Laser cleaning method for pollutants on surface of waste metal

Abstract

The invention provides a waste metal surface pollutant laser cleaning method which comprises the following steps: S1, carrying out three-dimensional imaging on the surface of a workpiece through scanning equipment, obtaining a surface model, carrying out grid division on the surface model into independent units, and determining curvature change characteristics of each independent unit; s2, grouping the pollutant thickness of each independent unit, judging whether the pollutant thickness exceeds a preset thickness threshold value or not, and if a thickness grouping result shows a high-density area, planning a cleaning path by adopting an A star algorithm to obtain a cleaning path for the high-density area; s3, laser parameters are adjusted according to the cleaning path in combination with the distribution density, a hierarchical processing sequence is determined, layer-by-layer scanning is executed on each independent unit through the hierarchical processing sequence, and preliminary cleaning effect data are obtained; and S4, if the preliminary cleaning effect data shows that pollutants remain, the A star algorithm is repeated to adjust the cleaning path, and an optimized cleaning result is obtained. The cleaning precision and efficiency of the complex workpiece are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent processing technology, and in particular to a laser cleaning method for contaminants on the surface of waste metal. Background Technology

[0002] In industrial production, workpiece surface cleaning technology is a crucial process that directly impacts product quality and production efficiency. Whether for metal parts or aerospace components, the removal of surface contaminants has a decisive influence on subsequent processing and performance. The importance of this technology lies not only in its relation to workpiece functionality but also in its implications for production costs and environmental protection, making it an indispensable key link in industrial manufacturing.

[0003] However, existing cleaning methods often fall short when dealing with complex workpieces. Many traditional techniques struggle to adapt to the diversity of workpiece shapes and the uneven distribution of contaminants, frequently resulting in poor cleaning effects or low efficiency. More seriously, when handling complex geometries, these methods are prone to causing unnecessary damage to the workpiece surface due to a lack of specificity, affecting the integrity of the substrate. This limitation is particularly pronounced in high-precision scenarios, necessitating a more flexible and precise solution.

[0004] Against this backdrop, the planning of cleaning paths has become a core challenge for technological breakthroughs. Complex workpiece surfaces often exhibit irregular geometric features, and the distribution of contaminants may vary across different areas. If a reasonable path design cannot be implemented based on the specific shape of the workpiece surface and the distribution characteristics of contaminants, the cleaning process will lack focus, potentially leading to over-treatment and damage in some areas while others remain uncleaned. For example, when cleaning an aerospace component with a multi-curved structure, if the path planning does not conform to the surface variations, uneven laser energy distribution may occur, resulting in contaminant residue in some recessed areas while protruding areas are worn down due to concentrated energy. This problem of uneven cleaning and substrate damage caused by unreasonable path planning persists throughout the entire cleaning process, forming a major technological bottleneck.

[0005] Therefore, how to accurately plan the path for the uneven distribution of contaminants on the surface of workpieces with complex geometries in order to achieve efficient cleaning while avoiding damage to the substrate has become a key problem that needs to be solved in this study. Summary of the Invention

[0006] To address the technical problems mentioned in the background section, this invention provides a laser cleaning method for contaminants on the surface of waste metal. The method includes: S1, performing three-dimensional imaging of the workpiece surface using a scanning device to obtain a surface model and dividing it into independent units, determining the curvature variation characteristics of each independent unit; S2, grouping the contaminant thickness of each independent unit, determining whether the contaminant thickness exceeds a preset thickness threshold, and if the thickness grouping result shows a high-density area, then using the A* algorithm to plan a cleaning path to obtain a cleaning path for the high-density area; S3, adjusting the laser parameters based on the cleaning path and distribution density to determine a layered processing sequence, and performing layer-by-layer scanning on each independent unit using the layered processing sequence to obtain preliminary cleaning effect data; S4, if the preliminary cleaning effect data indicates residual contaminants, then repeating the A* algorithm to adjust the cleaning path to obtain an optimized cleaning result.

[0007] Further, step S1 includes: step S11, the scanning device collects the height information and pollutant reflectance spectrum data of the workpiece surface to generate a surface model; step S12, the complex shape is decomposed into multiple independent units according to the surface model using a meshing method; step S13, for each independent unit, the curvature change characteristics are calculated and correlated with the pollutant distribution density to determine the curvature change characteristics of each independent unit.

[0008] Furthermore, step S11 includes: scanning the workpiece surface point by point with a scanning device to obtain height data at each location and form point cloud data; using reflectance spectral analysis technology to collect the reflection characteristics of pollutants to specific wavelengths of light and to preliminarily determine the type and thickness distribution of pollutants; and generating a three-dimensional surface model from the point cloud data using a triangulation algorithm.

[0009] Furthermore, step S13 includes: determining the curvature value by analyzing the changes in the surface normal of each independent unit.

[0010] Further, step S2 includes: step S21, extracting the pollutant thickness statistics of each independent unit as clustering input, and performing K-means clustering grouping; step S22, dividing into low, medium and high density groups according to the set thickness threshold of the grouping, and obtaining the thickness grouping result; step S23, if the thickness grouping result shows a high density area, then taking the independent unit of the high density area as the starting node and planning the cleaning path.

[0011] Furthermore, step S23 includes: step S231, generating a cleaning path using curvature change features as path cost weights; step S232, associating the cleaning path with each independent unit of the surface model to ensure that the path covers high-density areas; and step S233, outputting the cleaning path for high-density areas.

[0012] Furthermore, step S3 includes: step S31, dynamically setting the laser power pulse width and scanning speed according to the cleaning path and thickness grouping results, and adjusting the laser parameters; step S32, determining the layered processing sequence based on the adjusted laser parameters and distribution density; and step S33, performing a sequential layer-by-layer scan from the surface of the contaminant to the base layer on each independent unit through the layered processing sequence to obtain preliminary cleaning effect data.

[0013] Furthermore, step S31 includes:

[0014] The laser parameters are adjusted based on the thickness and distribution density of the contaminants. For high-density areas, the laser power is increased and the pulse width is shortened; for low-density areas, the laser power is decreased and the pulse width is increased.

[0015] Furthermore, step S4 includes: step S41, using the remaining high-density areas in the preliminary cleaning effect data as new input, and re-grouping them using K-means clustering; step S42, using the A* algorithm to plan the adjusted cleaning path based on the new thickness grouping results; and step S43, repeating the hierarchical processing sequence to obtain the optimized cleaning result.

[0016] Further, step S41 includes: step S411, extracting the statistical value of residual pollutant thickness as the input for new clustering, and performing the K-means clustering; step S412, determining whether the new grouping exceeds the preset residual thickness threshold, and if a new high-density area is displayed, generating a new starting node; step S413, combining the curvature change feature as the new path cost weight, and using the A* algorithm to obtain the adjusted cleaning path.

[0017] The technical solution provided by this invention has the following beneficial effects:

[0018] This invention discloses a laser cleaning method for contaminants on the surface of waste metals. It proposes a systematic solution to address the challenges of uneven contaminant distribution, diverse shapes, and high cleaning precision requirements on complex workpiece surfaces. This problem relates to the cleaning needs of workpieces of different materials and sizes in industrial production, such as metal parts and aerospace components, and how to achieve precise and efficient cleaning under complex geometries while avoiding damage to the substrate. This invention uses 3D imaging technology to accurately construct a workpiece surface model, combines contaminant distribution analysis and path planning, utilizes K-means clustering and A* algorithm to optimize the cleaning path, and dynamically adjusts laser parameters to achieve layered processing, ensuring a highly targeted and efficient cleaning process. Furthermore, residue detection and iterative path optimization ensure that the cleaning effect meets expectations. The technical advantages of this invention are significantly improved accuracy and efficiency in cleaning complex workpieces, reduced resource waste, protection of the workpiece substrate, applicability to various industrial scenarios, and high adaptability and practicality. Attached Figure Description

[0019] Figure 1 This is a flowchart of a laser cleaning method for contaminants on the surface of waste metal according to the present invention. Detailed Implementation

[0020] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] This invention provides a laser cleaning method for contaminants on the surface of waste metal, aiming to achieve efficient cleaning of complex workpiece surfaces through precise three-dimensional imaging, contaminant distribution analysis, and path planning. The technical solution of this invention is described in detail below with reference to specific embodiments to make the objectives, technical solutions, and advantages of this invention clearer. In one embodiment, the method provided by this invention is mainly aimed at situations where contaminants are unevenly distributed and have complex shapes on the workpiece surface, achieving precise cleaning through step-by-step processing. The overall process includes steps such as three-dimensional modeling of the workpiece surface, grouping contaminant thicknesses, planning cleaning paths, and adjusting laser parameters. The method is applicable to various industrial scenarios, such as oil stain cleaning after metal parts processing and coating removal from the surface of aerospace components. The specific implementation of each step will be described in detail below to ensure the operability of the technical solution.

[0022] like Figure 1 As shown, the method specifically includes S1, performing three-dimensional imaging of the workpiece surface using a scanning device, obtaining a surface model and dividing it into independent units, and determining the curvature variation characteristics of each independent unit.

[0023] Specifically, this step aims to acquire basic data on the geometric information and contaminant distribution of the surface of the scrap metal workpiece, providing a basis for subsequent cleaning path planning and parameter adjustment. The scrap metal workpiece can be metal sheet, mechanical parts, or composite material components, and its surface may contain contaminants such as oil, oxide layers, or coatings. By collecting surface information through scanning equipment, a comprehensive understanding of the complex shape and contaminant distribution of the workpiece surface can be obtained. In one possible implementation, step S1 is further broken down into more specific operational procedures:

[0024] Step S11: The scanning device collects the height information and pollutant reflectance spectrum data of the workpiece surface to generate a surface model.

[0025] Specifically, the scanning equipment can be a high-precision laser scanner or a structured light scanning system. By scanning the workpiece surface point by point, the height data of each location is obtained, forming point cloud data. Simultaneously, reflectance spectroscopy analysis technology is used to collect the reflection characteristics of contaminants to specific wavelengths of light, in order to preliminarily determine the type and thickness distribution of contaminants. After processing, the point cloud data generates a three-dimensional surface model. The model digitally represents the geometric features of the workpiece surface, such as protrusions, depressions, or flat areas. The model generation process uses a triangulation algorithm to connect the point cloud data into a continuous surface, ensuring the integrity and accuracy of the model. Step S12 involves decomposing the complex shape into multiple independent units based on the surface model using a mesh generation method. Specifically, mesh generation is the process of dividing the three-dimensional model into multiple small regions, each called an independent unit, typically a rectangular or triangular mesh. The purpose of mesh generation is to decompose the complex workpiece surface into individually processable units, facilitating subsequent analysis and cleaning operations. The mesh size can be adjusted according to the complexity of the workpiece surface. For example, for areas with large curvature changes, the mesh can be set smaller to capture detailed features; while for flat areas, the mesh can be appropriately enlarged to reduce computational load. After segmentation, each independent unit has a unique identifier and location information, which can be linked to subsequent pollutant distribution data. Step S13: For each independent unit, calculate the curvature change characteristics and correlate them with the pollutant distribution density to determine the curvature change characteristics of each independent unit.

[0026] Specifically, curvature variation features are used to characterize the geometric changes of the workpiece surface within a unit, such as whether it is flat, raised, or recessed. When calculating curvature variation features, the curvature value can be determined by analyzing the changes in the surface normal within each independent unit. Areas with larger curvature values ​​usually indicate complex shapes, which may have special requirements for cleaning paths and laser parameters. Simultaneously, combining curvature variation features with contaminant distribution density analysis can determine whether contaminants are concentrated in areas with large curvature variations; for example, more contaminants may accumulate in grooves. This correlation analysis helps prioritize complex areas during subsequent path planning. In one embodiment, taking the cleaning of metal parts as an example, the specific application of step S1 is illustrated. Assume the workpiece is a car engine block with machining oil stains and localized oxide layers on its surface. First, a laser scanner is used to scan the block surface to obtain height information and reflectance spectral data, generating a three-dimensional model. The model shows that the block surface has multiple grooves and raised areas, with contaminants mainly concentrated in the grooves. Next, the model is divided into multiple independent units with a mesh size of 2 mm by 2 mm to accommodate complex shapes. For each unit, the curvature variation characteristics were calculated, revealing that the groove region had a higher curvature value and the contaminant reflectance spectrum indicated a larger thickness. Through correlation analysis, these high-curvature units were identified as key areas for subsequent cleaning. This method accurately locates cleaning difficulties and provides data support for subsequent steps. In step S2, the contaminant thickness of each independent unit was grouped, and it was determined whether the contaminant thickness exceeded a preset thickness threshold. If the thickness grouping results showed high-density areas, the A* algorithm was used to plan the cleaning path, resulting in a cleaning path specifically for these high-density areas.

[0027] Specifically, the purpose of this step is to classify the distribution of pollutants, identify areas requiring focused cleaning, and plan reasonable cleaning paths for them to improve cleaning efficiency. In one possible implementation, step S2 is further refined into specific operations.

[0028] Step S21: Extract the pollutant thickness statistics of each independent unit as clustering input and perform K-means clustering.

[0029] Specifically, the statistical value of pollutant thickness can be obtained through reflectance spectral data analysis, for example, by inferring the thickness value through changes in spectral reflectance intensity. K-means clustering is an unsupervised learning method that groups independent units with similar thickness values ​​into the same group. The number of groups can be set according to actual needs, for example, three groups representing low, medium, and high thickness regions, respectively. The clustering process iteratively calculates the distance from each unit to the cluster center, gradually optimizing the grouping results to ensure the similarity of thickness distribution. Step S22: Based on the set thickness threshold, low, medium, and high density groups are divided to obtain the thickness grouping results.

[0030] Specifically, the thickness threshold can be preset according to the performance of the cleaning equipment and the type of contaminants. For example, for oil stain cleaning, the thickness threshold can be set to 0.5 mm, and units exceeding this value are classified as high-density groups. The grouping results are recorded in the form of a data table, showing the group to which each unit belongs, such as low-density group, medium-density group, or high-density group. The judgment results show the number and distribution of units in the high-density group, providing a basis for subsequent path planning. In step S23, if the thickness grouping results show a high-density area, the independent units in the high-density area are used as starting nodes to plan the cleaning path.

[0031] Specifically, high-density areas are typically areas with thicker contaminant layers and higher cleaning difficulty, requiring priority treatment. The A* algorithm is a heuristic search method that finds the optimal path from the starting node to the target node by evaluating path costs. During the planning process, independent cells in high-density areas are set as starting nodes, with the goal of covering all high-density cells while minimizing path length and cleaning time. In one embodiment, step S23 is further refined into a specific implementation method.

[0032] Step S231: Use curvature change features as path cost weights to generate a cleaning path.

[0033] Specifically, the path cost weight is used to measure the difficulty of traversing each unit in the path planning. Units with larger curvature variations usually mean higher cleaning difficulty, so their cost weights are set higher. When calculating the path, the A* algorithm comprehensively considers the path length and cost weight, prioritizing paths with lower costs. For example, for groove regions with large curvature variations, the algorithm will try to plan smooth paths to avoid frequent angle adjustments by the laser equipment. Step S232 associates the cleaning path with each independent unit of the surface model to ensure that the path covers high-density areas.

[0034] Specifically, after path planning is completed, the generated path is recorded as a series of cell identifiers to ensure that the path passes through all cells in the high-density region. Simultaneously, the path is aligned with the surface model to ensure that the laser device can accurately move to each cell position according to the path. The integrity of the path coverage is verified by checking whether all cells in the high-density region are included in the path. Step S233 outputs the cleaning path for the high-density region.

[0035] Specifically, the final output cleaning path is presented as a coordinate sequence, with each coordinate corresponding to the position of an independent unit. The path can be directly transmitted to the laser cleaning equipment as a guide for subsequent cleaning operations. Simultaneously, the path data can be stored as a file for easy adjustment or reuse. Path planning for high-density areas can effectively concentrate resources on areas with more contaminants, improving cleaning efficiency. In one embodiment, taking the cleaning of aerospace components as an example, the specific application of step S2 is illustrated. Assume the scrap metal workpiece is an aircraft wing component with localized coatings and oil stains on its surface. Based on the surface model and independent unit data generated in step S1, the contaminant thickness statistics for each unit are extracted, revealing that some areas have a thickness of 1.2 mm, far exceeding the preset thickness threshold of 0.5 mm. Using K-means clustering, the thickness data is divided into three groups: low, medium, and high, with the high-density group concentrated in the recessed areas at the wing edge. Next, it is determined that the high-density group has a larger number of units and needs to be cleaned first. Using these units as starting nodes, the A* algorithm is used to plan the path. During planning, curvature variation characteristics are used as cost weights. It was found that the curvature of concave areas is relatively large, so the path avoids sharp turns as much as possible to ensure stable operation of the laser equipment. The final generated path covers all high-density cells and is associated with the surface model, outputting a coordinate sequence. This method can accurately locate cleaning priorities and avoid resource waste. In another embodiment, for the cleaning scenario of mechanical parts, the diversified implementation of step S2 is further explained. Assume the scrap metal workpiece is a gear with processing oil stains on the surface and uneven thickness distribution. Through K-means clustering, it was found that the contaminant thickness in the gear tooth groove is relatively high, reaching 0.8 mm, exceeding the thickness threshold of 0.5 mm, and is classified as a high-density group. When planning the path, the cells in the tooth groove are used as starting nodes. Combining curvature variation characteristics, it was found that the curvature variation in the tooth groove area is drastic, resulting in a high path cost weight. The A* algorithm prioritizes paths that move continuously along the tooth groove direction during planning, reducing the number of turns of the laser equipment. The final path covers all high-density cells and is output as a coordinate sequence. This method can adapt to complex geometries and ensure a reasonable cleaning path. S3. Adjust the laser parameters according to the cleaning path and distribution density to determine the layered processing sequence. Perform layer-by-layer scanning on each independent unit through the layered processing sequence to obtain preliminary cleaning effect data.

[0036] Specifically, this step aims to dynamically adjust the operating parameters of the laser equipment based on the distribution of contaminants and the path planning results, ensuring the cleaning effect while avoiding damage to the workpiece substrate. In one possible implementation, step S3 is further refined into specific operations.

[0037] Step S31: Dynamically set the laser power, pulse width, and scanning speed according to the cleaning path and thickness grouping results, adjusting the laser parameters. The laser parameters are adjusted based on the contaminant thickness and distribution density. For example, for high-density areas, the laser power can be appropriately increased and the pulse width shortened to enhance cleaning intensity; while for low-density areas, the laser power can be decreased and the pulse width increased to avoid over-cleaning. The scanning speed is adjusted according to the path length and curvature changes; the speed is slowed down in areas with greater curvature to ensure that the laser energy acts uniformly on the surface. The parameter adjustment process can be achieved through a preset correspondence table; for example, for every 0.1 mm increase in thickness, the power increases by 10%. Step S32: Based on the adjusted laser parameters and distribution density, determine the layered processing sequence.

[0038] Specifically, the stratified treatment sequence refers to the order and method of removing contaminants layer by layer, typically starting from the surface layer and gradually progressing to the basal layer. When determining the sequence, the thickness and density of the contaminants are considered. For example, high-density areas may require multiple stratified treatments, removing a certain thickness of contaminants each time; while low-density areas may only require a single treatment. The stratified treatment sequence treats each independent unit as an object, recording the number of treatment layers and the laser parameters for each treatment to ensure the cleaning process proceeds in an orderly manner. Step S33 involves performing a sequential layer-by-layer scan from the contaminant surface to the basal layer using the stratified treatment sequence to obtain preliminary cleaning effect data.

[0039] Specifically, during scanning, the laser device moves to each independent unit position according to the cleaning path, acting on the contaminant surface layer by layer according to the layered processing sequence. After each scan, the cleaning status of that unit is recorded, such as the thickness of the removed contaminants or the residual amount. Preliminary cleaning effect data is obtained through surface inspection after scanning, for example, by measuring the residual thickness using reflectance spectroscopy or judging the surface cleanliness through visual inspection. Data is recorded on a unit-by-unit basis for easy subsequent analysis and adjustment. In one embodiment, taking the cleaning of metal sheets as an example, the specific application of step S3 is explained. Assume the workpiece is a steel sheet with oil and oxide layers on the surface, and the thickness distribution is uneven. Based on the cleaning path and thickness grouping results generated in step S2, it is found that the central area of ​​the steel sheet is a high-density area with a thickness of 0.7 mm. When adjusting the laser parameters, for the high-density area, the power is set to a higher value, the pulse width is shorter, and the scanning speed is slowed down to 10 mm per second to ensure that the energy is concentrated on the contaminant surface. When determining the layered processing sequence, the central area is set to 3 layers for processing, removing approximately 0.2 mm of contaminant thickness each time. During scanning, the laser device processes each unit sequentially along the path, starting from the surface and gradually penetrating deeper. After completing three layers of scanning, the residual thickness is detected using reflectance spectroscopy to generate preliminary cleaning effect data. This method effectively removes thick layers of contaminants while protecting the substrate from damage. In another embodiment, for the cleaning scenario of composite material components, the diversified implementation of step S3 is further explained. Assume the workpiece is a carbon fiber reinforced composite board with adhesive residue on the surface and a complex thickness distribution. Based on the cleaning path and thickness grouping results, the edge area of ​​the board is found to be a high-density area with a thickness of approximately 0.6 mm. When adjusting the laser parameters, for the high-density area, the power is set to a medium level, the pulse width is moderate, and the scanning speed is controlled at 15 mm per second to avoid thermal damage to the composite material substrate. When determining the layered processing sequence, the edge area is set to be processed in two layers, removing approximately 0.3 mm of contaminants each time. During scanning, the laser device processes each unit sequentially along the path, scanning layer by layer from the surface to the base layer. After completion, the surface cleanliness is judged by visual inspection to generate preliminary cleaning effect data. This method can adapt to sensitive materials, ensuring the cleaning process is safe and reliable while improving the cleaning effect. In one possible implementation, the flexibility of the layering sequence in step S3 is further explained. The determination of the layering sequence can also be combined with the characteristics of the workpiece material. For example, for harder metals, the number of layers can be appropriately increased, with each layer removing a smaller thickness to avoid surface overheating due to concentrated laser energy; while for softer composite materials, the number of layers is reduced, with each layer removing a larger thickness to shorten the processing time. After the sequence is determined, it can also be adjusted based on real-time feedback. For example, if the removal rate of contaminants in a certain unit is found to be slow during scanning, an additional layer can be added temporarily to ensure the cleaning effect. This flexibility can adapt to different workpieces and contaminant types, improving the versatility of the method.In another embodiment, taking industrial pipeline cleaning as an example, the application of step S3 in different scenarios is illustrated. Assume the workpiece is the inner wall of a stainless steel pipe with oil and rust layers on its surface, exhibiting uneven thickness distribution. Based on the cleaning path and thickness grouping results, the pipe's curved areas are identified as high-density regions, approximately 0.9 mm thick. When adjusting laser parameters, for high-density regions, the power is set to a higher value, the pulse width is shorter, and the scanning speed is slowed to 8 mm per second to ensure sufficient energy is applied to the curved area surface. When determining the layered processing sequence, the curved areas are set to be processed in 4 layers, removing approximately 0.2 mm of contaminant thickness each time. During scanning, the laser device is guided by a robotic arm, processing unit by unit along the path, starting from the surface layer and gradually penetrating to the base layer. After completion, the residual thickness is detected using reflectance spectroscopy to generate preliminary cleaning effect data. This method effectively handles workpieces with complex shapes, ensuring comprehensive cleaning coverage. It should be noted that the implementation of steps S1 to S3 can be further adjusted according to the actual application scenario. For example, the accuracy of the scanning device, the size of the grid division, the number of K-means clustering groups, and the adjustment range of laser parameters can all be flexibly set according to the workpiece type and contaminant characteristics. This diverse implementation ensures the method adapts to different industrial needs, while gradually optimizing the cleaning effect through step-by-step processing and data feedback. In one embodiment, the overall process of steps S1 to S3 is comprehensively explained using bearing component cleaning as an example. Assume the workpiece is a bearing outer ring with surface contamination from lubricating oil and metal shavings, exhibiting uneven thickness distribution. First, a three-dimensional image of the bearing outer ring surface is obtained using a laser scanner to acquire height information and reflectance spectral data, generating a surface model. The model shows multiple tiny grooves on the outer ring surface, with contaminants concentrated within these grooves. Next, the model is divided into multiple independent units with a grid size of 1 mm x 1 mm. The curvature variation characteristics of each unit are calculated, revealing higher curvature in the groove region. K-means clustering is used to extract statistical values ​​of contaminant thickness, grouping them into low, medium, and high groups. The high-density group is concentrated in the groove region, reaching a thickness of 0.6 mm, exceeding the thickness threshold of 0.4 mm. Starting with the high-density unit, the A* algorithm is used to plan the cleaning path, incorporating curvature variation characteristics as a cost weight to ensure the path covers all high-density areas. When adjusting laser parameters, for high-density areas, the power is set to a higher value, the pulse width is shorter, and the scanning speed is slowed to 12 millimeters per second. When determining the layered processing sequence, it is set to 3 layers, removing approximately 0.2 millimeters of thickness each time. During scanning, the laser device processes each unit along the path, and after completion, the residual thickness is detected by reflectance spectroscopy to generate preliminary cleaning effect data. This method can accurately locate the distribution of contaminants, optimize the cleaning path and parameters, and improve overall cleaning efficiency. In another embodiment, taking mold cleaning as an example, the comprehensive application of steps S1 to S3 is further illustrated. Assume the workpiece is an injection mold with residual plastic and oil stains on its surface, and a complex thickness distribution.A structured light scanning system was used to perform 3D imaging of the mold surface, acquiring height information and reflectance spectral data to generate a surface model. The model showed multiple small grooves and protrusions on the mold surface, with contaminants concentrated in the grooves. The model was then divided into multiple independent units with a mesh size of 1.5 mm x 1.5 mm. The curvature variation characteristics of each unit were calculated, revealing higher curvature in the groove regions. K-means clustering was used to extract statistical values ​​of contaminant thickness, grouping them into low, medium, and high density groups. The high-density group reached a thickness of 0.8 mm, exceeding the high-density thickness threshold of 0.5 mm. Using high-density units as starting nodes, the A* algorithm was employed to plan the cleaning path, incorporating curvature variation characteristics as cost weights to ensure a smooth path that covered all high-density areas. When adjusting laser parameters, for high-density areas, the power was set to a medium-high value, the pulse width was moderate, and the scanning speed was controlled at 10 mm per second. A 3-layer processing sequence was determined, removing approximately 0.25 mm of thickness each time. During scanning, the laser equipment processes each unit along the path, scanning layer by layer from the surface to the substrate. After completion, visual inspection is used to determine surface cleanliness and generate preliminary cleaning effect data. This method effectively handles complex mold surfaces, ensuring a precise and efficient cleaning process. In one possible implementation, the diversity of K-means clustering grouping in step S2 is further explained. The number of K-means clustering groups can be adjusted according to the complexity of the contaminant distribution. For example, for workpieces with relatively uniform thickness distribution, they can be divided into two groups, distinguishing only between low-density and high-density areas; while for workpieces with significant differences in thickness distribution, they can be divided into four or more groups to more finely classify contaminant density. The adjustment of the number of groups can be determined by analyzing the variance of the thickness data. Increasing the number of groups when the variance is large ensures that the grouping results are more representative. This flexibility can adapt to different contaminant distributions and improve the targeting of path planning. In another embodiment, taking the cleaning of ship components as an example, the application of steps S1 to S3 on large workpieces is illustrated. Assume the workpiece is a ship propeller blade with marine organisms and oil contamination on its surface, and uneven thickness distribution. A high-precision laser scanner was used to create a 3D image of the blade surface, acquiring height information and reflectance spectral data to generate a surface model. The model revealed multiple curved areas on the blade surface, with contaminants concentrated near the blade root. The model was then divided into multiple independent units with a grid size of 5 mm x 5 mm. The curvature variation characteristics of each unit were calculated, revealing higher curvature in the root region. K-means clustering was used to extract statistical values ​​of contaminant thickness, grouping them into low, medium, and high density groups. The high-density group reached a thickness of 1.5 mm, far exceeding the high-density thickness threshold of 0.8 mm. Using high-density units as starting nodes, the A* algorithm was employed to plan the cleaning path, incorporating curvature variation characteristics as cost weights to ensure the path covered all high-density areas. When adjusting laser parameters, for high-density areas, the power was set to a higher value, the pulse width shorter, and the scanning speed slowed to 20 mm per second.When determining the layered processing sequence, a 4-layer process is set, removing approximately 0.4 mm of thickness each time. During scanning, the laser device is guided by a large robotic arm, processing unit by unit along the path, scanning layer by layer from the surface to the base layer. After completion, the residual thickness is detected by reflectance spectroscopy to generate preliminary cleaning effect data. This method can adapt to large workpieces, ensuring that the cleaning process fully covers the contaminant area. It should be noted that the above embodiments are only some application scenarios of the method of the present invention. In actual applications, parameters and processes can be further adjusted according to workpiece size, material, and contaminant type. For example, the selection of scanning equipment can be adjusted according to workpiece size, the size of the grid division can be changed according to accuracy requirements, and the setting of laser parameters can be optimized according to material characteristics. Through this flexibility, the method of the present invention can play a role in various industrial cleaning scenarios, improving cleaning efficiency and quality. In one embodiment, the dynamic nature of laser parameter adjustment in step S3 is further explained. The adjustment of laser parameters can be based not only on contaminant thickness but also on the thermal sensitivity of the workpiece surface. For example, for materials with high thermal sensitivity, such as aluminum alloys or composite materials, the laser power and pulse width need to be further reduced to avoid heat accumulation leading to surface deformation or damage. When adjusting parameters, the thermal sensitivity level of the corresponding material can be queried through a preset material property database, and the parameter values ​​can be dynamically adjusted according to the level. This method can maximize the protection of the workpiece substrate and extend its service life while ensuring the cleaning effect. In another embodiment, taking the cleaning of electronic components as an example, the application of steps S1 to S3 on precision workpieces is explained. Assume that the workpiece is a circuit board bracket with welding residue and oil stains on the surface, and the thickness distribution is thin but complex. A small laser scanner is used to perform three-dimensional imaging of the bracket surface to obtain height information and reflectance spectrum data, generating a surface model. The model shows that there are multiple tiny protrusions and grooves on the bracket surface, and the contaminants are concentrated in the grooves. Then, the model is divided into multiple independent units with a grid size of 0.5 mm by 0.5 mm. The curvature change characteristics of each unit are calculated, and it is found that the curvature of the groove area is relatively high. Through K-means clustering, the statistical value of contaminant thickness is extracted and grouped into three groups: low, medium and high. The high-density group has a thickness of 0.3 mm, which exceeds the high-density thickness threshold of 0.2 mm. Starting with high-density cells, the A* algorithm is used to plan the cleaning path, incorporating curvature variation characteristics as cost weights to ensure the path covers all high-density areas. When adjusting laser parameters, for high-density areas, the power is set to a lower value, the pulse width is longer, and the scanning speed is controlled at 5 mm per second to avoid damage to precision components. The layered processing sequence is determined to be two layers, removing approximately 0.15 mm of thickness each time. During scanning, the laser equipment is guided by a high-precision robotic arm, processing cells one by one along the path, scanning layer by layer from the surface to the substrate. After completion, visual inspection is used to determine surface cleanliness and generate preliminary cleaning effect data.This method is adaptable to precision workpieces, ensuring a safe and reliable cleaning process. In one possible implementation, further explanation is provided regarding the precision control of 3D imaging and mesh generation in step S1. The precision of the scanning equipment directly affects the quality of the surface model. For example, for workpieces with micron-level precision, a higher-resolution scanner can be selected to ensure that the point cloud data can capture minute features. During mesh generation, the choice of mesh size needs to balance precision and computational load. For example, for large workpieces, the mesh can be appropriately enlarged to 10 mm x 10 mm to reduce data processing time; while for small precision workpieces, the mesh can be reduced to 0.1 mm x 0.1 mm to capture more detailed features. This precision control ensures the accuracy of subsequent analysis and cleaning operations. In step S4, if the preliminary cleaning effect data indicates residual contaminants, the A* algorithm is repeated to adjust the cleaning path.

[0040] This step involves replanning the cleaning path for residual contaminants to ensure that the secondary cleaning covers all problem areas, while optimizing the path to improve efficiency. The adjusted path will be more focused on areas with high residual density, avoiding repeated treatment of already cleaned areas. In one possible implementation, step S4 is further refined into specific operations.

[0041] Step S41: Using the remaining high-density areas in the preliminary cleaning effect data as new input, perform K-means clustering again.

[0042] Specifically, high-density residual areas refer to independent units where the residual thickness exceeds a threshold. The residual thickness statistics of these units are extracted as new input, and K-means clustering is performed again. The clustering process divides the residual pollutant thickness into three groups: low, medium, and high, with a focus on the distribution of the high-density group, providing a basis for pathway planning.

[0043] Optionally, this step also includes:

[0044] Step S411: Extract the statistical value of residual pollutant thickness as the input for new clustering and perform K-means clustering.

[0045] Specifically, the statistical value of residual contaminant thickness is obtained from the preliminary cleaning effect data. For each independent unit exceeding the threshold, its residual thickness value is recorded as a new input. The K-means clustering process iteratively calculates the distance from each unit to the cluster center, gradually optimizing the grouping results to ensure the similarity of the residual thickness distribution. After grouping, a new thickness grouping data table is generated, marking the location of high-density residual areas. In step S412, it is determined whether the new grouping exceeds the preset residual thickness threshold. If a new high-density area is displayed, a new starting node is generated.

[0046] Specifically, the new grouping results are compared with a preset residual thickness threshold. If the average thickness of a group exceeds the residual thickness threshold, it is identified as a new high-density region. The individual cells in these regions are selected as new starting nodes for subsequent path planning. The selection of starting nodes can prioritize the cells with the highest residual thickness, ensuring that cleaning resources are concentrated on the most difficult-to-process areas. In step S413, combining curvature change characteristics as the new path cost weight, the A* algorithm is used to obtain the adjusted cleaning path.

[0047] Specifically, the curvature variation characteristics, calculated in the preceding steps, serve as path cost weights to measure the difficulty of traversing each unit in path planning. When calculating a new path, the A* algorithm considers both path length and cost weights, prioritizing paths with lower costs. For example, for residual areas with significant curvature variations, the algorithm will try to plan smoother paths to avoid frequent angle adjustments by the laser equipment. The final adjusted cleaning path is generated, covering all new high-density areas. In one embodiment, taking metal sheet cleaning as an example, the specific application of steps S411 to S413 is illustrated. Assume the workpiece is a steel plate, and after initial cleaning, oil residue remains in the central area, reaching a thickness of 0.35 mm, exceeding the residual thickness threshold of 0.2 mm. The residual thickness statistics are extracted as new input, and K-means clustering is performed. The grouping results show that the central area is a high-density group. The average thickness of the new group exceeds the residual thickness threshold, confirming the central area as a new high-density area, and a new starting node is generated. Combining the curvature variation characteristics as path cost weights, it is found that the curvature variation in the central area is relatively small, so a straight path is prioritized during path planning, and the A* algorithm is used to obtain the adjusted cleaning path. This method can quickly locate key residual areas and optimize the secondary cleaning path. In one possible implementation, the versatility of path adjustment in step S4 is further explained. When planning the adjusted path, the A* algorithm can adjust the calculation method of the path cost weight according to the complexity of the workpiece surface. For example, for workpieces with large curvature changes, the weight of curvature features can be appropriately increased to ensure the path is as smooth as possible; while for flat workpieces, the weight of path length can be increased to prioritize the shortest path. This flexibility can adapt to different workpiece shapes and improve the adaptability of path planning. Step S42: The A* algorithm is used to plan the adjusted cleaning path based on the new thickness grouping results.

[0048] Specifically, the new thickness grouping results show the location and number of residual high-density areas. Using the independent units of these areas as new starting nodes, the cleaning path is replanned using the A* algorithm. During planning, path length and curvature variation characteristics are comprehensively considered to ensure that the path covers all residual high-density areas while minimizing equipment movement distance and turning frequency. Step S43: Repeat the layered processing sequence to obtain the optimized cleaning results.

[0049] Specifically, based on the adjusted cleaning path, the laser parameters and layered processing sequence are reset, and the residual contaminant area is scanned layer by layer. After scanning, the residual thickness is measured again to confirm whether the cleanliness requirements are met. If residues still exist, the above steps can be repeated until the residual thickness of all units is lower than the residual thickness threshold, and finally, the optimized cleaning result is obtained. In one embodiment, taking the cleaning of aerospace components as an example, the specific application of step S4 is explained. Assume the workpiece is an aircraft wing component, and after preliminary cleaning, there are still coating residues in the edge area, with a thickness of 0.4 mm, exceeding the residual thickness threshold of 0.3 mm. Using these high-density residual areas as new inputs, K-means clustering is performed again, and the grouping results show that the edge recessed areas are high-density groups. Then, according to the grouping results, the A* algorithm is used to plan the adjusted cleaning path, with the edge area unit as the starting node, combined with the curvature change characteristics, to ensure that the path covers all residual areas. The layered processing sequence is repeated, the laser parameters are adjusted, the power is set to medium to high, and the scanning speed is slowed to 10 mm per second, and the residual coating is removed layer by layer. Finally, the residual thickness is confirmed to be lower than the residual thickness threshold by reflectance spectroscopy detection, and the optimized cleaning result is obtained. This method effectively handles residual contaminants, ensuring the workpiece surface cleanliness meets requirements. In another embodiment, using a mechanical parts cleaning scenario, the diverse implementation of step S4 is further illustrated. Assume the workpiece is a bearing inner ring; after initial cleaning, some areas of the inner wall still have residual lubricating oil, reaching a thickness of 0.25 mm, exceeding the residual thickness threshold of 0.2 mm. Using these residual areas as new input, K-means clustering is re-performed, and the grouping results show that the inner wall grooves are a high-density group. Based on the grouping results, the A* algorithm is used to plan the adjusted cleaning path, using the groove area units as starting nodes and combining curvature change characteristics to ensure the path is smooth and covers all residual areas. The layered processing sequence is repeated, adjusting laser parameters, setting the power to medium, and controlling the scanning speed to 8 mm per second, removing residual oil contaminants layer by layer. Finally, surface cleanliness is confirmed through visual inspection, obtaining the optimized cleaning result. This method can adapt to complex geometries, ensuring precise and efficient secondary cleaning. In one possible implementation, the flexibility of the K-means clustering grouping in step S4 is further explained. When re-performing K-means clustering, the number of groups can be adjusted based on the distribution characteristics of residual pollutants. For example, if the residual areas are relatively concentrated, they can be divided into two groups, distinguishing only between low-density and high-density areas; while if the distribution is relatively dispersed, they can be divided into three or more groups to more finely classify the residual density. The adjustment of the number of groups can be determined by analyzing the distribution range of residual thickness data, ensuring that the grouping results are more targeted. This flexibility can adapt to different residual distributions and improve the efficiency of secondary path planning.

[0050] The above are only some preferred embodiments of the present invention, but the present invention is not limited thereto, and many improvements and modifications can be made. Any improvements and modifications made based on the basic principles of the present invention should be considered to fall within the protection scope of the present invention.

Claims

1. A laser cleaning method for contaminants on the surface of waste metal, characterized in that, include: S1. The surface of the workpiece is three-dimensionally imaged by a scanning device to obtain a surface model and divide it into independent units, and the curvature change characteristics of each independent unit are determined. S2, group the contaminant thickness of each independent unit and determine whether the contaminant thickness exceeds the preset thickness threshold. If the thickness grouping result shows a high-density area, the A* algorithm is used to plan the cleaning path to obtain a cleaning path for the high-density area. S3, adjust the laser parameters according to the cleaning path and distribution density to determine the layered processing sequence. Perform layer-by-layer scanning on each independent unit through the layered processing sequence to obtain preliminary cleaning effect data. S4, if the preliminary cleaning effect data indicates residual contaminants, repeat the A* algorithm to adjust the cleaning path to obtain the optimized cleaning result.

2. The method as described in claim 1, characterized in that, Step S1 includes: Step S11, the scanning device collects the height information and pollutant reflectance spectrum data of the workpiece surface to generate a surface model; Step S12, the complex shape is decomposed into multiple independent units according to the surface model using a meshing method; Step S13, for each independent unit, the curvature change characteristics are calculated and correlated with the pollutant distribution density to determine the curvature change characteristics of each independent unit.

3. The method as described in claim 2, characterized in that, Step S11 includes: scanning the workpiece surface point by point with a scanning device to obtain height data at each location and form point cloud data; using reflectance spectral analysis technology to collect the reflection characteristics of pollutants to specific wavelengths of light and to preliminarily determine the type and thickness distribution of pollutants; and generating a three-dimensional surface model from the point cloud data using a triangulation algorithm.

4. The method as described in claim 3, characterized in that, Step S13 includes: determining the curvature value by analyzing the changes in the surface normal of each independent unit.

5. The method as described in claim 1, characterized in that, Step S2 includes: Step S21, extracting the pollutant thickness statistics of each independent unit as clustering input, and performing K-means clustering grouping; Step S22, dividing into low, medium and high density groups according to the set thickness threshold of the grouping, and obtaining the thickness grouping result; Step S23, if the thickness grouping result shows a high density area, then taking the independent unit of the high density area as the starting node and planning the cleaning path.

6. The method as described in claim 5, characterized in that, Step S23 includes: Step S231, generating a cleaning path using curvature change features as path cost weights; Step S232, associating the cleaning path with each independent unit of the surface model to ensure that the path covers high-density areas; Step S233, outputting the cleaning path for high-density areas.

7. The method as described in claim 1, characterized in that, Step S3 includes: Step S31, dynamically setting the laser power pulse width and scanning speed according to the cleaning path and thickness grouping results, and adjusting the laser parameters; Step S32, determining the layered processing sequence based on the adjusted laser parameters and distribution density; Step S33, performing sequential layer-by-layer scanning from the surface of the contaminant to the base layer on each independent unit through the layered processing sequence to obtain preliminary cleaning effect data.

8. The method as described in claim 7, characterized in that, Step S31 includes: The laser parameters are adjusted based on the thickness and distribution density of the contaminants. For high-density areas, the laser power is increased and the pulse width is shortened; for low-density areas, the laser power is decreased and the pulse width is increased.

9. The method as described in claim 1, characterized in that, Step S4 includes: Step S41, using the remaining high-density areas in the preliminary cleaning effect data as new input, and re-grouping them using K-means clustering; Step S42, using the A* algorithm to plan the adjusted cleaning path based on the new thickness grouping results; Step S43, repeating the hierarchical processing sequence to obtain the optimized cleaning results.

10. The method as described in claim 9, characterized in that, Step S41 includes: Step S411, extracting the statistical value of residual pollutant thickness as the input for new clustering, and performing K-means clustering; Step S412, determining whether the new grouping exceeds the preset residual thickness threshold, and if a new high-density area is displayed, generating a new starting node; Step S413, combining the curvature change feature as the new path cost weight, and using the A* algorithm to obtain the adjusted cleaning path.

Images