A complex curved surface laser cleaning system based on dynamic focusing and machine vision

Abstract

The application discloses a kind of complex curved surface laser cleaning systems based on dynamic focusing and machine vision, including industrial computer and machine vision module, dynamic focusing module and cleaning light path module connected with industrial computer;Workpiece to be cleaned is placed on the object table, the machine vision module and cleaning light path module are located above workpiece;Machine vision module is used to construct the three-dimensional profile of complex curved surface on workpiece, to obtain the height information of each point of complex curved surface in real time, and data information is transmitted to dynamic focusing module;The dynamic focusing module controls cleaning light path module according to the received data information, dynamic focusing is carried out on the complex curved surface of workpiece, so that laser beam is focused to the corresponding position of complex curved surface to be cleaned, to realize accurate cleaning of complex curved surface of workpiece.The application has the characteristics of compact structure, convenient operation, high reliability, realizes all-round, blind area cleaning to complex curved surface with height difference, and significantly improves cleaning quality.

Description

Technical Field

[0001] This invention relates to the field of laser cleaning technology, specifically to a complex curved surface laser cleaning system based on dynamic focusing and machine vision, which can perform efficient and precise cleaning on complex curved surfaces with height differences. Background Technology

[0002] In industrial production, many parts have complex curved surface structures, and there are certain height differences between the curved surfaces, such as automobile engine blocks and aerospace engine blades. The cleaning quality of these parts directly affects their performance and lifespan.

[0003] When cleaning complex curved surfaces with height differences or abrupt contour changes, maintaining the laser focus on the workpiece surface is crucial for ensuring cleaning quality. Existing technologies, such as the Chinese patent application CN116586770A, disclose a scheme using a fixed-focal-length field lens combined with galvanometer planar scanning and XY mechanical displacement. This relies on a pre-processing 3D scanning reconstruction model to plan the mechanical motion path; the optical path itself lacks dynamic focal length adjustment capability, and Z-axis height adaptation depends entirely on adjusting the overall distance of the processing head via mechanical structure. This method, essentially mechanical path adaptation plus a fixed optical focal length, can only adapt to gently curved surfaces within the focal depth range. When the height difference of the surface exceeds the focal depth threshold, local defocusing and a sharp drop in energy density can easily occur, leading to incomplete cleaning or local overheating. Furthermore, the mechanical transmission method has high inertia and slow response speed, making it difficult to adapt to the high-speed, precise focusing requirements of micro-scale height differences and abrupt contour changes. Offline modeling cannot compensate for deviations caused by clamping errors or workpiece thermal deformation in real time, limiting cleaning efficiency and the threshold for process implementation. Therefore, there is an urgent need for a technical solution that can solve the problem of focusing on complex curved surfaces in real time from the underlying optical path. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to address the shortcomings of traditional cleaning methods when dealing with complex curved surfaces with height differences, such as incomplete cleaning, low efficiency, and inability to dynamically focus. The present invention provides a compact, easy-to-operate, stable and reliable laser cleaning system for complex curved surfaces based on dynamic focusing and machine vision.

[0005] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A laser cleaning system for complex curved surfaces based on dynamic focusing and machine vision includes an industrial control computer and a machine vision module, a dynamic focusing module, and a cleaning optical path module connected to the industrial control computer. The workpiece to be cleaned is placed on a stage, and the machine vision module and the cleaning optical path module are both located above the workpiece. The machine vision module is used to construct the three-dimensional contour of the complex curved surface on the workpiece to obtain the height information of each point on the complex curved surface in real time and transmit the data information to the dynamic focusing module. The dynamic focusing module controls the cleaning optical path module to dynamically focus on the complex curved surface of the workpiece according to the received data information, so that the laser beam is focused on the corresponding position of the complex curved surface to be cleaned, thereby achieving precise cleaning of the complex curved surface of the workpiece.

[0006] As a further improvement of the present invention, the machine vision module includes an industrial camera, a laser scanner, and an image processor connected to an industrial control computer; the industrial camera is used to acquire two-dimensional images of complex curved surfaces on a workpiece, the laser scanner emits a laser beam to scan the complex curved surfaces of the workpiece to obtain depth information of each point on the complex curved surfaces, and the image processor performs fusion processing on the two-dimensional images acquired by the industrial camera and the depth information acquired by the laser scanner to construct a three-dimensional contour of the complex curved surfaces on the workpiece.

[0007] As a further improvement of the present invention, the dynamic focusing module includes a focusing drive device and a controller; the controller is connected to an industrial computer and receives complex curved surface three-dimensional contour data transmitted by the machine vision module, analyzes and calculates to determine the position parameters that the cleaning optical path needs to focus at each moment, and sends a control signal to the focusing drive device, which is used to drive the cleaning optical path module to dynamically focus on the complex curved surface of the workpiece.

[0008] As a further improvement of the present invention, the focusing drive device employs a piezoelectric ceramic actuator.

[0009] As a further improvement of the present invention, the cleaning optical path module includes a laser, a reflector group and a focusing lens arranged in sequence; the laser is connected to an industrial control computer to emit a laser beam for cleaning, and the laser beam is reflected by the reflector group and guided to the focusing lens; the focusing lens is connected to a focusing drive device, and under the drive of the focusing drive device, the focusing lens focuses the laser beam to the current corresponding position of the complex curved surface to be cleaned, so as to achieve precise cleaning of the complex curved surface of the workpiece.

[0010] As a further improvement of the present invention, the reflector group adopts an adjustable reflector to assist in adjusting the direction of the optical path and realize the laser beam covering the complex curved surface to be cleaned.

[0011] As a further improvement of the present invention, a beam combiner is provided between the laser and the reflector assembly.

[0012] As a further improvement of the present invention, the industrial camera is a high-definition camera.

[0013] As a further improvement of the present invention, the image processor and controller are integrated into an industrial computer.

[0014] Compared with existing technologies, the complex curved surface laser cleaning system based on dynamic focusing and machine vision of the present invention has the following significant advantages: 1. By dynamically focusing and compensating for height differences in real time, the focus is ensured to fall precisely on the workpiece surface within the entire curved surface range, ensuring consistent cleaning energy density and fundamentally solving the problems of uneven cleaning quality and blind spots on complex curved surfaces with high elevation differences.

[0015] 2. The optical path of this invention is a multi-degree-of-freedom dynamically adjustable structure, consisting of an adjustable reflector group and a piezoelectric ceramic driven focusing lens. The reflector angle can be dynamically adjusted to expand the coverage of curved surfaces, and the focusing lens can reciprocate at high speed along the optical axis to adapt to height differences in real time. The optical path itself has dual-dimensional adjustment capabilities in focal length and direction, belonging to a dynamic optical path architecture. By acquiring height data in real time and dynamically adjusting the focus, it can adaptively compensate for height deviations caused by clamping errors and workpiece deformation, eliminating the need for high-precision tooling positioning and significantly reducing the threshold for process implementation.

[0016] 3. Optical dynamic focusing adjustment is achieved through piezoelectric ceramic actuators, eliminating mechanical motion inertia, with a response speed in the microsecond range and adjustment accuracy down to the submicron level. It can adapt to high-speed and precise focusing on curved surfaces with small local height differences and steeply changing contours. Height adaptation is completed simultaneously during high-speed scanning. The adjustment accuracy and response speed are 1 to 2 orders of magnitude higher than mechanical solutions, and the cleaning efficiency is significantly improved compared to mechanical adjustment solutions.

[0017] 4. By integrating the image processor and controller into the same industrial computer, a closed-loop system is achieved for visual data processing, focusing algorithm calculation, and drive signal output, significantly shortening the control chain and improving response speed and synchronization accuracy. Real-time closed-loop focusing control ensures high consistency in laser spot diameter and power density at any location on complex curved surfaces, avoiding the problem of excessive energy at high points and insufficient energy at low points in traditional solutions. This significantly improves cleaning uniformity and cleanliness consistency, making it particularly suitable for precision cleaning of components in high-reliability applications such as aerospace and rail transportation. Attached Figure Description

[0018] Figure 1 This is a schematic diagram illustrating the structural principle of a complex curved surface laser cleaning system based on dynamic focusing and machine vision in a specific embodiment of the present invention. Figure 2 This is a schematic diagram of the optical path of the cleaning optical path module in a specific embodiment of the present invention.

[0019] Legend: 1. Industrial computer; 2. Industrial camera; 3. Laser scanner; 4. Stage; 5. Laser; 6. Image processor; 7. Laser; 8. Beam combiner; 9. Focusing lens; 10. Workpiece; 11. Focusing drive device; 12. Reflector group; 13. Controller. Detailed Implementation

[0020] The present invention will be further described below with reference to the accompanying drawings and specific preferred embodiments, but this does not limit the scope of protection of the present invention.

[0021] In the description of this invention, it should be understood that the terms "side", "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", "axial", "radial", "circumferential", etc., indicating the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, are only for the convenience of describing this invention and simplifying the description, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this invention.

[0022] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more unless otherwise explicitly specified.

[0023] Example like Figure 1 As shown, the complex curved surface laser cleaning system based on dynamic focusing and machine vision of the present invention includes an industrial control computer 1 and a machine vision module, a dynamic focusing module, and a cleaning optical path module connected to the industrial control computer 1. In specific implementation, the industrial control computer 1, as the central processing unit of the system, is responsible for coordinating data interaction and command issuance between various modules. It integrates image processing algorithms and motion control logic to ensure the synchronization and accuracy of the entire cleaning process. It should be understood that although... Figure 1 The diagram shows the separate configuration of each module. However, in other implementations, the machine vision module and the cleaning optical path module can be integrated into a single processing head, or the functions of the industrial computer 1 can be distributed and embedded into the controllers of each sub-module, as long as overall collaborative control can be achieved.

[0024] The workpiece 10 to be cleaned is placed on the stage 4. Both the machine vision module and the cleaning optical path module are located above the workpiece 10, which facilitates vertical or near-vertical observation and processing of the workpiece 10 surface, reducing measurement errors or energy loss caused by excessive tilt angles. The stage 4 can be configured as a fixed platform, a rotary table, or a multi-axis moving platform according to actual needs to accommodate workpieces 10 of different sizes and shapes for cleaning operations. In this embodiment, the stage 4 primarily provides stable support and reference positioning for the workpiece 10, while adaptation to changes in surface height is mainly accomplished by the upper optical system, thereby reducing reliance on the dynamic performance of the mechanical motion mechanism.

[0025] The machine vision module is used to construct the three-dimensional contour of the complex curved surface on the workpiece 10, so as to acquire the height information of each point on the complex curved surface in real time and transmit the data information to the dynamic focusing module. Specifically, the "real-time acquisition" in this embodiment means that the data acquisition frequency of the machine vision module is matched with the scanning speed of the cleaning optical path, so that it can continuously and dynamically perceive the morphological changes of the workpiece 10 surface during the cleaning process, rather than performing a one-time offline modeling before cleaning. For example, when the cleaning optical path moves to a certain area of ​​the workpiece 10, the machine vision module can immediately provide feedback on the current actual height value of that area. Even if the workpiece 10 has unpredictable deformations such as clamping deformation or thermal expansion, the system can compensate through real-time perception. The online perception mechanism forms the data foundation for the closed-loop control of the system, effectively overcoming the processing deviation caused by the discrepancy between the model and reality in the traditional open-loop scheme.

[0026] The dynamic focusing module, based on received data, controls the cleaning optical path module to dynamically focus on the complex curved surface of workpiece 10, ensuring the laser beam is focused on the corresponding position on the surface to be cleaned, thus achieving precise cleaning of the complex curved surface of workpiece 10. In this process, the dynamic focusing module acts as a switching hub between perception and execution. It calculates in real-time the discrete or continuous height data output by the machine vision module into the focal length adjustment or position offset required by the cleaning optical path module, and drives the optical path components to make corresponding physical responses. This ensures that regardless of changes in the curvature of the workpiece 10 surface, the laser focus always falls precisely on the surface to be cleaned, maintaining a constant spot size and energy density. Compared to the traditional method of relying on mechanical Z-axis overall lifting to adapt to height changes, the dynamic focusing method used in this embodiment has higher response bandwidth and adjustment accuracy, fundamentally solving the defocusing failure problem during cleaning of large height differences and steep contours from an optical perspective, ensuring consistent cleaning quality across the entire curved surface.

[0027] like Figure 1As shown, the machine vision module includes an industrial camera 2, a laser scanner 3, and an image processor 6 connected to an industrial control computer 1. Specifically, the industrial camera 2 is used to acquire two-dimensional images of the complex curved surface on the workpiece 10, and the laser scanner 3 emits a laser beam to scan the complex curved surface of the workpiece 10 to obtain depth information of each point on the complex curved surface. In actual operation, the data acquisition of the industrial camera 2 and the laser scanner 3 is usually synchronous or time-series correlated. The laser scanner 3 can accurately obtain the absolute distance of the curved surface points in the optical axis direction through principles such as triangulation or time-of-flight method, but the point cloud data it generates often lacks surface texture, color, and edge features, and is prone to data gaps when facing highly reflective or light-absorbing materials. Conversely, the industrial camera 2 can capture rich surface texture and grayscale gradient information, which helps to identify dirt boundaries, scratch locations, and changes in surface material, but it is difficult to directly obtain high-precision absolute depth values ​​with monocular or multi-view vision alone, especially when the curvature of the curved surface changes drastically, which can easily produce distortion errors. Therefore, this embodiment ensures the completeness of the perceived information from a physical level through complementary acquisition of dual-source data.

[0028] Based on this, the image processor 6 fuses the two-dimensional image acquired by the industrial camera 2 and the depth information acquired by the laser scanner 3 to construct a three-dimensional contour of the complex surface on the workpiece 10. In this embodiment, the fusion process is not a simple superposition, but rather a high-precision registration and mapping of the pixel coordinates of the two-dimensional image with the coordinates of the three-dimensional laser point cloud based on a pre-calibrated extrinsic parameter matrix. For example, the laser point cloud can be used as a geometric skeleton, and edge features in the two-dimensional image can be used to interpolate and encrypt the point cloud or filter out noise; or image texture information can be used to fill in missing data in the shadow area or specular reflection area of ​​the laser scan. Through this deep fusion, the three-dimensional contour constructed by the system not only contains accurate spatial height information but also retains the semantic features of the surface, enabling the dynamic focusing module to not only know where the surface is but also understand its local characteristics, thereby avoiding focusing errors caused by misjudging edges or noise. It should be understood that the fusion algorithm can adopt conventional settings in the art, such as feature-based registration, intensity-based fusion, or deep learning-driven end-to-end reconstruction, as long as it can effectively integrate two-dimensional and three-dimensional data.

[0029] Furthermore, to improve the quality of the data source, the industrial camera 2 employs a high-definition camera. In laser cleaning scenarios, high definition is not merely a resolution indicator, but a crucial physical prerequisite for ensuring the accuracy of identifying minute height differences. Specifically, higher pixel density means that, under the same field of view, the physical size corresponding to each pixel is smaller, i.e., a higher spatial sampling rate. When there are micron-level stains, fine cracks, or gradually changing height transition areas on the surface of the workpiece 10, ordinary resolution cameras may fail to distinguish these features due to pixel aliasing, causing the 3D reconstruction results to become smooth or distorted in these areas, thus misleading the dynamic focusing module to produce defocus. High-definition cameras can clearly capture the texture gradients of these fine structures, and, combined with the sub-pixel-level edge extraction algorithm of the image processor 6, can significantly improve the restoration accuracy of the 3D contour in local details. For example, for areas with abrupt curvature changes such as the leading and trailing edges of aero-engine blades, high-definition imaging can accurately delineate the true contour curve, rather than a fitted approximate arc. This directly determines whether the piezoelectric ceramic actuator can make the correct focus compensation response within milliseconds. Of course, the specific selection of a high-definition camera can be adapted according to the size of the workpiece 10, the cleaning accuracy requirements, and the data processing bandwidth, and is not limited to a specific resolution value or sensor type.

[0030] like Figure 1As shown, the dynamic focusing module includes a focusing drive device 11 and a controller 13. The controller 13 is connected to the industrial computer 1 and receives complex curved surface 3D contour data transmitted by the machine vision module. It analyzes and calculates the position parameters that the cleaning optical path needs to focus on at each moment, and sends control signals to the focusing drive device 11. The focusing drive device 11 is used to drive the cleaning optical path module to dynamically focus on the complex curved surface of the workpiece 10. Specifically, the controller 13 runs a real-time focusing calculation algorithm. This algorithm converts the discrete or continuous height coordinate values ​​output by the image processor 6 into the physical displacement or voltage drive value required by the focusing drive device 11 through a preset optical mapping model. For example, when the machine vision module detects that the height change of the workpiece 10 surface at a certain moment is Δh, the controller 13 does not simply equate it to the lens movement. Instead, it combines the focal length of the current focusing lens 9, the optical path magnification, and the system calibration coefficient to calculate the precise compensation amount and generate the corresponding analog voltage or digital pulse signal. By employing a real-time conversion mechanism from "geometric perception" to "optical execution," the cleaning optical path is ensured to follow changes in surface topography at millisecond or even higher frequencies, thereby maintaining dynamic stability of the focal position during continuous scanning. It should be understood that although the controller 13 is described as a standalone component in this embodiment, in other implementations, its function can be integrated into the motherboard of the industrial computer 1 or embedded in the internal drive circuit of the focusing drive device 11, as long as real-time calculation and output of the three-dimensional contour data to the focusing control signal can be achieved. Alternatively, the image processor 6 and the controller 13 can be integrated into the industrial computer 1 to achieve an integrated closed loop of visual data processing, focusing algorithm calculation, and drive signal output, significantly shortening the control link and improving response speed and synchronization accuracy.

[0031] Furthermore, to meet the stringent dynamic performance requirements of high-speed cleaning of complex curved surfaces, the focusing drive device 11 employs a piezoelectric ceramic actuator. Compared to solutions using stepper motors or servo motors with ball screws, the piezoelectric ceramic actuator eliminates mechanical backlash, friction dead zones, and hysteresis errors. Its positioning resolution can reach sub-micron or even nanometer levels, with a response time on the order of microseconds. When the workpiece 10 surface exhibits steep edges, deep grooves, or areas of abrupt curvature changes, the focusing drive device 11 can complete significant focus adjustments in an extremely short time without overshooting or oscillation due to mechanical inertia, nor will the focus lag behind the scanning spot due to transmission delay. In contrast, while voice coil motors can also achieve a relatively fast response, they typically require complex closed-loop feedback and PID tuning to suppress jitter, and are prone to overheating and affecting accuracy under high-frequency reciprocating motion. The piezoelectric ceramic actuator, on the other hand, inherently possesses self-locking characteristics and extremely high rigidity, maintaining its position even when powered off, and exhibits superior thermal stability under high-frequency dynamic conditions. Therefore, using a piezoelectric ceramic actuator as the focusing drive device 11 is not only a technical means to achieve high-precision focusing, but also a key physical basis for solving the problem of cleaning failures caused by large height differences and steeply changing contours. The resulting microsecond-level response bandwidth and sub-micron-level repeatability accuracy enable the system to significantly improve scanning speed while ensuring cleaning quality, achieving a dual breakthrough in efficiency and accuracy. Of course, without departing from the inventive concept of this application, if the application scenario has lower requirements for response speed but is more sensitive to cost, alternative solutions such as high-precision linear motors or voice coil motors can be considered. However, the piezoelectric ceramic actuator remains the preferred implementation method for dealing with the cleaning needs of extremely complex curved surfaces.

[0032] In this embodiment, the cleaning optical path module includes a laser 7, a mirror group 12, and a focusing lens 9 arranged sequentially. For example... Figure 2 As shown, laser 7 is connected to industrial computer 1 to emit a laser beam for cleaning. The laser beam is reflected by mirror group 12 and guided to focusing lens 9. In this optical path architecture, laser 7 acts as the energy source, and its output parameters (such as power, pulse width, and frequency) are controlled in real-time by industrial computer 1 to adapt to cleaning thresholds for different materials and contaminants. Mirror group 12 performs beam transmission and path folding functions, enabling the entire optical path system to achieve a longer effective focal length or a specific incident angle within a limited space, optimizing the compactness of the processing head. Focusing lens 9 acts as the terminal actuator of the optical path, focusing the collimated or divergent high-energy laser beam into a micron-sized high-energy-density spot, directly acting on the surface of workpiece 10. It should be understood that although... Figure 2This application demonstrates a specific optical path arrangement. However, in other embodiments, the relative positions, number, and types of the optical elements can be adjusted according to actual working conditions. For example, a beam expander can be added to adjust the spot size, or a protective lens can be introduced to prevent splash contamination of the core lens. As long as the basic functional link of "light source → transmission → focusing" is maintained, it falls within the protection scope of this application.

[0033] To decouple the optical path from mechanical motion and improve dynamic response performance, the focusing lens 9 is connected to the focusing drive device 11. Driven by the focusing drive device 11, the focusing lens 9 focuses the laser beam to the corresponding position of the complex curved surface to be cleaned, thereby achieving precise cleaning of the complex curved surface of the workpiece 10. Here, the focusing lens 9 is designed as the only axially moving component in the optical path, and its mass is typically much smaller than the entire processing head or mechanical Z-axis module. This lightweight design allows the focusing drive device 11 to achieve high-frequency reciprocating motion with extremely low inertial load, thus completing dynamic compensation of the focal length within milliseconds. Compared to the traditional approach of moving the entire optical path system to adapt to height changes, this embodiment significantly reduces the risk of mechanical resonance by moving only the end lens, ensuring the stability of the focal position and the consistency of the spot shape during high-speed scanning cleaning. This is the physical basis for achieving high-quality cleaning of complex curved surfaces.

[0034] Furthermore, to address potential sidewalls, grooves, or large-angle areas on complex curved surfaces, the reflector group 12 employs adjustable reflectors to assist in adjusting the optical path direction, ensuring the laser beam covers the complex curved surface to be cleaned. Specifically, in this embodiment, "adjustable" means that the angle or position of the reflector is not fixed but can be dynamically deflected according to a preset program or real-time feedback signal. For example, the reflector group 12 can specifically employ a two-dimensional galvanometer scanning system, a MEMS micromirror array, or a rotating reflector driven by a precision servo motor. When there are obstructed areas on the surface of the workpiece 10 that cannot be vertically illuminated directly below the focusing lens 9, the adjustable reflector can change the beam's exit angle, allowing the laser beam to enter these areas at an angle. Combined with the Z-axis displacement of the focusing lens 9, this creates a two-dimensional spatial adaptation capability of "angle + focal length." This effectively compensates for the limitations of single-axis focusing in geometric coverage, ensuring that laser energy can be uniformly projected to every corner of the complex curved surface, fundamentally eliminating cleaning blind spots. It should be understood that the specific selection of the adjustable reflector depends on the geometric complexity of the workpiece 10 and the cleaning cycle requirements. For workpieces with gentle curvature changes, a low-frequency response motor-driven reflector can be used to reduce costs; while for precision parts with deep grooves or steep changes, a high-frequency response galvanometer system is preferred. This application does not limit this.

[0035] Furthermore, to improve system integration and operational accuracy, a beam combiner 8 is installed between the laser 7 and the reflector group 12. The beam combiner 8 typically employs a dichroic mirror or a beam splitter prism, its core function being to achieve coaxial transmission of multi-wavelength beams. Specifically, the beam combiner 8 can combine the high-power cleaning laser emitted by the laser 7 with the low-power visible light indicator laser (such as red or green light) into a single optical path. This allows the operator to accurately predict the actual landing point of the cleaning laser through the indicator spot observed with the naked eye, achieving "what you see is what you get" visual alignment and greatly simplifying the clamping and positioning of the workpiece 10 and the first-piece debugging process. It is understood that the beam splitting ratio, coating band, and installation angle of the beam combiner 8 can be customized according to the actual integrated functional modules.

[0036] In this embodiment, when cleaning a complex curved surface workpiece 10 with height differences, the industrial camera 2 and laser scanner 3 in the machine vision module are first activated. The industrial camera 2 acquires a two-dimensional image of the complex curved surface of the workpiece 10, and the laser scanner 3 emits a laser beam to scan the curved surface and obtain depth information at each point. The image processor 6 fuses the two-dimensional image and depth information to construct the three-dimensional contour of the complex curved surface and transmits this data to the controller 13 of the dynamic focusing module in real time.

[0037] The controller 13 analyzes the three-dimensional contour of the complex curved surface and calculates the position parameters that the laser beam needs to be focused at each moment during the cleaning process, as the workpiece 10 or the cleaning device moves. Subsequently, the controller 13 sends a control signal to the focusing drive device 11, which drives the focusing lens 9 to make corresponding position adjustments, such as... Figure 2 The three different height positions are a, b, and c.

[0038] Simultaneously, the laser 7 in the cleaning optical path module emits a laser beam, which is reflected by the reflector 12 and then directed towards the focusing lens 9. Under the action of the focusing drive device 11, the focusing lens 9 precisely focuses the laser beam onto the current position of the surface to be cleaned, performing precise cleaning. During the cleaning process, the reflector 12 adjusts the reflection angle appropriately according to a preset program or auxiliary control signals from the controller 13, ensuring that the laser beam can cover the entire complex curved surface of the workpiece 10, achieving comprehensive cleaning.

[0039] As the cleaning process proceeds, the machine vision module continuously acquires the three-dimensional contour data of the complex curved surface of the workpiece 10, and the dynamic focusing module continuously adjusts the focusing lens 9 to keep the laser beam focused on the curved surface of the workpiece 10 to be cleaned until the cleaning of the entire complex curved surface is completed.

[0040] This invention relates to a complex curved surface laser cleaning system based on dynamic focusing and machine vision. The system uses a machine vision module to collect surface height data in real time and directly use it as the input signal for dynamic focusing, constructing a fully real-time closed-loop control architecture of "real-time visual acquisition → 3D contour calculation → focusing parameter calculation → dynamic optical path adjustment." Compared to existing open-loop or semi-closed-loop solutions that rely on offline modeling and mechanical Z-axis adjustment, this invention achieves dynamic adaptation of the optical focal length from the optical path level. It can compensate for surface height differences in real time without relying on macroscopic displacement of mechanical structures, fundamentally solving the defocusing failure problem during cleaning of complex curved surfaces with large height differences and steep changes. This ensures the consistency of laser energy density and the uniformity of cleaning quality across the entire curved surface.

[0041] Meanwhile, the dynamic focusing module of this invention uses a piezoelectric ceramic actuator as the core execution element, leveraging its microsecond-level response speed and sub-micron-level positioning accuracy to overcome the shortcomings of traditional mechanical transmissions, such as large inertia and lag. This design enables the system to simultaneously complete instantaneous focal length compensation during high-speed scanning and cleaning, significantly improving cleaning efficiency and adaptively offsetting height deviations caused by workpiece clamping errors, thermal deformation, and uneven contaminant thickness. This reduces reliance on high-precision tooling positioning and significantly improves process robustness and adaptability. Furthermore, through the synergistic effect of the adjustable reflector group and the dynamic focusing lens, real-time adjustment of both the optical path direction and focal length is achieved, ensuring precise coverage and focusing of the laser beam at any position on complex curved surfaces. This is particularly suitable for the precision cleaning needs of high-reliability components in aerospace, rail transportation, and other fields.

[0042] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A complex curved surface laser cleaning system based on dynamic focusing and machine vision, characterized in that, The system includes an industrial computer (1) and a machine vision module, a dynamic focusing module, and a cleaning optical path module connected to the industrial computer (1); the workpiece (10) to be cleaned is placed on a stage (4), and the machine vision module and the cleaning optical path module are both located above the workpiece (10); the machine vision module is used to construct the three-dimensional contour of the complex surface on the workpiece (10) to obtain the height information of each point on the complex surface in real time and transmit the data information to the dynamic focusing module; the dynamic focusing module controls the cleaning optical path module to dynamically focus on the complex surface of the workpiece (10) according to the received data information, so that the laser beam is focused on the corresponding position of the complex surface to be cleaned, so as to achieve precise cleaning of the complex surface of the workpiece (10).

2. The complex curved surface laser cleaning system based on dynamic focusing and machine vision according to claim 1, characterized in that, The machine vision module includes an industrial camera (2), a laser scanner (3), and an image processor (6) connected to an industrial computer (1). The industrial camera (2) is used to acquire two-dimensional images of complex curved surfaces on the workpiece (10). The laser scanner (3) emits a laser beam to scan the complex curved surfaces of the workpiece (10) to obtain depth information of each point on the complex curved surfaces. The image processor (6) performs fusion processing on the two-dimensional images acquired by the industrial camera (2) and the depth information acquired by the laser scanner (3) to construct the three-dimensional contour of the complex curved surfaces on the workpiece (10).

3. The complex curved surface laser cleaning system based on dynamic focusing and machine vision according to claim 2, characterized in that, The dynamic focusing module includes a focusing drive device (11) and a controller (13); the controller (13) is connected to the industrial computer (1) and receives complex surface three-dimensional contour data transmitted by the machine vision module, analyzes and calculates to determine the position parameters that the cleaning optical path needs to focus at each moment, and sends the control signal to the focusing drive device (11). The focusing drive device (11) is used to drive the cleaning optical path module to dynamically focus on the complex surface of the workpiece (10).

4. The complex curved surface laser cleaning system based on dynamic focusing and machine vision according to claim 3, characterized in that, The focusing drive device (11) employs a piezoelectric ceramic actuator.

5. The complex curved surface laser cleaning system based on dynamic focusing and machine vision according to claim 4, characterized in that, The cleaning optical path module includes a laser (7), a reflector group (12), and a focusing lens (9) arranged in sequence. The laser (7) is connected to the industrial control computer (1) to emit a laser beam for cleaning. The laser beam is reflected by the reflector group (12) and guided to the focusing lens (9). The focusing lens (9) is connected to the focusing drive device (11). Under the drive of the focusing drive device (11), the focusing lens (9) focuses the laser beam to the current corresponding position of the complex curved surface to be cleaned, so as to achieve precise cleaning of the complex curved surface of the workpiece (10).

6. The complex curved surface laser cleaning system based on dynamic focusing and machine vision according to claim 5, characterized in that, The reflector group (12) uses adjustable reflectors to assist in adjusting the direction of the optical path, so as to achieve laser beam coverage of the complex curved surface to be cleaned.

7. The complex curved surface laser cleaning system based on dynamic focusing and machine vision according to claim 5, characterized in that, A beam combiner (8) is provided between the laser (7) and the reflector group (12).

8. The complex curved surface laser cleaning system based on dynamic focusing and machine vision according to any one of claims 2 to 7, characterized in that, The industrial camera (2) is a high-definition camera.

9. The complex curved surface laser cleaning system based on dynamic focusing and machine vision according to any one of claims 3 to 7, characterized in that, The image processor (6) and controller (13) are integrated into the industrial computer (1).

Images