization within a sub-region and path optimization between adjacent sub-regions.

[0010] The specific implementation of the path optimization between adjacent sub-regions is as follows: For the straight line vertex located on the path of the adjacent sub-region, a small random displacement perturbation is applied, causing the start and end points of the scan line at the boundary of the adjacent sub-region to no longer be precisely aligned; the perturbation is carried out within a specific range perpendicular to the original scan line direction; the path optimization within the sub-region is as follows: each scan line in the sub-region is divided into several shorter segments. When the laser is turned on to scan the sub-region, after scanning a segment, the laser does not immediately move to the start point of the next segment, but moves to a distant, unprocessed position in the sub-region according to a preset micro-jump mode to continue processing.

[0011] The specific implementation method of path optimization within the above sub-region is as follows: 3.a.1) Divide each scan line within the sub-region into several shorter segments; the division is based on fixed length or dynamic length; the fixed length is set to an optimal segment length based on the thermal diffusivity of the material and the laser parameters; the dynamic length division is based on a real-time thermal simulation model, and the division is performed when the predicted temperature is about to reach a critical value; 3.a.2) Create a processing task pool based on the segments obtained in step 3.a.1); the segments are unordered and discretely distributed in the task pool; 3.a.3) Turn on the laser and select any segment from the processing task pool created in step 3.a.2) to start scanning; after scanning the segment, select the next segment as the scanning target from the remaining unprocessed segments in the processing task pool according to an objective function; the objective function is either a distance-maximizing objective function or a temperature-minimizing objective function; the distance-maximizing objective function selects the segment with the greatest spatial distance from the current scanning segment; the temperature-minimizing objective function prioritizes selecting the segment near the area with the lowest surface temperature in the current sub-region.

[0012] The above-mentioned global adaptive filling and boundary stitch suppression method also includes, after step 3), step 4), co-optimizing the key parameters when scanning all sub-regions in the processing area.

[0013] The collaborative optimization in step 4) above includes boundary energy gradient control and handling the overlap of adjacent sub-regions at the stitching point; the key parameters of the boundary energy gradient control include the rated power of the scanning laser and the scanning speed; the key parameters of the handling of the overlap of adjacent sub-regions at the stitching point include the overlap width, the scanning speed, and the scanning frequency.

[0014] The advantages of this invention are: This invention provides a global adaptive filling and boundary stitching suppression method, which includes: 1) obtaining the surface to be processed and dividing it into multiple sub-regions; 2) generating a scanning filling path in each sub-region obtained in step 1); 3) optimizing the scanning filling paths of all sub-regions in the surface to be processed obtained in step 2) to complete global adaptive filling and boundary stitching suppression. This invention systematically solves key technical problems in large-format processing, such as uneven etching depth and edge stitching marks caused by thermal accumulation effects, platform deformation, and laser focus drift, by introducing adaptive partitioning and adaptive filling techniques. This invention employs an original adaptive planar region partitioning mechanism, constructing a dynamic partitioning algorithm based on competitive seed point growth and iterative centroid optimization. Through random seeding, distance assignment, centroid calculation, and multiple iterations, the processing surface is automatically divided into multiple convex polygonal sub-regions with uniform area and regular shape. The core innovation of this mechanism lies in transforming the traditional large-format processing problem into a parallel processing problem of multiple controllable sub-regions, laying the foundation for subsequent fine processing from the source. Simultaneously, employing a global fill path planning strategy with thermal management capabilities, this invention innovatively proposes a micro-jump scanning mode. By dividing the scanning path into discrete segments and placing them into the processing task pool, and using an objective function of maximizing distance or minimizing temperature for thermal optimization scheduling, it effectively breaks the thermal accumulation cycle caused by continuous scanning. Combined with random perturbation technology at the splicing boundary vertices, it achieves the dual objectives of thermal distribution control and splicing trace elimination. Furthermore, this invention employs a multi-parameter collaborative splicing trace elimination technology, establishing a boundary energy gradient control system and a precise parameter matching mechanism for overlapping areas. Through precise coordination of power modulation and scanning parameters, seamless splicing is achieved at the boundaries of sub-regions. Specifically, this invention has the following advantages: 1) Dynamic mesh partitioning ensures the processing independence of each sub-region, reducing mutual influence between regions. 2) Discrete staggered fill paths combined with energy gradient control fundamentally eliminate splicing traces caused by path overlap and energy abrupt changes in traditional methods. 3) By optimizing the processing path and parameters, production efficiency is significantly improved while ensuring processing quality. 4) This invention is applicable to the precision laser processing of various hard and brittle materials. Attached Figure Description

[0015] Figure 1 is a schematic diagram of the overall process of the global adaptive filling and boundary stitching suppression method provided by the present invention; Figure 2 is a flowchart of the processing area division algorithm used in the present invention; Figure 3 is a schematic diagram of the processing area division effect used in the present invention; Figure 4 is a schematic diagram (a) of the processing boundary stitching effect of the planned area caused by the straight line filling method and a schematic diagram (b) of the actual processing effect; Figure 5 is a schematic diagram (a) of the processing boundary stitching effect of the planned area using the straight line boundary misalignment filling method and a schematic diagram (b) of the actual processing effect using the straight line boundary misalignment filling method, processing task pool strategy, processing parameter optimization and stitching trace elimination strategy. Detailed Implementation

[0016] Referring to Figure 1, this invention provides a global adaptive filling and boundary stitch suppression method, which includes the following steps: 1) Adaptive planar region partitioning: Based on breadth-first search and multi-point parallel growth, region partitioning is achieved through a competition mechanism, dividing the processing area into multiple sub-regions, specifically including: 1.1) Random seeding: A certain number of initial coordinate points, called seed points, are randomly generated within the processing area. These seed points are the core of each region's growth and the starting point for subsequent calculations.

[0017] 1.2) Region Division: Traverse every point on the processing area, where the coordinates are (X, Y). Calculate the distance from each point on the processing area to all seed points obtained in step 1.1). Find the closest seed point and mark it as its region ID. Generate a partition map based on the convex hull calculation method. This map records which region each area's coordinates belong to; the partition map is the initial sub-region. 1.3) Centroid Calculation: Calculate a centroid for each region generated in the previous step. For a region, sum and average all X coordinates belonging to that region to obtain a new X coordinate X_avg. Similarly, sum and average all Y coordinates to obtain a new Y coordinate Y_avg. This average coordinate (X_avg, Y_avg) is the centroid of the region.

[0018] 1.4) Iterative Optimization: Use all the centroids calculated in the previous step as the new seed points, replacing the old seed points. Repeat steps 1.2) and 1.3), typically for 10-20 iterations. As the iterations proceed, the seed points will continuously move to the center of the region, making the area of ​​each region more and more uniform and the shape of the region more regular.

[0019] 1.5) Collect boundary points: For each region finally generated in step 1.4), extract its boundary information. Traverse all vertex coordinates of a region. If the neighboring vertices of a vertex in the up, down, left, right, or diagonal directions belong to other regions, then this vertex is a boundary point. Collect all such boundary points.

[0020] 1.6) Calculate the convex hull: The collected set of boundary points is scattered and contains internal points, and its shape may be an irregular concave shape. Perform convex hull calculation on this set of boundary points to ensure that the polygon (irregular concave shape) is free of jagged edges. For example, the convex hull calculation can be performed using the Graham Scan algorithm, which transforms the set of boundary points for each region into a smooth convex polygon defined by a small number of vertices.

[0021] Finally, after the above steps, the regional planning of the entire area is achieved. The specific process is shown in Figure 2, and the zoning graphic effect is shown in Figure 3.

[0022] 2) Global Fill Path Planning: After completing the adaptive partitioning of the image area, the processing object is transformed from a single large-format area into a series of relatively independent sub-regions with controllable dimensions. For each convex polygon sub-region, the scan fill path is generated according to the following steps:

[0023] 2.1) Sub-region Orientation and Scan Line Generation: Determine the optimal filling direction based on the shape and position of the area. Calculate the line spacing S = d × (1 - η) based on the effective spot diameter d of the femtosecond laser and the required overlap ratio η. In the determined filling direction, generate a set of parallel scan lines covering the entire sub-region with a line spacing S.

[0024] 2.2) Scan line truncation and path segment generation: Determine the scanning direction, fill the entire area with a straight line along the scanning direction, and then truncate the filling scanning line at the boundary of two adjacent sub-regions (i.e., the parallel scanning line as described above), and finally obtain the scan filling line of all regions.

[0025] 2.3) Path Connection and Thermal Management Optimization: Straight lines are used to fill sub-regions, and the vertices of the stitching boundary scan lines are randomly perturbed along the scan line direction to make them intersect, thus eliminating the influence of stitching boundary traces. This optimization mainly involves dividing each straight line into several shorter segments. After processing one segment, the laser does not immediately move to the starting point of the next segment, but instead moves to a distant, unprocessed position within that sub-region according to a preset micro-jump pattern to continue processing. This strategy breaks the inherent thermal cycle of continuous scanning, forcing heat to diffuse and dissipate more evenly within the sub-region, avoiding the formation of local hot spots. The main process is as follows:

[0026] 2.3.1) Splicing Vertex Perturbation: For the vertices of straight lines located on the splicing boundary, a small random displacement perturbation is applied. This perturbation is usually performed within a specific range perpendicular to the original scan line direction. This operation makes the start and end points of the scan lines of adjacent sub-regions no longer precisely aligned at the boundary, thereby macroscopically eliminating the regular splicing traces caused by path alignment.

[0027] 2.3.2) Path Segmentation: The system no longer treats the long scan line as a whole, but divides it into several shorter segments. The segmentation can be based on: a) Fixed length: An optimal segment length is set based on the thermal diffusivity of the material and the laser parameters. b) Dynamic segmentation: Segmentation is performed based on a real-time thermal simulation model when the predicted temperature is about to reach a critical value.

[0028] 2.3.3) Creating a processing task pool based on the path: All these segmented segments are no longer arranged in their original geometric order, but are placed into a virtual processing task pool. At this point, these segments are spatially disordered and discretely distributed.

[0029] 2.3.4) Micro-jump Machining and Thermal Optimization Scheduling: The laser selects the first segment from the machining task pool for processing. After completing the machining of this segment, the algorithm does not directly proceed to process the next adjacent segment. Instead, the system performs a micro-jump. Based on an objective function, the next optimal target is selected from the remaining unprocessed segments in the machining task pool. The core of this objective function is usually: a) Maximize distance: Select the segment with the greatest spatial distance from the current machining position. b) Minimize temperature: Integrate thermal imaging or a thermal model to prioritize selecting segments near the area with the lowest surface temperature of the current workpiece. c) Execution: The laser head quickly moves to the starting point of the selected new segment and begins machining. This process is repeated until all segments in the machining task pool have been processed.

[0030] 3) Processing Parameter Optimization: After completing the global path planning, to achieve seamless splicing between sub-regions and eliminate visible splicing marks, key processing parameters need to be optimized collaboratively. Splicing marks mainly originate from abrupt changes in laser parameters at the boundaries of sub-regions and path overlaps / gaps. Step 3) addresses this through the following parameter strategies:

[0031] 3.1) Boundary Energy Gradual Control: This aims to prevent abrupt energy changes at the sub-region boundary due to laser switching delays or galvanometer acceleration / deceleration, which could create deep grooves or protrusions. Laser power modulation is integrated inside the sub-region boundary. Using the laser's analog modulation port, the acousto-optic modulator is controlled to smoothly and linearly transition the laser power from 85% of the rated power at the boundary to 100% inside the region. The scanning speed remains constant within this boundary transition zone to avoid dwell time differences caused by speed variations.

[0032] 3.2) Precise Design and Parameter Matching of the Overlapping Zone: Ensure that adjacent sub-regions have just the right amount of material removed at the splicing point, with neither residue nor overcutting. Set a fixed splicing overlap width, typically 1.5-3 laser spot diameters. Within the overlapping zone, the repetition frequency and scanning speed must be precisely matched to ensure that the pulse overlap rate is consistent with the area inside. Single or double trimming scans can be performed on the overlapping zone, using slightly lower pulse energy and a higher scanning speed to smoothly remove any possible fine residue without introducing new heat input.

[0033] The final results of the processing strategy are compared in Figures 4 and 5. Figure 4 is a schematic diagram of the filling processing result using only the straight line filling method, and Figure 5 is a schematic diagram of the processing optimization result using the misalignment filling, processing task pool strategy, processing parameter optimization and splicing trace elimination strategy of this method.

[0034] This embodiment describes in detail a femtosecond laser etching process for a silicon carbide adsorption stage with a diameter of 450 mm.

[0035] 1. Processing Object and Equipment: Processing object: a reaction sintered silicon carbide flat blank with a diameter of 450mm and a thickness of 30mm, requiring the processing of a precision mesh flow channel with a depth of 50μm and a width of 80μm on its surface.

[0036] Processing equipment: Femtosecond laser processing center: laser wavelength 1030nm, pulse width 350fs, repetition frequency adjustable range 100kHz-2MHz, maximum single pulse energy 50μJ.

[0037] Precision motion platform: XY travel 500mm×500mm, positioning accuracy ±1μm.

[0038] High-speed galvanometer scanning system: scanning speed 0-5m / s.

[0039] Online visual positioning system (CCD).

[0040] Acousto-optic modulator (AOM): Used for rapid fine-tuning of laser power.

[0041] 2. Specific implementation steps 1) Region division based on adaptive competition mechanism 1.1) Random seeding: 120 initial seed points are randomly generated within a circular processing area with a diameter of 450mm.

[0042] 1.2) Iterative optimization: The processing area is discretized into a 1μm × 1μm vertex mesh. The first region partitioning and centroid calculation are performed, with 15 iterations. After iteration, the 120 seed points are moved to the center of their respective regions, forming 120 uniformly sized sub-regions.

[0043] 1.3) Convex Hull Generation: Boundary vertices are collected for each region. The Graham Scan algorithm is used to calculate the convex hull of each boundary point set, transforming jagged boundaries into smooth convex polygons. Finally, the entire processing area is divided into 120 convex polygon sub-regions with regular shapes and similar sizes.

[0044] 2) Global Fill Path Planning 2.1) Scan Line Generation: The effective laser spot diameter d = 15 μm, the overlap rate η = 30%, and the line spacing S = 15 × (1 - 0.3) = 10.5 μm are calculated. Scan lines parallel to the scanning direction and with a line spacing of 10.5 μm are generated for the area.

[0045] 2.2) Path segment generation: Intersect and cut off each scan line with the region boundary, and then randomly perturb the intersection point of each boundary scan line to obtain a series of intersecting straight line processing segments.

[0046] 2.3) Micro-jump thermal management path segmentation: All processing segments with a length exceeding 500μm are divided into shorter sub-segments with a fixed length of 200μm.

[0047] Create a processing task pool: Place all sub-segments (approximately 1500 segments) within this sub-region into a virtual "processing task pool" and shuffle their spatial order.

[0048] Thermal optimization scheduling: a) The laser retrieves the first segment from the task pool for processing. b) After processing, the algorithm selects the segment furthest from the current position in the task pool as the next processing target, based on the principle of maximizing distance. c) The laser head performs a micro-jump, quickly moving to the starting point of the target segment for processing. d) This process is repeated until the task pool is empty. This strategy ensures uniform heat distribution within the sub-region; the maximum temperature rise in this region during actual processing does not exceed 45℃, effectively avoiding thermal damage.

[0049] 3) Processing Parameter Optimization: At the sub-region boundary, the following strategies are implemented to ensure seamless splicing: 3.1) Boundary Energy Gradual Control: A transition zone with a width of 50 μm is set inside the sub-region boundary. By controlling the laser power, it linearly increases from 85% of the rated power at the boundary (corresponding to 4.25 μJ) to 100% (5 μJ) inside the region within this transition zone. The scanning speed remains constant at 2 m / s within the transition zone.

[0050] 3.2) Precise Design of the Overlapping Zone: The overlap width is set to two spot diameters (30 μm). Within the overlap zone, the laser repetition frequency (1 MHz) and scanning speed (2 m / s) are precisely matched to maintain the same pulse overlap rate as within the region. After all sub-regions are processed, a secondary trimming scan is performed on all seams. The trimming parameters are: pulse energy 3 μJ, scanning speed 3 m / s. This smoothly removes any nanoscale residues that may exist at the seams without introducing significant heat input.

Claims

1. A global adaptive filling and boundary stitch suppression method, characterized in that: The global adaptive filling and boundary stitch suppression method includes the following steps: 1) Obtain the surface to be processed and divide it into multiple sub-regions; 2) Generate a scan filling path in each sub-region obtained in step 1); 3) Optimize the scan filling paths of all sub-regions in the surface to be processed obtained in step 2) to complete the global adaptive filling and boundary stitch suppression.

2. The global adaptive filling and boundary stitching suppression method according to claim 1, characterized in that: The specific implementation of step 1) is as follows: 1.1) Randomly generate a certain number of seed points within the area to be processed; 1.2) Form an initial sub-region based on the seed points obtained in step 1.1); 1.3) Calculate the centroid for each initial sub-region generated in step 1.2); 1.4) Iteratively optimize the centroid obtained in step 1.3) to obtain the optimized sub-region; 1.5) Obtain the boundary points of the optimized sub-region obtained in step 1.4); 1.6) Perform convex hull calculation on the boundary points of the optimized sub-region obtained in step 1.5) to obtain the sub-region of the convex polygon, thus completing the division of the area to be processed.

3. The global adaptive filling and boundary stitching suppression method according to claim 2, characterized in that: The specific implementation of step 1.2) is as follows: traverse each point on the surface to be processed, where the coordinates of the point are (X, Y), calculate the distance from each point on the surface to be processed to all seed points obtained in step 1.1), find the closest seed point, and mark this point as the region ID of the seed point; generate a partition map according to the convex hull calculation method, and the partition map is the initial sub-region; the specific implementation of step 1.3) is as follows: for each initial sub-region, sum up all the X coordinates belonging to this initial sub-region and calculate the average to obtain a new X coordinate X_avg; sum up all the Y coordinates and calculate the average to obtain a new Y coordinate Y_avg; this average coordinate (X_avg, Y_avg) is the centroid of the initial sub-region; the specific implementation of step 1.4) is as follows: use the centroids of all the initial sub-regions calculated in step 1.3) as the seed points of the new round, replace the seed points in step 1.1), and repeat steps 1.2) and 1.3) until the area of ​​each sub-region is uniform and the shape of each sub-region is regular, and obtain the optimized sub-region.

4. The global adaptive filling and boundary stitch suppression method according to claim 3, characterized in that: The specific implementation of step 2) is as follows: 2.1) Based on the shape and position of the sub-region obtained in step 1), determine the scanning path filling direction of each sub-region; 2.2) Based on the scanning path filling direction determined in step 2.1), generate parallel scanning lines covering the entire sub-region within the sub-region. 2.3) Cut off the parallel scan lines at the boundary of two adjacent sub-regions to obtain the scan fill path of all sub-regions in the area to be processed.

5. The global adaptive filling and boundary stitch suppression method according to claim 4, characterized in that: The row spacing of the parallel scan lines in step 2.2) is S, and S is calculated as follows: S = d × (1 - η); where d is the effective spot diameter of the femtosecond laser; and η is the overlap ratio required by the process.

6. The global adaptive filling and boundary stitch suppression method according to claim 5, characterized in that: The optimization in step 3) includes path optimization within a sub-region and path optimization between adjacent sub-regions.

7. The global adaptive filling and boundary stitch suppression method according to claim 6, characterized in that: The specific implementation of the path optimization between adjacent sub-regions is as follows: For the straight line vertices located on the path between adjacent sub-regions, a small random displacement perturbation is applied, causing the start and end points of the scan lines at the boundary of the adjacent sub-regions to no longer be precisely aligned. The disturbance is performed within a specific range perpendicular to the original scan line direction; the path optimization within the sub-region is as follows: each scan line within the sub-region is divided into several shorter sub-segments. When the laser is turned on to scan the sub-region, after scanning one sub-segment, the laser does not immediately move to the starting point of the next sub-segment, but moves to a distant, unprocessed position within the sub-region according to a preset micro-jump mode to continue processing.

8. The global adaptive filling and boundary stitching suppression method according to claim 7, characterized in that: The specific implementation of path optimization within the sub-region is as follows: 3.a.1) Divide each scan line within the sub-region into several shorter segments; the division is based on a fixed length or a dynamic length; the fixed length is determined by setting an optimal segment length based on the thermal diffusivity of the material and the laser parameters; the dynamic length division is based on a real-time thermal simulation model and is performed when the predicted temperature is about to reach a critical value; 3.a.2) Create a processing task pool based on the segments obtained in step 3.a.1); the segments are unordered and discretely distributed in the task pool; 3.a.3) Turn on the laser and select any segment from the processing task pool created in step 3.a.2) to start scanning; after scanning the segment, select the next segment as the scanning target from the remaining unprocessed segments in the processing task pool according to an objective function; the objective function is either a distance-maximizing objective function or a temperature-minimizing objective function; the distance-maximizing objective function selects the segment with the greatest spatial distance from the current scanning segment; the temperature-minimizing objective function prioritizes selecting the segment near the area with the lowest surface temperature in the current sub-region.

9. The global adaptive filling and boundary stitch suppression method according to any one of claims 1-8, characterized in that: The global adaptive filling and boundary stitch suppression method further includes, after step 3), step 4), co-optimizing the key parameters when scanning all sub-regions in the processing area.

10. The global adaptive filling and boundary stitching suppression method according to claim 9, characterized in that: The collaborative optimization in step 4) includes boundary energy gradient control and handling the overlap of adjacent sub-regions at the stitching point; the key parameters of the boundary energy gradient control include the rated power of the scanning laser and the scanning speed; the key parameters of the handling of the overlap of adjacent sub-regions at the stitching point include the overlap width, the scanning speed and the scanning frequency.

Images