Real-time optical inspection and processing system and method for mirrors in position

Abstract

The present disclosure relates to a mirror in-situ real-time optical detection and processing system and method, belonging to the field of mirror processing. The system comprises a processing module for processing a mirror according to a preset processing track; an optical detection module for collecting optical data of the mirror when the mirror is in a processing position, and for removing a polishing liquid film in a mirror detection area when the optical data is collected; an edge processing module for real-time analyzing surface error information of the mirror based on the optical data; and a closed-loop feedback control module for real-time converting the surface error information into processing module residence time compensation information, and real-time correcting the preset processing track according to the processing module residence time compensation information. The present disclosure can penetrate the interference of polishing medium to realize real-time in-situ detection, realize the synchronization of processing and detection, and improve the processing efficiency and the accuracy of the processed surface.

Description

Technical Field

[0001] This disclosure relates to the field of mirror processing, and in particular to a system and method for in-situ real-time optical inspection and processing of mirrors. Background Technology

[0002] In the grinding and polishing process of the optical surface of a mirror, the detection of surface accuracy is the core step.

[0003] The current mainstream mirror processing method adopts a cyclical model of "processing-shutdown-cleaning-transfer-environmental stabilization-de-inspection". That is: process on a polishing machine for a period of time; stop the machine and thoroughly clean the polishing fluid (grinding wheel fluid) from the mirror surface; move the mirror blank to an inspection tower or use an interferometer for inspection; for large-diameter mirrors, the mirror blank can weigh hundreds of kilograms or even several tons; wait for the mirror blank to reach thermal equilibrium with the environment (usually several hours); adjust the processing parameters based on the inspection results, and reassemble for the next round of processing. This method has the following disadvantages:

[0004] The auxiliary time is extremely long: the time for moving (i.e., handling), cleaning and thermal balancing far exceeds the actual processing time, resulting in low production efficiency, especially for large-diameter mirrors, where the cycle is often measured in years;

[0005] Repeated clamping error: Each time the machine is moved and re-adjusted, mechanical stress and positioning error are introduced, which disrupts the consistency of the machining coordinate system;

[0006] Unable to provide real-time feedback: This is a form of "post-processing compensation" and cannot be dynamically adjusted based on transient thermal deformation during the manufacturing process.

[0007] Therefore, there is an urgent need to design a system and method that can achieve real-time in-situ detection without machine downtime or handling, and can penetrate the interference of polishing media, thereby synchronizing processing and detection, significantly shortening the processing cycle and improving surface accuracy. Summary of the Invention

[0008] Therefore, it is necessary to provide an in-situ real-time optical inspection and processing system and method for mirrors to address the problem of low processing efficiency and processing effect.

[0009] To solve the above problems, the present disclosure adopts the following technical solution:

[0010] In a first aspect, this disclosure provides an in-situ real-time optical inspection and processing system for a mirror, including:

[0011] The processing module is used to process the reflector according to a preset processing trajectory;

[0012] An optical inspection module is used to acquire optical data of the reflector when it is in the processing position, and to remove the polishing liquid film in the inspection area of ​​the reflector when acquiring the optical data.

[0013] An edge processing module is used to acquire the optical data in real time and analyze the surface error information of the reflector in real time based on the optical data. The surface error is the difference between the real-time surface data and the ideal surface data.

[0014] The closed-loop feedback control module is used to acquire surface error information in real time, convert the surface error information into processing module dwell time compensation information in real time, and correct the preset processing trajectory in real time based on the processing module dwell time compensation information.

[0015] In a preferred embodiment, the optical detection module includes an optical data acquisition device and a liquid displacement device. The optical data acquisition device is used to acquire optical data of the reflector, and the liquid displacement device is used to displace the polishing liquid film in the detection area of ​​the reflector when the optical data acquisition device acquires the optical data.

[0016] In a preferred embodiment, the optical data acquisition device is a range sensor or a laser scanner, or the optical data acquisition device is a stripe projector and a camera.

[0017] In a preferred embodiment, the liquid discharge device is an air curtain nozzle, a rotary centrifuge device, or an ultrasonic discharge head.

[0018] In a preferred embodiment, the optical data is image data; the input of the edge processing module includes the spindle coordinates of the machine tool and the optical data. The real-time analysis of the reflector's surface error information based on the optical data specifically involves: real-time multispectral narrowband filtering and information decoupling processing of the optical data to obtain phase perturbation components; real-time phase extraction using Fourier transform; real-time calculation of the influence of fluid refractive index changes on the optical path based on the spindle coordinates of the machine tool and phase compensation correction to obtain phase compensation results; real-time unwrapping of the phase compensation results to restore the true continuous phase distribution; real-time construction of the spatiotemporal distribution of fluid motion characteristics in the processing area; and real-time obtaining surface error information based on the fluid motion characteristics. The real-time conversion of the surface error information into processing module dwell time compensation information specifically involves real-time processing of the surface error information to obtain a 3D surface residual cloud map and a dwell time deviation vector, with the dwell time deviation vector serving as the processing module dwell time compensation information.

[0019] In a preferred embodiment, the dwell time compensation information of the processing module is obtained through a deep reinforcement learning method. The input vector corresponding to the deep reinforcement learning method is: current surface residual, processing head speed, grinding pressure, real-time temperature, and track position; the corresponding action space is: dwell time correction, feed rate, and coolant flow rate; and the corresponding reward function is: surface convergence speed and surface roughness smoothness.

[0020] In a preferred embodiment, the edge processing module is specifically used to acquire the optical data and the spindle coordinates of the corresponding machine tool in real time, and to analyze the surface error information of the reflector in real time based on the optical data and the spindle coordinates of the machine tool.

[0021] In a preferred embodiment, the system further includes a surface shape judgment module, which is used to judge whether the surface shape meets the standard based on the surface shape error information; the closed-loop feedback control module is used to start when the surface shape judgment module judges that the surface shape does not meet the standard.

[0022] In a preferred embodiment, the system further includes a monitoring module and a safety judgment module. The monitoring module monitors the pressure and temperature rise of the processing module and sends the data to the safety judgment module. The safety judgment module determines whether the processing operation is safe based on the pressure and temperature rise of the processing module. When the safety judgment module determines that the operation is safe, the processing module processes the reflector according to a preset processing trajectory. When the safety judgment module determines that the operation is unsafe, the processing module stops the processing operation. And / or, an alarm module issues a safety alarm when the safety judgment module determines that the operation is unsafe.

[0023] Secondly, the present invention discloses a method for in-situ real-time optical inspection and processing of a reflector, including:

[0024] The reflector is processed according to a preset processing trajectory;

[0025] Optical data of the reflector is acquired when the reflector is in the processing position, and the polishing liquid film in the detection area of ​​the reflector is displaced when the optical data is acquired;

[0026] The optical data is acquired in real time, and the surface shape error information of the reflector is analyzed in real time based on the optical data. The surface shape error is the difference between the real-time surface shape data and the ideal surface shape data.

[0027] The surface error information is acquired in real time and converted into processing module dwell time compensation information in real time. The preset processing trajectory is then corrected in real time based on the processing module dwell time compensation information.

[0028] The aforementioned in-situ real-time optical inspection and processing system and method for reflectors eliminates the need for cleaning, handling, and thermal balancing, thus improving processing efficiency. It enables the acquisition of optical data from the reflector while it is in the processing position, allowing for real-time in-situ inspection even through polishing media interference. This disclosure eliminates the need for machine downtime, achieving real-time in-situ optical inspection of the reflector and synchronizing processing and inspection, significantly shortening the processing cycle. Furthermore, this disclosure not only eliminates the need for repeated clamping, avoiding errors from repeated clamping, but also provides real-time compensation information for the dwell time of the processing module to compensate for thermal deformation generated during processing, thereby correcting the preset processing trajectory in real time, improving the accuracy of the processed surface shape, and resulting in excellent processing effects. Attached Figure Description

[0029] Figure 1 This is a schematic diagram of the system structure in one embodiment of the present disclosure;

[0030] Figure 2 This is a flowchart illustrating a method in one embodiment of the present disclosure. Detailed Implementation

[0031] The technical solutions of this disclosure will now be described in detail with reference to the accompanying drawings and preferred embodiments.

[0032] See Figure 1 This embodiment provides an in-situ real-time optical inspection and processing system for a reflector, the system comprising:

[0033] The processing module is used to process the reflector according to a preset processing trajectory;

[0034] An optical inspection module is used to acquire optical data of the reflector when it is in the processing position, and to remove the polishing liquid film in the inspection area of ​​the reflector when acquiring the optical data.

[0035] An edge processing module is used to acquire the optical data in real time and analyze the surface error information of the reflector in real time based on the optical data. The surface error is the difference between the real-time surface data and the ideal surface data.

[0036] The closed-loop feedback control module is used to acquire surface error information in real time, convert the surface error information into processing module dwell time compensation information in real time, and correct the preset processing trajectory in real time based on the processing module dwell time compensation information.

[0037] In this embodiment, the processing module includes a polishing head or a grinding wheel. It can be understood that the polishing head or grinding wheel is a processing execution device, or a processing head. In one embodiment, the processing module is a machine tool, and the machine tool has a polishing head. The machine tool can be a gantry-type structure. It can be understood that the polishing head or grinding wheel is used to achieve deterministic finishing of the mirror surface through a computer-controlled motion path (and pressure distribution). It can be understood that correcting the preset processing trajectory is equivalent to correcting the preset processing trajectory of the polishing head or grinding wheel.

[0038] Understandably, the processing module is used to process the reflector according to the latest preset processing trajectory.

[0039] The processing module can be a traditional mechanical polishing device, or a magneto-rheological finishing (MRF) device, an ion beam polishing (IBF) device, or a laser micro-etching device.

[0040] The mirror being in the processing position means that the mirror has not been removed from the processing equipment. Specifically, the mirror that has not been processed has not been removed from the processing machine tool and is still located on the processing machine tool.

[0041] Silicon carbide (SiC) is typically the preferred material for large space telescopes and high-resolution camera mirrors due to its high specific stiffness and high thermal stability. This disclosure applies not only to mirrors made of silicon carbide but also to mirrors made of other materials. This disclosure applies to large-aperture mirrors as well as small-sized mirrors.

[0042] The optical detection module includes an optical data acquisition device and a liquid displacement device. The optical data acquisition device is used to acquire optical data of the reflector, and the liquid displacement device is used to displace the polishing liquid film in the detection area of ​​the reflector when the optical data acquisition device acquires the optical data.

[0043] The liquid dissipation device is used to dissipate the polishing liquid film in the detection area at the instant detected by the optical data acquisition device. The liquid dissipation device can be an air curtain nozzle, a rotary centrifuge (using the lens rotation to spin dry the polishing liquid), or an ultrasonic dissipation head; in the finishing stage, a specially designed transparent dissipation head / nozzle can also be used to directly dissipate the liquid for observation.

[0044] In a preferred embodiment, the optical data acquisition device includes a stripe projector and a camera, wherein the camera referred to herein can be understood as a multispectral camera; the liquid displacement device employs an air curtain nozzle. The stripe projector is used to emit stripe light and illuminate it onto a reflector, and the camera is used to obtain the reflected image of the stripe light.

[0045] In one specific embodiment, the edge processing module is used to acquire the optical data in real time, analyze the optical data in real time to obtain the mirror surface gradient information of the reflector, and convert the mirror surface gradient information into surface error information.

[0046] Specifically, the camera is directly connected to the edge processing module via a Gigabit Ethernet Vision interface, which uses an FPGA (Field-Programmable Gate Array) or a GPU (Graphics Processing Unit).

[0047] In the preferred embodiment, the stripe projector can be a structured light projector. In this embodiment, the camera is an industrial camera, specifically a high-speed industrial camera. In this embodiment, the pressure of the air curtain nozzle is 0.4-0.6 MPa; in other embodiments, the air curtain nozzle is also a high-pressure air curtain nozzle. It is understood that the high-pressure air curtain nozzle uses high-pressure gas to form an air curtain, with the high pressure range typically between 2.5 MPa and 3.5 MPa. It is understood that the air pressure setting of the air curtain nozzle is related to the thickness of the polishing liquid film, as well as the size of the reflector, the camera's shooting area, etc.

[0048] In one specific embodiment, at the moment of camera exposure, the air curtain nozzle is pulsed open to "blow open" a transparent window on the polishing liquid film; the structured light projector projects stripes of a specific wavelength (such as 530nm green light), and a narrow-band filter is installed at the camera end to filter out most of the ambient light and the chaotic refraction of the polishing liquid.

[0049] The illumination area of ​​the stripe projector, the shooting area of ​​the camera, and the spraying area of ​​the air curtain nozzle overlap. Preferably, the illumination area of ​​the stripe projector is greater than or equal to the shooting area of ​​the camera, and further, the spraying area of ​​the air curtain nozzle is not smaller than the shooting area of ​​the camera.

[0050] When using high-pressure air curtain technology to detect polishing fluid flowing instantaneously across a mirror surface, preferably, the optical detection module is also used to combine multispectral imaging to filter residual scattering, i.e., the optical data is an image filtered of residual scattering.

[0051] The optical inspection module can be directly fixed to the side of the machining spindle using a shockproof bracket. For example, a high-speed camera and a structured light projector can be directly mounted on the crossbeam of the machine tool to form an optical inspection module that moves synchronously with the machining head.

[0052] In other embodiments, the optical data acquisition device may be a laser interferometer, a rangefinder sensor (LVDT), or a laser scanner. It is understood that other feasible devices may also be used, and these will not be exhaustively listed here.

[0053] The optical data satisfies the following condition: based on the optical data and the ideal surface shape data, the surface shape error of the reflector can be calculated. For example, the optical data can be a reflection image of fringe light (using a fringe projector and camera), can be interference fringes with alternating brightness and darkness, can be distance data measured by a range sensor, or can be three-dimensional coordinate point cloud data of the reflector surface (obtained using a laser scanner).

[0054] The edge processing module can acquire the spindle coordinates of the machine tool spindle and the optical data in real time, ensuring that the acquired optical data has corresponding spatial coordinates. The spindle coordinates are CNC (Computer Numerical Control) data (understandably, the machine tool includes a CNC system). That is, the edge processing module is used to acquire the optical data and the corresponding spindle coordinates of the machine tool in real time, and based on the optical data and spindle coordinates, analyze the surface error information of the reflector in real time.

[0055] In one embodiment, the optical data is image data (e.g., the optical data acquisition device is a stripe projector and a camera, the camera is used to capture the image of the light beam emitted by the stripe projector reflected by a mirror to obtain the reflected image of the stripe light; the liquid dissipation device is an air curtain nozzle). Then, the input of the edge processing module includes the spindle coordinates of the machining tool and the optical data. The edge processing module is used to perform multispectral narrowband filtering and information decoupling processing on the optical data in real time to obtain the phase perturbation component. The phase is extracted in real time using the Fourier transform method. Combined with the spindle coordinates of the machining tool, the influence of the fluid refractive index change on the optical path is calculated in real time and the phase is compensated and corrected to obtain the phase compensation result. The phase (phase compensation result) is unwrapped in real time to restore the real continuous phase distribution. Then, the spatiotemporal distribution of the fluid motion characteristics of the processing area is constructed in real time. The surface error information is obtained in real time based on the fluid motion characteristics. The closed-loop feedback control module processes the surface error information in real time to obtain a 3D surface residual cloud map and a dwell time deviation vector. The dwell time deviation vector is used as the dwell time compensation information of the machining module.

[0056] Specifically, the inputs to the edge processing module include optical data and CNC data. Optical data consists of an image set acquired by a multispectral camera, covering different center wavelengths. In some embodiments, the image set is a multi-channel interference fringe image set (e.g., channel wavelengths of 632.8 nm, 785 nm, and 1064 nm), and each fringe pattern contains distortion information introduced by fluid disturbance. CNC data includes the spatial coordinates (x, y, z) of the machine tool spindle and the instantaneous feed rate. The edge processing module performs the following steps:

[0057] Step 1: Perform multispectral narrowband filtering and information decoupling;

[0058] This step is not simply stray light filtering, but rather a collaborative filtering and decoupling of the images in the image set along the spectral dimension, based on the differences in refractive index-temperature-pressure sensitivity of different wavelengths of light in the fluid medium. Specifically:

[0059] Principle basis: Refractive index of fluid medium It is temperature ,pressure ,concentration function Different wavelengths of light pair , , The sensitivities of light differ (e.g., longer wavelengths are more sensitive to temperature, while shorter wavelengths are more sensitive to concentration). It is understood that the pressure refers to the thermodynamic pressure exerted on the fluid, and the concentration refers to the proportion of a component in the fluid.

[0060] Processing Method: Information decoupling employs a multispectral decoupling algorithm based on Principal Component Analysis (PCA) or Independent Component Analysis (ICA) to decompose the image into phase perturbation components independently contributed by the temperature, pressure, and concentration fields. This process is equivalent to constructing a system of equations in the "spectral dimension" to solve for the independent influence of each physical field on optical path distortion. Output: The decoupled multi-physics phase perturbation components are obtained, laying the foundation for subsequent accurate compensation.

[0061] Step 2: Perform phase extraction and fluid refractive index influence compensation;

[0062] Fourier transforms are performed on each phase perturbation component output in step 1 to extract the original phase distribution. Combining this with the real-time spindle coordinates provided by the CNC, the influence of fluid refractive index changes on the optical path is calculated in real time, and the phase data (original phase distribution) is compensated and corrected to obtain the phase compensation result. Specifically, using a fluid dynamics model (such as the simplified Navier-Stokes equation model, where Navier-Stokes refers to Navier-Stokes) and an optical refractive index model, the theoretical optical path offset caused by temperature, pressure, and concentration gradients at the current processing point is calculated. The theoretical optical path offset is then differentially fused with the phase perturbation components of each physical field (temperature field, pressure field, concentration field) extracted in step 1 to obtain the compensated phase data, i.e., the phase compensation result, thereby offsetting the measurement errors caused by temperature, pressure, and concentration changes.

[0063] Step 3, Phase unwrapping and surface reconstruction;

[0064] The compensated phase data is spatially unwrapped to restore the true continuous phase distribution. Based on the unwrapped phase, the spatiotemporal distribution of fluid motion characteristics (such as vorticity field and velocity gradient field) in the processing area (i.e., the processed area of ​​the reflector) is constructed. Finally, the true surface error information of the workpiece surface is derived from the fluid motion characteristics (i.e., the surface error information is obtained), thus eliminating the influence of fluid interference on the measurement.

[0065] Based on the results of step 3, the closed-loop feedback control module performs closed-loop feedback and control command generation: The closed-loop feedback control module processes the above-mentioned surface error information to generate a 3D surface residual cloud map and a residence time deviation vector; The 3D surface residual cloud map: as a phase gradient cloud map, it intuitively displays the spatial distribution of phase changes, reflects the characteristics of fluid velocity gradient, vortex structure, etc., and is used to judge the quality of the processed surface; The residence time deviation vector: as residence time compensation information of the processing module, it guides the magnetorheological or ion beam polishing equipment to perform non-uniform residence time correction in the next processing cycle, so as to achieve accurate convergence of surface error.

[0066] In one embodiment, the edge processing module may use a spaceborne NPU (Neural Processing Unit) or an FPGA. In another embodiment, the edge processing module adopts a distributed cloud computing architecture, transmitting data to an edge server for processing via a 5G or 6G low-latency link, and then transmitting it back to the machine tool / processing module.

[0067] Preferably, the dwell time compensation information of the processing module is obtained through deep reinforcement learning (DRL) method, and the input vector (State) corresponding to the deep reinforcement learning method is: current surface residual, processing head rotation speed, grinding pressure, real-time temperature, and track position;

[0068] The action space corresponding to the deep reinforcement learning method is: dwell time correction, feed rate, and coolant flow rate;

[0069] The reward function corresponding to the deep reinforcement learning method is: surface convergence speed and surface roughness smoothness.

[0070] Predictive maintenance involves the system predicting the lifespan of the machining actuator (polishing head or grinding wheel) based on historical wear data and providing early reminders for replacement.

[0071] In this embodiment, the system further includes a surface shape judgment module, used to determine whether the surface shape meets the standard based on the surface shape error information. If the surface shape meets the standard, the processing is completed; otherwise, the closed-loop feedback control module is activated. That is, the closed-loop feedback control module is used to activate its function when the surface shape judgment module determines that the surface shape does not meet the standard.

[0072] In this embodiment, the system further includes a monitoring module and a safety judgment module. The monitoring module monitors the pressure and temperature rise of the processing module and sends the data to the safety judgment module. Specifically, the system may also include an alarm module. The safety judgment module determines whether the processing operation is safe based on the pressure and temperature rise of the processing module. If the determination result is safe, the processing module processes the reflector according to a preset processing trajectory. That is, the processing module continues to process the reflector according to the preset processing trajectory when the safety judgment module determines that the operation is safe. If the determination result is unsafe, the processing is stopped (the machine is stopped and the spindle is returned) and / or the alarm module is activated to issue a safety alarm. In other words, the processing module stops working when the safety judgment module determines that the operation is unsafe, and / or the alarm module issues a safety alarm when the safety judgment module determines that the operation is unsafe.

[0073] Furthermore, the monitoring module is also used to monitor the vibration information of the processing module, and the safety judgment module is used to judge whether the processing operation of the processing module is safe based on the vibration information. If the judgment result is safe, the processing module processes the reflector according to the preset processing trajectory. If the judgment result is unsafe, the processing is stopped and / or the alarm module is activated to issue a safety alarm.

[0074] In one specific embodiment, the edge processing module and the security judgment module together serve as the control center.

[0075] This disclosure provides a method for in-situ real-time optical inspection and processing of a mirror, including:

[0076] The reflector is processed according to a preset processing trajectory;

[0077] Optical data of the reflector is acquired when the reflector is in the processing position, and the polishing liquid film in the detection area of ​​the reflector is displaced when the optical data is acquired;

[0078] The optical data is acquired in real time, and the surface shape error information of the reflector is analyzed in real time based on the optical data. The surface shape error is the difference between the real-time surface shape data and the ideal surface shape data.

[0079] The surface error information is acquired in real time and converted into processing module dwell time compensation information in real time. The preset processing trajectory is then corrected in real time based on the processing module dwell time compensation information.

[0080] In one specific embodiment, such as Figure 2 As shown, the method includes:

[0081] S1, process the reflector according to the preset processing trajectory;

[0082] S2, determine whether the detection sampling point has been reached. If it has been reached (the judgment result is yes), proceed to S3; otherwise (the judgment result is no), repeat S2.

[0083] S3, when the reflector is in the processing position, acquire the optical data of the reflector, and when acquiring the optical data, dissipate the polishing liquid film in the reflector detection area; specifically, turn on the stripe projector, turn on the air curtain nozzle to dissipate the polishing liquid film in the reflector detection area, and at the same time, the camera acquires the reflection image of the stripe light;

[0084] S4, acquire the optical data (reflection image of striped light) in real time, and analyze the surface error information of the reflector in real time based on the optical data. The surface error is the difference between the real-time surface data and the ideal surface data.

[0085] S5: Determine whether the surface shape meets the standard based on the surface shape error information. If it meets the standard (the judgment result is yes), the processing is completed; otherwise (the judgment result is no), proceed to S6.

[0086] S6, acquire surface error information in real time, convert the surface error information into processing module dwell time compensation information in real time, and correct the preset processing trajectory in real time according to the processing module dwell time compensation information;

[0087] S7. Obtain the pressure and temperature rise of the processing module and send them to the safety judgment module. Determine whether the processing operation is safe based on the pressure and temperature rise of the processing module. If the judgment result is safe (the judgment result is yes), the processing module processes the reflector according to the preset processing trajectory. If the judgment result is unsafe (the judgment result is no), the processing is stopped or the alarm module is activated to issue a safety alarm.

[0088] The disclosed solution has been verified by both computer simulation and prototype experiment. The specific results are as follows:

[0089] Environmental interference simulation experiment:

[0090] The effect of an air curtain nozzle on a 2 mm thick polishing liquid film was simulated using fluid dynamics (CFD). The results show that under pressures of 0.4–0.6 MPa, the air curtain can instantaneously form an interferometric measurement window with a diameter greater than 50 mm on the mirror surface, reducing the residual thickness of the liquid film to the nanometer level, thus meeting the measurement conditions for the fringe reflection method.

[0091] Algorithm robustness simulation:

[0092] The processing methods of the edge processing module and the closed-loop feedback control module were simulated in the Matlab environment. A polishing environment with strong noise and stray light refraction was simulated. The results show that the stripe contrast was improved by 4.5 times after using a 530nm narrowband filter, and the surface reconstruction error (RMS) was less than λ / 50 (λ=632.8nm, where λ represents wavelength), which proves the effectiveness of the algorithm in harsh environments.

[0093] Processing efficiency comparison experiment (prototype):

[0094] A comparative experiment was conducted on a 500mm diameter silicon carbide mirror.

[0095] Traditional solution: The machine needs to be stopped for cleaning, handling and testing every 2 hours of processing, with a total auxiliary time of about 6 hours.

[0096] The disclosed solution involves automatically performing a 10-second in-situ scan every 15 minutes during the processing.

[0097] Conclusion: Experimental data show that the present invention improves the single-cycle convergence efficiency by more than 75%. Due to the avoidance of repeated clamping, the final surface accuracy after system error compensation is better than the result of traditional displacement detection compensation.

[0098] This disclosure presents a system and method for real-time in-situ optical inspection and processing of a reflector, eliminating the need for cleaning, handling, and thermal balancing, thus improving processing efficiency. This disclosure eliminates the need for repeated clamping, avoiding errors caused by repeated clamping. Thermal deformation during processing can cause changes in the acquired optical data; this disclosure can obtain corresponding processing module dwell time compensation information in real time, correcting the preset processing trajectory in real time, resulting in better processing effects and improved accuracy of the processed surface. This disclosure can acquire the reflector's optical data while the reflector is in the processing position, enabling real-time in-situ inspection that penetrates polishing media interference. This disclosure eliminates the need for machine downtime; real-time in-situ optical inspection of the reflector can be achieved during processing intervals or by pausing for a few seconds, synchronizing processing and inspection and significantly shortening the processing cycle.

[0099] This disclosure achieves "zero repeat clamping error" and physically unifies the machining coordinate system and the detection coordinate system. Based on this disclosure, a data storage module is set up, which enables digital recording of the entire process from rough grinding to fine polishing, and allows traceability of the state at any moment during processing.

[0100] This disclosure can be applied to multiple fields, such as: in the semiconductor field, specifically for real-time flatness monitoring during the polishing process of large-size silicon wafers; in the civilian precision manufacturing field, specifically for automated production lines of high-precision automotive windshield molds and large-aperture amateur astronomical telescope lenses; in the aerospace high-temperature components field, specifically for precision grinding and inspection of ceramic matrix composite turbine blades, achieving dimensional control in high-temperature and multi-coolant environments; and in the nuclear power equipment field, specifically for ultra-precision grinding of nuclear reactor sealing surfaces, enabling non-contact automatic quality monitoring in environments with radiation and high humidity.

[0101] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0102] The embodiments described above are merely illustrative of several implementations of this disclosure, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this disclosure, and these all fall within the protection scope of this disclosure. Therefore, the protection scope of this patent should be determined by the appended claims.

Claims

1. A mirror in situ real-time optical inspection and processing system, characterized in that, include: The processing module is used to process the reflector according to a preset processing trajectory; An optical inspection module is used to acquire optical data of the reflector when it is in the processing position, and to remove the polishing liquid film in the inspection area of ​​the reflector when acquiring the optical data. An edge processing module is used to acquire the optical data in real time and analyze the surface error information of the reflector in real time based on the optical data. The surface error is the difference between the real-time surface data and the ideal surface data. The closed-loop feedback control module is used to acquire surface error information in real time, convert the surface error information into processing module dwell time compensation information in real time, and correct the preset processing trajectory in real time according to the processing module dwell time compensation information. The optical detection module includes an optical data acquisition device and a liquid displacement device. The optical data acquisition device is used to acquire optical data of the reflector, and the liquid displacement device is used to displace the polishing liquid film in the detection area of ​​the reflector when the optical data acquisition device acquires the optical data. The optical data is image data; the input of the edge processing module includes the spindle coordinates of the machine tool and the optical data. The real-time analysis of the reflector's surface error information based on the optical data specifically involves: real-time multispectral narrowband filtering and information decoupling processing of the optical data to obtain phase perturbation components; real-time phase extraction using Fourier transform; real-time calculation of the influence of fluid refractive index changes on the optical path based on the spindle coordinates of the machine tool and phase compensation correction to obtain phase compensation results; real-time unwrapping of the phase compensation results to restore the true continuous phase distribution; real-time construction of the spatiotemporal distribution of fluid motion characteristics in the processing area; and real-time obtaining surface error information based on the fluid motion characteristics. The real-time conversion of the surface error information into processing module dwell time compensation information specifically involves real-time processing of the surface error information to obtain a 3D surface residual cloud map and a dwell time deviation vector, with the dwell time deviation vector serving as the processing module dwell time compensation information.

2. The in-situ real-time optical inspection and processing system for a reflector according to claim 1, characterized in that, The optical data acquisition device is a range sensor or a laser scanner, or the optical data acquisition device is a stripe projector and a camera.

3. The in-situ real-time optical inspection and processing system for a reflector according to claim 1, characterized in that, The liquid discharge device is an air curtain nozzle, a rotary centrifuge device, or an ultrasonic discharge head.

4. The in-situ real-time optical inspection and processing system for a reflector according to claim 1, characterized in that, The dwell time compensation information of the processing module is obtained through a deep reinforcement learning method. The input vector of the deep reinforcement learning method is: current surface residual, processing head speed, grinding pressure, real-time temperature, and trajectory position; the corresponding action space is: dwell time correction, feed rate, and coolant flow rate; and the corresponding reward function is: surface convergence speed and surface roughness smoothness.

5. The in-situ real-time optical inspection and processing system for a reflector according to claim 1, characterized in that, The edge processing module is specifically used to acquire the optical data and the corresponding spindle coordinates of the machine tool in real time, and to analyze the surface error information of the reflector in real time based on the optical data and the spindle coordinates of the machine tool.

6. The in-situ real-time optical inspection and processing system for a reflector according to claim 1, characterized in that, The system also includes a surface shape judgment module, which is used to judge whether the surface shape meets the standard based on the surface shape error information; the closed-loop feedback control module is used to start when the surface shape judgment module judges that the surface shape does not meet the standard.

7. The in-situ real-time optical inspection and processing system for a reflector according to claim 1, characterized in that, The system also includes a monitoring module and a safety judgment module. The monitoring module monitors the pressure and temperature rise of the processing module and sends the data to the safety judgment module. The safety judgment module determines whether the processing operation is safe based on the pressure and temperature rise of the processing module. When the safety judgment module determines that the operation is safe, the processing module processes the reflector according to a preset processing trajectory. When the safety judgment module determines that the operation is unsafe, the processing module stops the processing operation. And / or, an alarm module issues a safety alarm when the safety judgment module determines that the operation is unsafe.

Images