Mask optical film self-adaptive pressing equipment and control method thereof

Abstract

The invention discloses a mask optical film self-adaptive press-fit device and a control method thereof, and relates to the technical field of mask press-fit, the mask optical film self-adaptive press-fit device comprises an operation counter, the operation surface of the operation counter is provided with a P plate fixing frame and a G plate fixing frame; the G plate fixing frame is provided with a multi-cylinder pushing structure, the multi-cylinder pushing structure comprises a plurality of telescopic cylinders, the output ends of the telescopic cylinders are all connected with the G plate clamp, and the output ends of the telescopic cylinders are evenly distributed in the circumferential direction of the frame of the G plate; a P plate clamp is fixed on the P plate fixing frame, and a P plate is mounted on the P plate clamp; the control system can control the telescopic length of each telescopic cylinder. Pressure sensor arrays are arranged on the mounting surface of the G plate clamp and the mounting surface of the P plate clamp, and the multi-cylinder propelling structure is used for pressing a G plate on a P plate. By means of the scheme, high-precision dynamic parallel pressing of the G plate and the P plate can be achieved, the local flatness defect can be automatically corrected, and reworking is not needed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of mask pressing, in particular to a mask optical film self-adaptive pressing device and a control method thereof. BACKGROUND

[0002] Traditional IC mask optical film pressing devices adopt single-point motor-driven pressing mechanisms, as shown in the figure, and realize parallel pressing of G masks (MASK clamps) and P masks (optical film clamps) through mechanical leveling. Figure 1 The pressing parameters depend on manual selection of recipes, and the operator needs to manually select parameters, which is prone to errors and low in efficiency; the pressure uniformity depends on mechanical precision, lacks real-time calibration capability, and single-point driving is prone to uneven force on the pressing surface, affecting the bonding quality of the optical film and the MASK, resulting in poor pressure uniformity; manual detection of flatness is required after pressing, real-time detection of pressing parallelism and pressure distribution is not possible, and the efficiency is low and the film is prone to wrinkles or adhesion failure due to uneven local pressure, and lacks intelligent calibration; and there is no automatic re-pressing mechanism after pressing, and unqualified products need to be returned to work. SUMMARY

[0003] The present application is to solve the problems of the prior art, and aims to provide a mask optical film self-adaptive pressing device and a control method thereof, which can realize high-precision dynamic parallel pressing of G masks and P masks, and automatically correct local flatness defects without the need for rework.

[0004] The present application is achieved by the following technical solutions:

[0005] A mask optical film self-adaptive pressing device, comprising:

[0006] An operation counter, the operation surface of the operation counter is provided with a P mask fixing frame and a G plate fixing frame respectively; the G plate fixing frame is provided with a multi-cylinder propulsion structure, the multi-cylinder propulsion structure comprises a plurality of telescopic cylinders, the output ends of the plurality of telescopic cylinders are connected with a G mask clamp, the G mask clamp is used for mounting a G plate, and the output ends of the plurality of telescopic cylinders are evenly distributed along the circumference of the frame of the G plate; the P mask fixing frame is fixed with a P mask clamp, and the P mask clamp is used for mounting a P plate;

[0007] A control system, the control system can control the extension length of each telescopic cylinder;

[0008] The mounting surface of the G mask clamp and the P mask clamp is provided with a pressure sensor array, and the multi-cylinder propulsion structure is used to press the G plate onto the P plate.

[0009] Further solutions:

[0010] The present application also provides a control method of a mask optical film self-adaptive pressing device, comprising the following steps:

[0011] S1: fixing the G plate on the G plate fixture and fixing the P plate on the P plate fixture;

[0012] S2: controlling the control system to control a plurality of telescopic cylinders to extend synchronously, and pressing the G plate on the P plate according to a preset pressure; in the pressing process, the pressure distribution is detected in real time by a pressure sensor array and feedback is performed, and the control system adjusts the output pressure of the plurality of telescopic cylinders in real time according to the feedback value, so that the pressure distribution is uniform;

[0013] S3: after the first pressing is completed, the pressing surface is scanned by a 3D laser scanner, and flatness data is obtained, the control system compares the flatness data with ideal flatness data, and judges whether the standard is met;

[0014] S4: when the standard is not met, entering a re-pressing process, calculating and identifying a defect area with unqualified flatness, and driving the corresponding telescopic cylinder at the defect area by the control system to apply gradient pressure to the defect area until the flatness meets the standard.

[0015] Further optimization, in the step S2, the specific step of pressing the G plate on the P plate according to the preset pressure comprises:

[0016] Taking an image of the film frame silk screen printing by a camera;

[0017] The control system automatically identifies the film frame silk screen printing image, and compares the similarity with a historical formula database stored by itself; the historical formula database includes a plurality of stored film frame silk screen printing images and respective corresponding optimal pressing formulas;

[0018] When the similarity comparison is performed, the similarity comparison calculation is performed on the text recognition, geometric size and pattern feature respectively by image processing, and the similarity of the three is fused to obtain a matching degree;

[0019] The maximum matching degree is selected, and the image corresponding to the maximum matching degree and the optimal pressing formula corresponding to the maximum matching degree are selected as the output result.

[0020] Further optimization, in the step S2, the specific step of adjusting the output pressure of the plurality of telescopic cylinders in real time by the control system according to the feedback value, so that the pressure distribution is uniform, comprises:

[0021] Real-time acquisition of pressure sensor data of two pressure sensor arrays, and synchronous acquisition of pressure states output by a plurality of telescopic cylinders;

[0022] Calculating stiffness adaptive fusion weights wG and wP based on the material stiffness characteristics of the G plate and the P plate respectively;

[0023] According to the stiffness adaptive fusion weight, double version pressure data weighted fusion is performed to obtain accurate pressure distribution data;

[0024] Based on the accurate pressure distribution data, pressure standard deviation calculation is performed, and a pressure uniformity evaluation index is obtained according to the pressure standard deviation;

[0025] According to the pressure uniformity evaluation index, a pressure adjustment instruction of the telescopic cylinder is generated.

[0026] Further optimization, the calculation formula of the adaptive fusion weight wG and wP is respectively:

[0027] ;

[0028] ;

[0029] The calculation formula of the double version pressure data weighted fusion is:

[0030] ;

[0031] In the above formula, StiffnessG and StiffnessP are the stiffness coefficients of G version and P version respectively; P sensor G(x,y,t) is the G version pressure sensor array data, P sensor P(x,y,t) is the P version pressure sensor array data, x and y are the sensor grid coordinates, and t is the time sequence.

[0032] Further optimization, the calculation formula of the pressure standard deviation is:

[0033] ;

[0034] The calculation formula of the pressure uniformity evaluation index is:

[0035] ;

[0036] In the above formula, m and n are the row and column numbers of the sensor grid respectively, and μP(t) is the average pressure value at time t.

[0037] Further optimization, the step S4 further includes the following specific steps:

[0038] Based on the surface flatness scanning data obtained by the 3D laser scanner, threshold judgment is performed through the generated defect positioning mask, and the defect area with unqualified flatness is output;

[0039] The equivalent stiffness of each layer of material of the P version is calculated respectively to obtain the overall equivalent stiffness in series;

[0040] Based on the overall equivalent stiffness and the degree of defects in the defect region, the gradient pressure is calculated, and the recommended number of re-pressure cycles is determined based on historical data.

[0041] Based on the gradient repress pressure and the number of repress times, the repair effect is verified by rescanning after repressing.

[0042] Further optimization, the formula for determining the defect location mask is as follows:

[0043] ;

[0044] In the above formula, H scan (x,y) is the flatness scan data matrix, x,y are the scan grid coordinates, and H ideal τ represents the ideal flatness height (mm). defect The defect detection threshold is (mm).

[0045] Further optimization, the formula for calculating the overall equivalent stiffness is as follows:

[0046] ;

[0047] In the formula, k eq k is the overall equivalent stiffness coefficient (N / mm). Al k is the stiffness coefficient of the aluminum alloy frame (N / mm). film k is the stiffness coefficient of the cellulose film (N / mm). adhesive The stiffness coefficient of the acrylic soft rubber (N / mm);

[0048] The formula for calculating gradient pressure is:

[0049] ;

[0050] In the formula, F rework (x,y) represents the gradient complex pressure distribution, P base The basic pressure is (kPa), α is the pressure gradient coefficient (5-15 kPa·mm / N), and k eq Equivalent stiffness (N / mm).

[0051] Further optimization includes steps for continuous health prediction of key equipment components and supporting stress components during the pressing process, specifically including:

[0052] Real-time reading of component stress grid data, equipment monitoring sensor data, and component material fatigue parameters;

[0053] The equivalent stress of each grid point is calculated using the von Mises equivalent stress formula, and the risk level is classified based on the equivalent stress.

[0054] Then, the cumulative strain is calculated and the damage model is derived, and the remaining service life is calculated based on the damage model.

[0055] By combining risk level and remaining useful life, a health prediction result is generated.

[0056] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0057] 1. The present invention provides an adaptive pressing device and control method for optical films on photomasks. Using this solution, high-precision dynamic parallel pressing of G-plate and P-plate can be achieved, and local flatness defects can be automatically corrected without rework.

[0058] 2. The present invention provides an adaptive pressing device and control method for optical films on photomasks. Using this solution, the edge / center pressure error is ≤±1% through multi-cylinder distributed drive + real-time sensing calibration; the AI ​​re-pressing module makes the repair rate of local unevenness defects >95%.

[0059] 3. The present invention provides an adaptive pressing device and control method for optical films on photomasks. Using this solution, the mask model and size can be automatically identified, the pressing parameter matching speed is increased by 90%, the dependence on manual parameter selection is eliminated, and the pressing parameter matching is fully automatic.

[0060] 4. The present invention provides an adaptive pressing device for optical films and its control method, which can build a health prediction model for the device, provide early warning of structural deformation risks, and achieve a structural deformation prediction accuracy of ≥85%, thereby reducing losses from sudden downtime. Attached Figure Description

[0061] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be considered as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings:

[0062] Figure 1 This is a schematic diagram of a single-point motor film application device in the prior art;

[0063] Figure 2 This is a schematic diagram showing the positional relationship between the optical film and the photomask provided by the present invention;

[0064] Figure 3 This is a schematic diagram of the adaptive pressing device provided by the present invention;

[0065] Figure 4 A schematic diagram of the structure of the G-plate fixing frame provided by the present invention;

[0066] Figure 5 This is a schematic diagram of the structure of the P-plate fixing frame provided by the present invention;

[0067] Figure 6 This is a schematic diagram of the control method provided by the present invention.

[0068] The attached diagram shows the markings and corresponding component names:

[0069] 1-Operating counter, 2-P-board mounting bracket, 3-G-board mounting bracket, 4-Multi-cylinder propulsion structure, 5-Pressure sensor array, 6-3D laser scanner, 7-Camera, 8-Display. Detailed Implementation

[0070] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0071] Example 1: This Example 1 provides an adaptive pressing device for optical films on photomasks, such as... Figures 2-5 As shown, it specifically includes an operating counter 1, on which a P-plate fixing frame 2 and a G-plate fixing frame are respectively provided; the G-plate fixing frame is provided with a multi-cylinder propulsion structure 4, which includes several telescopic cylinders, which are evenly distributed along the circumference of the film frame, and the output ends of the telescopic cylinders are all connected to the G-plate fixture. The mounting surface of the G-plate fixture is provided with a temporary fixing block for installing the G-plate, i.e., installing the photomask; while the P-plate fixing frame 2 is fixed with a P-plate fixture, and the mounting surface of the P-plate fixture has an L-shaped placement block, and a telescopic clamping mechanism is provided on one side of the L-shaped placement block to install the P-plate, i.e., the optical film.

[0072] Secondly, the adaptive pressing equipment also includes a pressure sensing system, which specifically comprises a pressure sensor array 5 distributed on the mounting surfaces of the G-plate fixture and the P-plate fixture. The pressure sensor array 5 can be a thin-film pressure sensor array 5 or a piezoelectric ceramic sensor array to achieve pressure monitoring at several array points on the plane. The control system can dynamically adjust the stroke of each telescopic cylinder based on the data fed back by the pressure sensor array 5.

[0073] The adaptive lamination equipment also includes a 3D contour scanning module, an automatic image recognition system, and a health prediction system. The 3D contour scanning module, comprising a 3D laser scanner 6, uses a line-scanning laser sensor to perform a 3D scan of the laminated optical film surface, generating a topological map of the aluminum alloy film frame's flatness to identify and locate defective areas, allowing for re-lamination of these areas. The automatic image recognition system uses a camera 7 to capture the silkscreened model number on the aluminum alloy film frame, and a 3D line-scanning laser to scan the mask size and thickness, automatically matching the lamination parameters (pressure and lamination time at different lamination stages) from the database. The health prediction system constructs a structural deformation mesh model using historical data from resistance strain gauge pressure sensors to predict future deformation trends; when the predicted deformation exceeds a threshold, it automatically triggers a maintenance alarm and generates a spare parts order.

[0074] In addition, the present invention can also be used in fields requiring high-precision planar lamination, such as PCB lamination machines and display panel bonding equipment; the AI ​​lamination module can be adapted to hydraulic / pneumatic drive systems.

[0075] Example 2: This Example 2 further optimizes Example 1 and provides a control method for an adaptive pressing device for photomask optical films, such as... Figure 6 As shown, the specific steps include the following:

[0076] 1. Equipment initialization: The system is powered on, and each sensor (camera 7, 3D laser scanner 6, pressure sensor array 5) and the multi-cylinder propulsion structure 4 perform self-tests.

[0077] 2. Manually fix the optical film (acrylic side up) to the P plate (lower template) and place the MASK (mask) in the designated position.

[0078] 3. The vision system captures the silkscreen markings on the membrane frame (e.g., "AL-2024-T6"), and the 3D laser scanner 6 acquires the precise dimensional point cloud data of the mask. This data will serve as input for AI parameter matching. The system will compare the collected data with a historical formula database using multiple features to automatically recommend the optimal pressing formula (pressure, temperature, time). If the matching confidence level is below a threshold (e.g., 0.8), the process will proceed to "manual review." Specifically, this includes:

[0079] 3.1. First, acquire the screen printing image (RGB matrix) using camera 7. vision The 3D scanned point cloud data LscanMASK(x,y,z) and MASK-sized data are stored in the historical recipe database within the control system. database .

[0080] 3.2. Subsequently, the similarity of text recognition, geometric dimensions, and pattern features was calculated respectively:

[0081] Text feature similarity Stext: from I vision Extract the silkscreen text (e.g., "AL-2024-T6") from (x, y) and compare it with the text in the database. Calculate the similarity using the edit distance (Levenshtein distance) algorithm:

[0082] ;

[0083] In the formula, T current This is the current text, written by I. vision (x,y) is extracted from T. database It is database text, Stext ranges from [0,1], the larger the value, the more similar the text.

[0084] Pattern feature similarity S pattern :fromI vision Extract image features (such as color histogram or texture) from (x,y), and calculate using histogram intersection:

[0085] ;

[0086] In the formula, Hcurrent and Hdatabase are the histograms of the current image and the database image, respectively, and n is the number of histogram buckets.

[0087] Geometric similarity (Sgeometry): Compares dimensions (such as diagonal length or area) based on LscanMASK(x,y,z) data with dimensions in a database. The formula is:

[0088] ;

[0089] In the formula, L current This is the current size (derived from point cloud computing), L database It is the database size, L max This is the maximum permissible deviation (e.g., 0.1 mm); S geometry The range is [0,1], with a larger value for smaller deviations. By comparing the current mask scan size with the size deviation in the database, the Gaussian kernel function is used to calculate the geometric similarity between the current mask size and each recipe in the database. The σ parameter controls the leniency of the matching.

[0090] 3.3 Subsequently, multi-feature similarity fusion is performed:

[0091] ;

[0092] In the formula, w1, w2, and w3 are weighting coefficients (each defaults to 1 / 3, and w1 + w2 + w3 = 1), S text For text similarity, Sgeometry For geometric similarity (Formula 9), S pattern For pattern feature similarity, S total The range is [0,1], with higher values ​​indicating better matching. By using image processing techniques (such as OCR to extract text) and point cloud registration to calculate geometric size similarity, the accuracy of recipe matching is improved by integrating multi-dimensional similarity of text recognition, geometric size, and pattern features.

[0093] 3.4 Finally, the formulation process parameter with the highest total value is taken as the Recipe. optimal Process results, and calculation of confidence level. match =Stota, if Confidence match If the value is less than 0.8 (adjustable threshold), the system can trigger manual review. (Note: The last part, "Recipe," appears to be an unrelated fragment and is left untranslated.) optimal For optimal compression formulation parameters (structure), Confidence match To match the confidence level (0-1 scalar).

[0094] 4. After determining the optimal pressing data, the initial pressing is initiated. The control system synchronously extends several telescopic cylinders, pressing the G plate onto the P plate according to the preset pressure. A mesh of thin-film pressure sensors provides real-time feedback on the pressure distribution, and the AI ​​algorithm dynamically adjusts the stroke and pressure of each electric cylinder to ensure the pressure across the entire plane remains stable within the target range (e.g., 80 ± 0.2 kPa). This step innovatively employs weighted fusion of pressure data from the G and P plates, effectively eliminating unilateral measurement errors and improving measurement accuracy. Furthermore, dynamic weight calculation based on stiffness characteristics ensures the fused pressure data better reflects actual stress conditions. Pressure standard deviation and uniformity indicators provide quantitative basis for AI decision-making, achieving a complete closed-loop control from pressure sensing to adjustment, ensuring the stability of the pressing quality.

[0095] The initial pressing process specifically includes the following steps:

[0096] 4.1 Data Acquisition Phase: Real-time acquisition of pressure sensor data from the G and P plates, simultaneously acquiring the output pressure status of the electric cylinders. Specifically, PsensorG(x,y,t) represents the data from the G plate's 5-cell thin-film pressure sensor array (m×n matrix), acquiring the pressure values ​​at each grid point of the G plate in real-time, in kPa. This data reflects the pressure distribution of the upper template, where x and y represent the sensor grid coordinates, and t represents the time series. PsensorP(x,y,t): Data from the P plate's 5-cell thin-film pressure sensor array (m×n matrix), acquiring the pressure values ​​at each grid point of the P plate in real-time, in kPa. This data reflects the pressure distribution of the lower template, complementing the G plate data. Factuator(k,t): Current output pressure data for each electric cylinder (k=1,2,…,N), representing the real-time output status of the actuator, used for control feedback and dynamic adjustment. MASKgeometry: Mask geometric parameters, including size and shape features, used to establish the desired pressure distribution pattern and analysis benchmark.

[0097] 4.2 Data preprocessing stage: The sensor data is filtered and standardized to eliminate noise and baseline drift.

[0098] 4.3 Weight Calculation Stage: Adaptive fusion weights wG and wP are calculated based on material stiffness characteristics; where: , Where StiffnessG and StiffnessP are the stiffness coefficients of version G and version P, respectively. This step dynamically adjusts the fusion weights based on the relative stiffness of versions G and P, making the fused pressure data more consistent with actual stress conditions.

[0099] 4.4 Data Fusion Stage: Subsequently, the two versions of pressure data are fused to obtain a more accurate pressure distribution P. fused (x,y,t). Where:

[0100] ;

[0101] In the formula, P sensor G(x,y,t) represents the data from pressure sensor array 5 (version G), P sensor P(x,y,t) represents the data from pressure sensor array 5 (version P), where x and y are the sensor grid coordinates, and t is the time series. This step weights and fuses the pressure sensor data from versions G and P to eliminate one-sided measurement errors and improve the accuracy of pressure measurement.

[0102] 4.5 Uniformity Assessment Stage: Calculate the pressure standard deviation σP(t) and the uniformity index P respectively. uniformity (t); where:

[0103] The formula for calculating the standard deviation of pressure is:

[0104] ;

[0105] μP(t) is the average pressure value at time t:

[0106] ;

[0107] The formula for calculating the pressure uniformity evaluation index is as follows:

[0108] ;

[0109] In the formula, m and n are the number of rows and columns of the sensor grid, respectively. This step can calculate the spatial variability of the pressure distribution, reflecting the severity of pressure non-uniformity. It also quantifies the uniformity of the pressure distribution, providing a quantifiable evaluation standard for AI decision-making. The pressure uniformity index (0-1 scalar) quantifies the uniformity of the pressure distribution, with 1 representing complete uniformity and 0 representing complete non-uniformity. This index provides a quantitative basis for AI decision-making.

[0110] 4.6 Control Decision Stage: Generate electric cylinder pressure adjustment command ΔF based on uniformity index. adjust (k,t+Δt); ΔF adjust (k,t+Δt) represents the dynamic adjustment of the electric cylinder pressure, calculated based on the real-time pressure distribution, with the unit being kPa. This parameter directly guides the adjustment of the actuator's movements.

[0111] 4.7 Execution Feedback Phase: Implement pressure adjustments and monitor the effects to form a closed-loop control.

[0112] 5. After the initial pressing, a 3D laser scanner 6 scans the pressed surface, generating a flatness cloud map and calculating the deviation from the ideal plane. The system automatically judges based on a preset defect threshold (e.g., 0.01mm). If the threshold is met, the system proceeds to the health inspection stage; otherwise, the re-pressing process is initiated. This step uses intelligent defect identification and an adaptive threshold algorithm to accurately identify various flatness defects. Pressure is applied in stages according to the severity of the defects to avoid over-correction or under-correction. For example, based on the depth of the defect and the material stiffness characteristics, the system can calculate the gradient pressure curve to be applied to the area (e.g., 85kPa→90kPa→85kPa). By considering the stiffness characteristics of multi-layered materials, the accuracy of the re-pressing parameter calculation is improved, and the re-pressing strategy is optimized based on historical pressure distribution data to continuously improve the repair effect. During re-pressing, only the specific electric cylinder corresponding to the defect area is activated to apply the calculated gradient pressure. After completion, the process returns to the flatness inspection step for re-verification, forming a closed-loop correction circuit until the flatness meets the standard or the maximum number of re-pressing cycles is reached.

[0113] The re-pressure process specifically includes the following steps:

[0114] 5.1 Flatness Data Acquisition: Surface morphology data H was acquired using a 3D line-scanning laser system. scan (x,y), H scan (x,y) represents 3D line-scanned laser flatness scan data (W×H matrix), in millimeters; this data was acquired using a high-precision laser scanner and reflects the three-dimensional morphological features of the optical film surface after lamination, where x,y represent the scan grid coordinates. P was also collected. history (x,y,t) and Material properties P history (x,y,t) represents the historical pressure distribution data sequence, recording the pressure distribution at each time point from the initial pressing. This data is used to analyze the correlation between pressure distribution and smoothness defects, providing a basis for optimizing re-pressing parameters. properties This is a set of parameters for optical film materials, including the stiffness characteristics of each layer: k Al k is the stiffness coefficient of the aluminum alloy frame (N / mm). film k is the stiffness coefficient of the cellulose film (N / mm). adhesive The stiffness coefficient of the acrylic soft rubber (N / mm).

[0115] 5.2 Defect Region Identification: Generating a Defect Location Mask M Based on Threshold Judgment defect (x,y):

[0116] ;

[0117] In the above formula, H scan (x,y) is the flatness scan data matrix, x,y are the scan grid coordinates, and H ideal τ represents the ideal flatness height (mm). defect This is the defect detection threshold (mm). The threshold is used to automatically identify defective areas that do not meet the flatness standards, providing a target area for localized re-pressuring.

[0118] 5.3 Material property analysis: Calculation of the equivalent stiffness k of multilayer materials eq :

[0119] ;

[0120] In the formula, k eq k is the overall equivalent stiffness coefficient (N / mm). Al k is the stiffness coefficient of the aluminum alloy frame (N / mm). film k is the stiffness coefficient of the cellulose film (N / mm). adhesiveThe stiffness coefficient (N / mm) of the acrylic flexible rubber is given. This is achieved by considering the series stiffness characteristics of the multi-layer optical film structure (aluminum alloy frame + cellulose film + acrylic flexible rubber) to accurately calculate the overall equivalent stiffness.

[0121] 5.4 Calculation of Gradient Pressure Parameters: Calculate the gradient pressure F based on the degree of defect. rework (x,y):

[0122] ;

[0123] In the formula, F rework (x,y) represents the gradient complex pressure distribution, P base The basic pressure is (kPa), α is the pressure gradient coefficient (5-15 kPa·mm / N), and k eq The equivalent stiffness (N / mm) is calculated based on the flatness deviation and material stiffness characteristics, determining the pressure distribution required for gradient compressive stress.

[0124] 5.5 Determine the recommended number of re-pressure cycles N based on historical data rework Based on the severity of defects and historical data, this method is optimized to ensure effective repair while avoiding excessive pressing.

[0125] 5.6 Effect Verification: After performing the re-compression, rescan to verify the repair effect.

[0126] 6. Throughout the entire process of executing all the above steps, system health monitoring and early warning are also required. Specifically, throughout the process, the system continuously monitors the performance indicators (such as the standard deviation of pressure fluctuations and the deformation detected by strain gauges) of key components and supporting load-bearing components (such as electric cylinders). If an abnormal trend is detected (such as a 30% increase in fluctuations), the system predicts the component's lifespan and automatically generates an early warning message to notify maintenance personnel. Finally, the process ends.

[0127] The specific steps for equipment health prediction include:

[0128] 6.1 Real-time reading of F structural (t), S monitor (t) and M aterialfatigue ;F structural (t) represents the structural stress mesh data, S monitor (t) represents the data from the device's monitoring sensors, M aterialfatigue These are material fatigue parameters.

[0129] 6.2 The equivalent stress at each grid point is calculated using the von Mises equivalent stress formula, which converts the multiaxial stress state into a uniaxial equivalent stress, facilitating comparison with the material's yield strength and fatigue life prediction.

[0130] ;

[0131] Wherein, input parameter F structural (t) to provide stress component σ xx , σ yy , σ zz , τ xy , τ yz , τ zx (For plane stress state, σ can be simplified) zz =0,τ yz =τ zx =0); and output σ von (t) is used to calculate the deformation risk level Riskdeformation(t).

[0132] 6.3 According to σ von (t) / σ y Ratio-based classification of Riskdeformation(t):

[0133] ;

[0134] Where, σ y The input parameter M aterialfatigue The yield strength within the range. Risk level classification is based on engineering standards (such as ASME), with a warning triggered when the level is ≥4.

[0135] 6.4 Subsequently, the cumulative strain ε was calculated. cum (t), and derive the damage D(t), where the cumulative strain ε cum (t) is obtained by integrating the strain rate and is the basis for fatigue damage assessment. The formula is:

[0136] ;

[0137] Wherein, the strain rate ε˙(τ) is determined by the input parameter S monitor (t) and stress data derivation, for example based on Hooke's law ε˙=σ˙ von / E, where E is Young's modulus (from M) aterialfatigue ), σ˙ von This is the rate of change of von Mises stress. This step calculates the total cumulative strain through strain rate integration, providing fundamental data for material fatigue damage assessment.

[0138] The cumulative damage D(t) is calculated using Miner's linear accumulation rule:

[0139] ;

[0140] In the formula, n i It is the stress level σ i The number of loops, Nfi It is the fatigue life at this stress level (by M) aterialfatigue (The SN curve is determined in the text).

[0141] To adapt to real-time monitoring, the discrete formula is converted into a continuous integral form:

[0142] ;

[0143] In the formula, The strain rate is measured directly using a sensor (such as a strain gauge); This represents the differential damage increment; The fatigue life corresponding to the current stress; Instantaneous damage rate; For integration (summation), it sums up all instantaneous damage rates from time zero to the current time t, thus obtaining the total cumulative damage D(t) from the beginning to the present.

[0144] 6.5 RUL(t) is calculated based on the damage model, where RUL is calculated based on the current damage and failure threshold. It is assumed that failure occurs when D(t) ≥ 1:

[0145] ;

[0146] In the formula, D˙(t) is the damage change rate, which is fitted from real-time data.

[0147] 6.6 Finally, combining the risk level and RUL, a maintenance plan is generated. If Riskdeformation(t) ≥ 4 or RUL(t) < 72h, an emergency maintenance recommendation (such as immediate component replacement) is output. Otherwise, a preventative plan is output (to be checked during the next scheduled maintenance). The algorithm runs periodically (e.g., once every time the optical film is pressed and once every hour) to ensure real-time performance.

[0148] Example 3: Based on Example 2, this Example 3 provides a specific implementation method.

[0149] 1. Initial bonding of the optical film and the mask:

[0150] The optical film (acrylic adhesive side up) is fixed to the P plate; camera 7 captures the silkscreened "AL-2024-T6" on the film frame, and the 3D laser scan shows the MASK size to be 6 inches; the system automatically calls the parameters: pressure 80kPa, temperature 25℃, time 120s; multiple electric cylinders push the G plate down, and the film sensor provides real-time feedback of the pressure value to the AI ​​algorithm; the AI ​​first pressing module dynamically adjusts the electric cylinder stroke to stabilize the pressure on all four sides to 79.8~80.2kPa; after pressing, a linear laser scan generates a flatness cloud map (local depression 0.05mm).

[0151] 2. AI re-pressure correction process:

[0152] The pressure module locates the recessed area as a Grid-C7 grid; the two electric cylinders at the corresponding positions are activated to apply gradient pressure (85kPa→90kPa→85kPa); a second scan shows that the recess has been eliminated and the flatness meets the standard (≤0.01mm).

[0153] 3. Health Warning:

[0154] The system detected abnormal pressure fluctuations in electric cylinder No. 3 (standard deviation increased by 30%). Calculations showed that D(t) = 0.6, D˙(t) = 0.0083h−1 (i.e., RUL = (1−0.6) / 0.0083≈48h). The mesh model predicted that the deformation of its supporting structure would exceed the limit within 48 hours. An automatic alarm was pushed: "Replace the guide rail of electric cylinder No. 3, spare part code #E3-7".

[0155] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A photomask optical film adaptive lamination device, characterized in that, include: The operating counter (1) is provided with a P-plate fixing frame (2) and a G-plate fixing frame (3) on its operating surface. The G-plate fixing frame (3) is provided with a multi-cylinder propulsion structure (4). The multi-cylinder propulsion structure (4) includes several telescopic cylinders. The output ends of the telescopic cylinders are all connected to the G-plate clamp. The G-plate clamp is used to install the G-plate, and the output ends of the telescopic cylinders are evenly distributed along the circumference of the edge of the G-plate. The P-plate fixing frame (2) is fixed with a P-plate clamp, which is used to install the P-plate. A control system, wherein the control system is capable of controlling the extension and retraction length of each of the telescopic cylinders respectively; Pressure sensor arrays (5) are provided on the mounting surfaces of both the G-plate fixture and the P-plate fixture, and the multi-cylinder propulsion structure (4) is used to press the G-plate onto the P-plate.

2. The control method for an adaptive pressing device for optical films on a photomask according to claim 1, characterized in that, Includes the following steps: S1: Fix the G plate to the G plate fixture and fix the P plate to the P plate fixture; S2: The control system controls several telescopic cylinders to extend synchronously and press the G plate onto the P plate according to the preset pressure; during the pressing process, the pressure distribution is detected in real time by the pressure sensor array (5) and feedback is given, and the control system adjusts the output pressure of several telescopic cylinders in real time according to the feedback value so that the pressure distribution is uniform. S3: After the first pressing is completed, the pressing surface is scanned by a 3D laser scanner (6) and flatness data is obtained. The control system compares the flatness data with the ideal flatness data and judges whether it meets the standard. S4: When the flatness does not meet the standard, the process of re-pressing is initiated. The defective areas that do not meet the flatness standard are calculated and identified. The control system drives the telescopic cylinder corresponding to the defective area to apply gradient pressure to the defective area until the flatness meets the standard.

3. The control method for an adaptive pressing device for optical films on a photomask according to claim 2, characterized in that, In step S2, the specific steps of pressing the G plate onto the P plate according to the preset pressure include: The screen-printed image of the membrane frame is captured by camera (7); The control system automatically identifies the screen-printed image of the membrane frame and compares it with its own stored historical formula database; the historical formula database includes several stored screen-printed images of membrane frames and their corresponding optimal pressing formulas; When performing similarity comparison, image processing is used to calculate the similarity of text recognition, geometric dimensions and pattern features respectively, and the similarity of the three is fused to obtain the matching degree; Select the image with the highest matching degree and its corresponding optimal pressing formula as the output.

4. The control method for an adaptive pressing device for optical films on a photomask according to claim 2, characterized in that, In step S2, the specific steps by which the control system adjusts the output pressure of several telescopic cylinders in real time according to the feedback value to make the pressure distribution uniform include: Real-time acquisition of pressure sensor data from two pressure sensor arrays (5), and synchronous acquisition of pressure status output from several telescopic cylinders; The stiffness adaptive fusion weights wG and wP are calculated based on the material stiffness characteristics of plates G and P, respectively. Based on the stiffness adaptive fusion weight, a weighted fusion of the two versions of pressure data is performed to obtain accurate pressure distribution data; Based on accurate pressure distribution data, the pressure standard deviation is calculated, and a pressure uniformity evaluation index is obtained based on the pressure standard deviation. The pressure adjustment command for the telescopic cylinder is generated based on the pressure uniformity evaluation index.

5. The control method for an adaptive pressing device for optical films on a photomask according to claim 4, characterized in that, The calculation formulas for the adaptive fusion weights wG and wP are as follows: ； ； The formula for calculating the weighted fusion of the two versions of pressure data is as follows: ； In the above formula: StiffnessG and StiffnessP are the stiffness coefficients of version G and version P, respectively; P sensor G(x,y,t) represents the data from the G-type pressure sensor array, P sensor P(x,y,t) represents the data of the P-type pressure sensor array, where x and y are the sensor grid coordinates and t is the time series.

6. The control method for an adaptive pressing device for optical films on a photomask according to claim 5, characterized in that, The formula for calculating the standard deviation of pressure is: ； The formula for calculating the pressure uniformity evaluation index is as follows: ； In the above formula, m and n are the number of rows and columns of the sensor grid, respectively, and μP(t) is the average pressure value at time t.

7. The control method for an adaptive pressing device for optical films on a photomask according to claim 2, characterized in that, Step S4 further includes the following specific steps: Based on the surface flatness scanning data obtained by the 3D laser scanner (6), the threshold judgment is performed by the generated defect location mask, and the defect area with substandard flatness is output. Calculate the equivalent stiffness of each layer of material in plate P separately to obtain the overall equivalent stiffness of the series connection; Based on the overall equivalent stiffness and the degree of defects in the defect region, the gradient pressure is calculated, and the recommended number of re-pressure cycles is determined based on historical data. Based on the gradient repress pressure and the number of repress times, the repair effect is verified by rescanning after repressing.

8. The control method for an adaptive pressing device for optical films on a photomask according to claim 7, characterized in that, The formula for determining the defect location mask is: ； In the above formula, H scan (x,y) is the flatness scan data matrix, x,y are the scan grid coordinates, and H ideal τ represents the ideal flatness height (mm). defect The defect detection threshold is (mm).

9. The control method for an adaptive pressing device for optical films on a photomask according to claim 8, characterized in that, The formula for calculating the overall equivalent stiffness is: ； In the formula, k eq k is the overall equivalent stiffness coefficient (N / mm). Al k is the stiffness coefficient of the aluminum alloy frame (N / mm). film k is the stiffness coefficient of the cellulose film (N / mm). adhesive The stiffness coefficient of the acrylic soft rubber (N / mm); The formula for calculating gradient pressure is: ； In the formula, F rework (x,y) represents the gradient complex pressure distribution, P base The basic pressure is (kPa), α is the pressure gradient coefficient (5-15 kPa·mm / N), and k eq Equivalent stiffness (N / mm).

10. The control method for an adaptive pressing device for optical films on a photomask according to claim 8, characterized in that, It also includes steps for continuous health prediction of key equipment components and supporting stress components during the pressing process, specifically including: Real-time reading of component stress grid data, equipment monitoring sensor data, and component material fatigue parameters; The equivalent stress of each grid point is calculated using the von Mises equivalent stress formula, and the risk level is classified based on the equivalent stress. Then, the cumulative strain is calculated and the damage model is derived, and the remaining service life is calculated based on the damage model. By combining risk level and remaining useful life, a health prediction result is generated.

Images