A method and system for compensating alignment error of double-sided coating based on machine vision

Abstract

The application discloses a double-sided coating alignment error compensation method and system based on machine vision, and relates to the field of automatic control technology. The application realizes the collaborative compensation of the overall deviation of a base material and the non-linear deformation in the width direction of the base material by detecting the lateral deviation of the base material and the local distortion of the front pattern in different regions, and eliminates the deviation in different regions caused by uneven tension or local stretching. The application realizes the accurate triggering of the longitudinal boundary in the conveying mileage domain by the encoder mileage synchronization and the preset fixed stroke from the camera to the coating head, and overcomes the speed fluctuation and the space stroke conversion error.

Description

Technical Field

[0001] This invention relates to the field of automatic control technology, and in particular to a method and system for compensating for alignment errors in double-sided coating based on machine vision. Background Technology

[0002] Double-sided coating is a core process in lithium-ion battery electrode manufacturing, optical film preparation, and special functional coating processing. The alignment accuracy of the coating patterns on both sides in the substrate transport direction and width direction directly determines the quality and safety of the final product. With the continuous increase in substrate width and operating speed, the alignment control of double-sided coating faces increasingly stringent precision requirements and dynamic response challenges.

[0003] Currently, the detection and compensation of alignment errors in double-sided coating mainly rely on closed-loop control of machine vision and correction actuators. Existing solutions typically use a vision camera to acquire images of the substrate edge or coating pattern to obtain positional deviation information, which then drives the correction actuator to perform lateral displacement compensation. However, existing technologies still have the following shortcomings: lateral correction schemes generally treat the substrate width direction as a rigid whole, ignoring the non-uniform lateral deformation caused by uneven tension, material tolerances, or local stretching, making it impossible to achieve fine-grained compensation in different areas when there are significant differences in offset between the edge and center regions; longitudinal alignment schemes rely on the theoretical design distance of the camera installation position, failing to effectively solve the spatial travel conversion problem between the front vision detection position and the reverse coating execution position, with alignment accuracy limited by speed fluctuations and cumulative mileage errors; furthermore, existing solutions fail to effectively distinguish between overall substrate misalignment and local deformation, resulting in insufficient targeting and adaptability of compensation strategies. Summary of the Invention

[0004] The purpose of this section is to outline some aspects of the embodiments of the present invention and to briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section, as well as in the abstract and title of the present application, to avoid obscuring the purpose of this section, the abstract and title of the invention. Such simplifications or omissions shall not be used to limit the scope of the present invention.

[0005] In view of the aforementioned existing problems, the present invention is proposed.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a machine vision-based method for compensating alignment errors in double-sided coating, comprising: acquiring longitudinal boundary and lateral edge data of the front coating pattern as a following reference for the reverse coating; detecting the lateral offset of the substrate within each width partition, and calculating the local distortion control amount of each width partition based on the position of the front pattern edge and the substrate edge; calculating the average value of the lateral offset of all width partitions, and comparing the offset difference between the edge partition and the center partition: if it is less than a set value, the overall correction control amount is taken as the average value and the fine-tuning control amount of each width partition is taken as the distortion control amount; otherwise, the overall correction control amount is still taken as the average value, and the fine-tuning control amount of the edge partition is superimposed on the corresponding width partition. The difference between the lateral offset of the partition and the mean is used to determine the distortion control amount for the fine-tuning control of the remaining width partitions. For reverse coating, the nozzle opening trigger position and nozzle closing trigger position of each width partition are determined based on the leading edge boundary position and trailing edge boundary position of each width partition in the longitudinal boundary, the preset fixed stroke from the camera to the coating head, and the encoder mileage. When the encoder's real-time accumulated mileage reaches the nozzle opening trigger position, the coating nozzle of the corresponding width partition is opened, and when it reaches the nozzle closing trigger position, the coating nozzle of the corresponding width partition is closed. At the same time, based on the overall correction control amount and the fine-tuning control amount of each width partition, the overall correction actuator and the fine-tuning actuator corresponding to each width partition are controlled to perform lateral displacement.

[0007] In a preferred embodiment of the present invention, the step of obtaining the longitudinal boundary and lateral edge data of the front coating pattern includes: acquiring an image of a substrate with completed front coating using a front vision camera; dividing the substrate image into multiple width partitions along the substrate width direction; extracting the leading edge and trailing edge positions of the front coating pattern in each width partition in the substrate conveying direction as longitudinal following reference data; and extracting the left edge and right edge positions of the front coating pattern in each width partition in the substrate width direction as lateral following reference data.

[0008] As a preferred embodiment of the present invention, the step of detecting the lateral offset of the substrate within each width partition includes: acquiring images of the physical edges of the substrate at each width partition using a reverse vision camera in front of the reverse coating head; detecting the current pixel position of the physical edge of the substrate in the image; and comparing the current pixel position with the theoretical pixel position of the substrate in a pre-calibrated state without offset to obtain the overall lateral offset of the substrate within each width partition.

[0009] As a preferred embodiment of the present invention, the step of calculating the local distortion compensation amount of each width partition based on the position of the front pattern edge and the substrate edge includes: pre-calibrating a fixed design margin between the front coating pattern edge and the physical edge of the substrate in each width partition; determining the theoretical expected position of the reverse coating target edge at the current substrate position based on the current actual position of the physical edge of the substrate in each width partition and the fixed design margin; comparing the actual edge position of the front coating pattern in each width partition with the theoretical expected position to obtain the local distortion compensation amount introduced by the substrate width change or the previous coating error in each width partition.

[0010] In a preferred embodiment of the present invention, determining the overall correction amount and the fine-tuning amount of each width zone based on the comparison results includes: calculating the arithmetic mean of the overall lateral offset of the substrate within all width zones, and using the arithmetic mean as the overall correction amount; determining the left edge zone, right edge zone, and center zone in the width direction of the substrate; calculating a first difference in the lateral offset between the left edge zone and the center zone, and a second difference in the lateral offset between the right edge zone and the center zone; when both the first difference and the second difference are less than a set value, the fine-tuning amount of each width zone is adjusted. The adjustment amount is set to the local distortion compensation amount corresponding to each width partition; when either the first difference or the second difference is greater than or equal to the set value, the fine adjustment amount of the left edge partition is set to the difference between the horizontal offset of the left edge partition and the arithmetic mean, plus the local distortion compensation amount corresponding to the left edge partition; the fine adjustment amount of the right edge partition is set to the difference between the horizontal offset of the right edge partition and the arithmetic mean, plus the local distortion compensation amount corresponding to the right edge partition; and the fine adjustment amount of all middle width partitions is set to the local distortion compensation amount corresponding to each middle width partition.

[0011] In a preferred embodiment of the present invention, the set value is a judgment threshold obtained by pre-calibration through static repeatability testing. The static repeatability testing involves collecting multiple sets of sample data on the absolute value of the difference in lateral offset between the central partition and the edge partition within a preset period of continuous and stable operation of the device, and determining the judgment threshold based on the statistical distribution characteristics of the multiple sets of sample data.

[0012] As a preferred embodiment of the present invention, the triggering of the longitudinal coating start and end of each width partition includes: pre-determining a fixed physical distance along the substrate conveying direction between the image acquisition position of the front vision camera and the coating execution position of the reverse coating head; real-time acquisition of the cumulative travel distance of the substrate conveying; determining the image domain distance corresponding to the longitudinal boundary of each width partition based on the leading edge boundary position and the trailing edge boundary position of each width partition; superimposing the fixed physical distance and the image domain distance, and combining it with the reference travel distance corresponding to the moment when the front vision camera acquires the substrate image, determining the nozzle opening trigger position and nozzle closing trigger position of the reverse coating in each width partition; when the cumulative travel distance reaches the nozzle opening trigger position, opening the coating nozzle of the corresponding width partition; when the cumulative travel distance reaches the nozzle closing trigger position, closing the coating nozzle of the corresponding width partition.

[0013] In a preferred embodiment of the present invention, the reference travel distance is obtained by synchronous latching of the encoder; the synchronous latching refers to synchronously reading and locking the current cumulative pulse value of the transmission encoder at the start of exposure of the substrate image acquired by the front vision camera, and converting the latched current cumulative pulse value into a physical travel distance as the reference travel distance.

[0014] In a preferred embodiment of the present invention, the step of driving the overall correction actuator and the fine-tuning actuators corresponding to each partition includes: converting the overall correction amount into a physical lateral overall displacement amount, driving the overall correction actuator to perform a lateral overall displacement to compensate for the overall lateral deviation of the substrate; converting the fine-tuning amount of each width partition into a physical lateral local displacement amount corresponding to each width partition, driving the fine-tuning actuator corresponding to each width partition to perform an independent local lateral displacement, wherein the fine-tuning actuators are set in a one-to-one correspondence with the width partitions; through the combined action of the overall displacement of the overall correction actuator and the local displacement of the fine-tuning actuators of each width partition, the longitudinal boundary of the reverse coating in each width partition is aligned with the longitudinal boundary of the front coating pattern, and the lateral edge of the reverse coating in each width partition follows the lateral edge position of the front coating pattern.

[0015] On the other hand, a machine vision-based double-sided coating alignment error compensation system is also provided, including: a front reference extraction module, used to acquire the longitudinal boundary and lateral edge data of the front coating pattern as a following reference for the reverse coating; a partition distortion calculation module, used to detect the lateral offset of the substrate within each width partition, and calculate the local distortion control amount of each width partition according to the position of the front pattern edge and the substrate edge; and an error diversion decision module, used to calculate the average value of the lateral offset of all width partitions, and compare the offset difference between the edge partition and the center partition: if it is less than a set value, the overall correction control amount is taken as the average value and the fine-tuning control amount of each width partition is taken as the distortion control amount; otherwise, the overall correction control amount is still taken as the average value, and the fine-tuning control amounts of the edge partitions are superimposed. The difference between the lateral offset and the mean of the corresponding width partition is used as the distortion control value for the fine-tuning control of the remaining width partitions. The follow-up coating execution module is used for reverse coating. Based on the leading edge boundary position and trailing edge boundary position of each width partition in the longitudinal boundary, the preset fixed stroke from the camera to the coating head and the encoder mileage, the nozzle opening trigger position and nozzle closing trigger position of each width partition are determined respectively. When the encoder's real-time accumulated mileage reaches the nozzle opening trigger position, the coating nozzle of the corresponding width partition is opened, and when it reaches the nozzle closing trigger position, the coating nozzle of the corresponding width partition is closed. At the same time, based on the overall correction control value and the fine-tuning control value of each width partition, the overall correction actuator and the fine-tuning actuator corresponding to each width partition are controlled to perform lateral displacement.

[0016] The beneficial effects of this invention are as follows: This invention achieves coordinated compensation for overall substrate misalignment and nonlinear deformation in the width direction by detecting lateral offset of the substrate and local distortion of the front pattern in partitioned detection, eliminating partitioned deviations caused by uneven tension or local stretching; by synchronizing encoder mileage and setting a fixed stroke from the camera to the coating head, precise triggering of the longitudinal boundary is achieved in the conveying mileage domain, overcoming speed fluctuations and spatial stroke conversion errors; at the same time, the control strategy is adaptively switched according to the degree of difference between the offset of the edge partition and the center partition, taking into account both uniform deviation stability correction and targeted compensation for nonlinear deformation, effectively reducing the coating defect rate, and is especially suitable for high-precision double-sided alignment control in wide-width high-speed coating scenarios. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein: Figure 1 This is the process for acquiring the outline data of the front coating pattern as shown in the present invention.

[0018] Figure 2This is the process for generating the lateral offset detection and compensation strategy as shown in this invention. Detailed Implementation

[0019] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0020] Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort should fall within the scope of protection of this invention.

[0021] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0022] According to an embodiment of the present invention, in combination Figures 1-2 The flowchart shown illustrates a machine vision-based method for compensating alignment errors in double-sided coating, comprising: S1: Obtain the longitudinal boundary and lateral edge data of the front coating pattern as a reference for the reverse coating.

[0023] S1.1: After the front coating process, a front visual inspection station is set up along the substrate conveying direction. A front visual camera is installed at the front visual inspection station, with its optical axis perpendicular to the substrate width and its field of view covering at least the entire width of the substrate. The front visual camera acquires an image of the substrate after front coating and transmits the acquired image to the image processing unit.

[0024] S1.2: After receiving the substrate image, the image processing unit divides the substrate image into N equal-width partitions along the width direction of the substrate. The value of N matches the number of coating nozzles or coating units arranged along the width direction of the reverse coating head. Each width partition is arranged continuously along the width direction of the substrate, covering the entire width range of the substrate, and a one-to-one spatial position mapping relationship is formed between each width partition and each coating nozzle of the reverse coating head.

[0025] It should be noted that the lateral offset distribution of the substrate in the width direction is usually not a rigid translation. The offset amount in the central region and the edge region of the substrate may differ, and the width of the substrate itself may also change locally due to tension fluctuations or material tolerances. By dividing the width direction of the substrate into multiple independent zones, subsequent lateral offset detection can be performed on a zone-by-zone basis, thereby achieving refined detection of non-uniform lateral offset.

[0026] S1.3: For each width partition, the image processing unit scans the grayscale values ​​of the image pixels within that partition line by line along the substrate conveying direction, detecting the position of the pixel row where the grayscale value changes. When the pixel grayscale value changes from the substrate background grayscale value to the coating area grayscale value, this position is determined as the leading edge boundary position of the front coating pattern within that width partition; when the pixel grayscale value changes from the coating area grayscale value to the substrate background grayscale value, this position is determined as the trailing edge boundary position of the front coating pattern within that width partition.

[0027] The grayscale transition detection mentioned above can be implemented using edge detection operators, including but not limited to gradient operators, Laplacian operators, or Canny edge detection operators.

[0028] Record the row coordinates of the leading pixel of each width partition i. and trailing edge pixel row coordinates The pixel row coordinates represent the position of the boundary in the image coordinate system along the substrate transport direction. The leading and trailing pixel row coordinates of all width partitions are summarized to form the longitudinal following reference data.

[0029] S1.4: For each width partition, the image processing unit scans the pixel grayscale values ​​within the vertical range of the front coating pattern in that partition column by column along the width direction of the substrate, and detects the position of the pixel column where the grayscale value changes. When the pixel grayscale value changes from the grayscale value of the uncoated area to the grayscale value of the coated area, the position is determined to be the left edge position of the front coating pattern in that width partition; when the pixel grayscale value changes from the grayscale value of the coated area to the grayscale value of the uncoated area, the position is determined to be the right edge position of the front coating pattern in that width partition.

[0030] Record the left edge pixel column coordinates of each width partition i and the right edge pixel column coordinates The pixel column coordinates represent the position of the edge along the width direction of the substrate in the image coordinate system. The pixel column coordinates of the left and right edges of all width partitions are combined to form the lateral following reference data.

[0031] S1.5: Pre-calibrate and store the fixed design margin between the edge of the front coating pattern and the physical edge of the substrate within each width partition, including the fixed design margin between the left edge of the front coating pattern and the physical left edge of the substrate within each width partition. And the fixed design margin between the right edge and the physical right edge of the substrate. .

[0032] The fixed design margin is determined by the design parameters of the coating process. Specifically, it is the nominal distance in the width direction between the edge of the front coating pattern and the physical edge of the substrate when the substrate has no lateral offset and is in the standard tape-running position. The fixed design margin is pre-stored in the storage unit of the control system in the form of a data table, corresponding one-to-one with the partition index of each width zone, and is used in subsequent steps to calculate the local distortion compensation amount of each width zone.

[0033] S1.6: The longitudinal following reference data, lateral following reference data, and fixed design margins are summarized to form the following reference data for reverse coating. The following reference data is organized by width partition as an index. The data corresponding to each width partition includes at least the partition index, leading edge pixel row coordinates, trailing edge pixel row coordinates, left edge pixel column coordinates, right edge pixel column coordinates, left fixed design margin, and right fixed design margin. The following reference data is stored in the storage unit of the control system for the reverse coating control unit to access in subsequent processes.

[0034] Through the above method, this embodiment obtains precise positional references for the front coating pattern at each partition across the full width. By finely partitioning the width, subsequent lateral offset detection can be performed independently on a partition-by-part basis, overcoming the shortcomings of traditional methods that treat the substrate as a rigid whole, resulting in the inability to detect and compensate for non-uniform offsets. This provides an accurate data foundation for improving the alignment accuracy of double-sided coating.

[0035] S2: Detect the lateral offset of the substrate within each width zone, and calculate the local distortion control amount of each width zone based on the position of the front pattern edge and the substrate edge.

[0036] S2.1: A reverse vision inspection station is set up in front of the reverse coating head (upstream of the reverse coating head along the substrate conveying direction). A reverse vision camera is installed at the reverse vision inspection station. The optical axis of the reverse vision camera is perpendicular to the substrate width, and its field of view covers at least the edge area of ​​the substrate width direction.

[0037] It should be noted that, unlike the front vision camera which needs to cover the entire width of the substrate, the main detection object of the back vision camera is the physical edge of the substrate. Since the physical edge of the substrate is located on both sides of the substrate width, two independent cameras can be used to perform local imaging of the left and right edges of the substrate respectively, or a wide-angle camera can be used to image both edges of the substrate simultaneously.

[0038] A reverse vision camera captures grayscale images of the physical edges of the substrate at each width section in front of the reverse coating head. In the grayscale images, there is a significant grayscale difference between the substrate area and the background area, and the physical edges of the substrate are represented in the image as pixel columns where the grayscale value changes abruptly.

[0039] In this embodiment, the reverse vision camera captures the physical edge image of the substrate before the reverse coating head. This is to obtain the lateral offset state of the substrate detected by the reverse vision camera, specifically the actual offset state of the substrate as it approaches the reverse coating head. Lateral deviation control based on this information can minimize additional errors introduced by substrate jitter or guide roller misalignment between the detection and execution positions.

[0040] S2.2: The reverse vision camera transmits the acquired grayscale image to the image processing unit. The image processing unit analyzes the grayscale image, scanning the pixel regions corresponding to each width partition in the grayscale image column by column along the width direction of the substrate, and detecting the column coordinates of pixels where the grayscale value jumps.

[0041] Specifically, for detecting the left edge of the substrate, the image is scanned from left to right. When a pixel grayscale value jumps from the background area (low grayscale value) to the substrate area (high grayscale value), the jump position is the pixel column coordinate of the physical left edge of the substrate in the current image. For detecting the right edge of the substrate, the image is scanned from right to left. When a pixel grayscale value jumps from the background area to the substrate area, the jump position is the pixel column coordinate of the physical right edge of the substrate in the current image.

[0042] Since the substrate is divided into multiple width sections along its width direction, the physical edge position of the substrate within each width section may differ due to variations in the substrate's width or local wrinkles. Therefore, this embodiment detects the current pixel column coordinates of the physical edge of the substrate corresponding to each width section i. Specifically, for the i-th width section, the pixel column coordinates of the physical edge of the substrate at the corresponding longitudinal position in the substrate transport direction are detected and recorded as the current edge pixel column coordinates of each width section. .

[0043] Current edge pixel column coordinates This represents the actual pixel position of the physical edge of the substrate at the i-th width partition in the image coordinate system along the width direction under the current substrate transport state.

[0044] S2.3: Obtain the current edge pixel column coordinates of each width partition. Then, subtract the pre-defined theoretical edge pixel column coordinates corresponding to each width partition from the current edge pixel column coordinates of each width partition. The resulting difference is used as the overall lateral offset of the substrate within each width zone. .

[0045] Theoretical edge pixel column coordinates This refers to the calibration coordinates of the physical edge of the substrate in the image when the substrate has no lateral offset and is in the standard conveyor belt position. These calibration values ​​are pre-acquired during the equipment commissioning phase. Specifically, under the standard conveyor belt state with no substrate offset, images of the substrate edge are acquired using a reverse vision camera. The pixel column coordinates of the physical edge of the substrate at each width partition are detected and recorded as theoretical values ​​and stored in the control system's storage unit.

[0046] The sign of the difference indicates the direction of the substrate's lateral offset, and the absolute value of the difference indicates the magnitude of the offset. For example, if the current edge pixel column coordinates are greater than the theoretical edge pixel column coordinates, it indicates a positive offset of the substrate in the image width direction; conversely, it indicates a negative offset of the substrate in the image width direction. This direction definition corresponds to the execution direction of subsequent correction control. When the overall lateral offset is positive, the overall correction actuator moves in the opposite direction to pull the substrate back to the standard position.

[0047] It should be noted that the overall lateral offset It was measured separately for each width section, and the measurements were taken between each section. There may be differences. When each partition... When they are basically consistent, it indicates that the substrate has undergone overall rigid translation; when each zone is consistent... When significant differences exist between the two, it indicates that the substrate exhibits non-uniform deformation in the width direction, such as localized stretching, wrinkling, or width variation. This distinction is crucial for subsequently determining the corrective control strategy.

[0048] S2.4: Before calculating the local distortion compensation amount for each width partition, the image processing unit retrieves the pre-calibrated fixed design margin for each width partition from the storage unit of the control system, that is, the fixed design margin between the left edge of the front coating pattern and the physical left edge of the substrate within each width partition. And the fixed design margin between the right edge and the physical right edge of the substrate. .

[0049] The calibration process for fixed design margins has been described in detail in Section S1 and will not be repeated here. Fixed design margins are stored using partition indices as keys; during retrieval, the corresponding margin is read based on the currently processed partition index i. and Numerical value.

[0050] S2.5: Based on the current actual position of the physical edge of the substrate within each width zone and the fixed design margin, determine the theoretical expected position of the target edge for reverse coating at the current substrate position.

[0051] In this embodiment, the theoretically desired position of the target edge of the reverse coating is specifically defined as follows: if the positional relationship between the reverse coating and the physical edge of the substrate is completely consistent with that of the front coating, that is, the positional deviation of the reverse coating pattern relative to the physical edge of the substrate is equal to the positional deviation of the front coating pattern relative to the physical edge of the substrate, then the reverse coating should be applied to the theoretically desired position on the substrate surface. Specifically: For the left edge, the theoretically expected position is the current actual position of the physical left edge of the substrate within each width partition plus the fixed design margin on the left. That is, the left edge of the reverse coating target should be located inside the physical left edge of the substrate at a distance of [missing information]. The theoretical expected position is located at the given position. In the image coordinate system, this is expressed as: + ,in The lateral pixel equivalent (unit: mm / pixel) for the image acquisition system is used to convert fixed design margins from physical dimensions to the pixel domain. The lateral pixel equivalent is a fixed parameter pre-calibrated by the system. The specific calibration method is as follows: under standard conditions, a calibration block of known size is photographed, and the lateral pixel equivalent is determined based on the ratio of the actual physical size of the calibration block to its pixel size in the image. The value.

[0052] For the right edge, the theoretically expected position is the current actual position of the physical right edge of the substrate within each width partition minus the fixed design margin on the right side. That is, the right edge of the reverse coating target should be located inside the physical right edge of the substrate at a distance of [missing information]. At the location.

[0053] Due to fixed design margins and This reflects the nominal positional relationship of the front coating relative to the physical edge of the substrate. However, the substrate may have already shifted laterally. Therefore, the theoretical expected position is the target position that the edge of the reverse coating should reach in order to maintain the same relative positional relationship with the front coating under the current substrate position.

[0054] S2.6: Compare the actual edge position of the front coating pattern in each width partition with the theoretical expected position to obtain the amount of local distortion compensation introduced by the substrate width change or the previous coating error in each width partition.

[0055] Specifically, for the i-th width partition, the actual edge position of the front coating pattern is the left edge pixel column coordinate extracted in section S1. and the right edge pixel column coordinates Calculate separately: Left-side local distortion compensation amount: ,in, The theoretical expected position on the left side, calculated in Section 2.5; and the local distortion compensation amount on the right side. ,in This is the theoretical expected position on the right side calculated in Section 2.5.

[0056] Since the left and right edges of the front-coated pattern may shift within each width partition, in a preferred embodiment, the average of the left and right local distortion compensation amounts is taken as the local distortion compensation amount for that width partition. .

[0057] Among them, local distortion compensation amount The meaning of positive and negative signs and the overall horizontal offset Maintain consistency. When When the value is positive, it indicates that the front coating pattern is biased towards the positive direction relative to the current physical edge of the substrate, and the subsequent local fine-tuning actuator should move in the opposite direction to compensate for this deviation; when When the value is negative, fine-tuning is performed in the opposite direction.

[0058] As can be seen, this embodiment compares the actual edge position of the front coating pattern with the theoretical expected position calculated based on the fixed design margin and the current substrate edge position. It quantifies factors such as substrate width variation, local tensile deformation, and coating position error of the previous coating process into a local distortion compensation amount. This allows the subsequent reverse coating lateral correction to simultaneously eliminate two different types of deviations: overall substrate misalignment and local substrate deformation, achieving fine alignment across the entire width range.

[0059] S3: Calculate the average value of the lateral offset of all width zones, and compare the offset difference between the edge zone and the center zone: if it is less than the set value, the overall correction control amount is taken as the average value and the fine-tuning control amount of each width zone is taken as the distortion control amount; otherwise, the overall correction control amount is still taken as the average value, the fine-tuning control amount of the edge zone is added to the difference between the lateral offset of the corresponding width zone and the average value, and the fine-tuning control amount of the remaining width zones is taken as the distortion control amount.

[0060] S3.1: In step S2, the overall lateral offset of the substrate within each wide section was obtained by detecting it using a reverse vision camera. (i=1,2,...,N). Before performing the control logic calculations in this step, the image processing unit or control system first calculates the overall lateral offset of all N width partitions. The data is then aggregated to form a complete horizontal offset dataset.

[0061] The horizontal offset dataset is arranged in partition index order, and each data entry contains at least the partition index i and the overall horizontal offset corresponding to that partition. During the aggregation process, optionally, a validity check can be performed on the lateral offset dataset to remove invalid data points caused by abnormal image acquisition or edge detection failure. When the lateral offset of a certain partition exceeds a preset reasonable range, the data point is determined to be invalid, removed, and replaced with the interpolated value of the lateral offset of the adjacent partition to ensure the stability and reliability of subsequent calculations.

[0062] S3.2: The control system calculates the overall lateral offset of all N width zones. The arithmetic mean, denoted as .

[0063] When the substrate as a whole undergoes uniform lateral deviation, each zone... Both are close to When the substrate exhibits non-uniform deformation in the width direction, the deformation of each zone... around They exhibit varying degrees of discrete distribution.

[0064] Arithmetic mean As the overall correction amount in subsequent steps The base value is used to drive the overall correction actuator to compensate for the lateral deviation of the substrate. Regardless of the subsequent judgment result, that is, whether the substrate is in a uniform deviation state or a nonlinear deformation state, the overall correction amount is... The average value is This ensures that the overall lateral position of the substrate is always controlled near the standard conveyor belt position.

[0065] S3.3: Before comparing the offset difference between the edge partition and the center partition, the control system first determines the position of each key partition in the width direction: For the left edge partition, its partition index is i=1, which is the first width partition divided along the width direction of the substrate, corresponding to the left edge region of the substrate width direction; for the right edge partition, its partition index is i=N, which is the last width partition divided along the width direction of the substrate, corresponding to the right edge region of the substrate width direction.

[0066] For the center partition, when the total number of width partitions N is odd, the partition index of the center partition is i = (N+1) / 2; when the total number of width partitions N is even, the partition index of the center partition is i = N / 2, that is, the one closer to the left of the two middle partitions, or i = N / 2+1, that is, the one closer to the right of the two middle partitions. In this embodiment, for the sake of consistency, it is pre-defined that when N is even, the center partition is i = N / 2.

[0067] The aforementioned key partitions are pre-defined during the equipment commissioning phase and stored as configuration parameters in the control system's storage unit. During subsequent operation, the control system directly calls these configuration parameters without recalculation.

[0068] It should be noted that in the position representation in the width direction, the substrate width is observed with the substrate conveying direction as the reference. The identification of the left edge and the right edge is related to the observation direction, but the definition of the operating side and the non-operating side in the actual coating equipment shall prevail. This embodiment does not limit this, as long as the definition of the left edge partition and the right edge partition is consistent in the whole control logic.

[0069] S3.4: The control system calculates the absolute value of the difference in lateral offset between the left edge partition and the center partition, and the absolute value of the difference in lateral offset between the right edge partition and the center partition.

[0070] Specifically, calculate the absolute value of the first difference. ,in This represents the overall horizontal offset of the left edge partition. This represents the overall lateral offset of the central region. The absolute value of this first difference characterizes the degree of lateral offset difference between the left edge region and the central region in the width direction of the substrate.

[0071] Calculate the absolute value of the second difference ,in This represents the overall lateral offset of the right edge region. The absolute value of this second difference characterizes the degree of lateral offset difference between the right edge region and the center region in the width direction of the substrate.

[0072] The larger of the absolute values ​​of the first and second differences is taken as the deviation judgment value: Among them, the deviation judgment value It characterizes the maximum degree of lateral offset difference between the edge region and the center region in the width direction of the substrate.

[0073] It should be noted that when the substrate is rigidly translated as a whole, the lateral offset of each width section is basically equal, and the difference between the edge section and the center section is small. When the substrate experiences nonlinear deformation such as local stretching, wrinkling, or width variation in the width direction, the offset between the edge region and the center region will differ, and the greater the difference, the more severe the nonlinear deformation of the substrate. By comparing the deviation judgment value with the preset set value, it is possible to quantitatively determine whether the substrate is in a uniform deviation state or a nonlinear deformation state, thereby determining which correction control strategy to adopt.

[0074] S3.5: When the deviation judgment value When the value is less than the preset storage value T (i.e. the set value), the control system determines that the lateral offset distribution in the width direction of the substrate is basically uniform, the substrate is in an overall rigid translation state, and there is no significant nonlinear deformation.

[0075] In this case, the control system executes the following control logic: Overall correction amount Set as the arithmetic mean of the lateral offsets of all width zones. Fine-tuning of each width section (i=1,2,...,N) are all set as the local distortion compensation amount corresponding to each width partition. .

[0076] Under the control strategy of the first branch, the overall lateral deviation of the substrate is compensated by the overall correction actuator through overall displacement, while the local pattern misalignment of each width zone is compensated by the corresponding fine-tuning actuator of each zone through independent local displacement. Overall correction and local fine-tuning target different types of deviation sources, do not interfere with each other, and work synergistically.

[0077] The rationale behind the above control logic lies in the fact that when the substrate is in a uniform deviation state, the lateral offset of each zone is basically the same, and the arithmetic mean of the lateral offset of all width zones is... It can accurately reflect the actual deviation of the entire substrate. At this time, the main correction actuator follows... The substrate can be pulled back to the standard transport position by performing overall displacement, without the need for additional nonlinear compensation for specific zones.

[0078] S3.6: When the deviation judgment value When the value is greater than or equal to the preset storage value T, the control system determines that there is a significant nonlinear deformation in the width direction of the substrate, that is, there is a substantial difference in the lateral offset between the edge area and the center area of ​​the substrate, and the substrate may have undergone local stretching, wrinkling or width change due to uneven tension.

[0079] In this case, the control system executes the following control logic: Overall correction amount It is still set to the arithmetic mean of the lateral offsets of all width zones. That is, regardless of whether the substrate undergoes nonlinear deformation, the overall correction actuator always performs overall displacement according to the average value of the lateral offset of all width zones, so as to eliminate the average deviation of the substrate as a whole.

[0080] For the left edge width partition (i=1), its fine-tuning amount Set as: Overall lateral offset of the left edge width partition Subtract the arithmetic mean The difference obtained is then superimposed with the local distortion compensation amount corresponding to the left edge width partition. .Right now .

[0081] For the right edge width partition (i=N), the fine adjustment amount Set as: Overall horizontal offset of the right edge width partition Subtract the arithmetic mean The difference obtained is then superimposed with the local distortion compensation amount corresponding to the right edge width partition. .

[0082] For all intermediate width partitions (i.e., i=2,3,...,N-1) except for the left and right edge width partitions, including the center width partition, the fine-tuning amount is... All are set as the local distortion compensation amount corresponding to each intermediate width zone. ,Right now .

[0083] Under the control strategy of the second branch, the fine-tuning amount obtained by each width partition consists of two parts: the first part is the difference between the overall lateral offset and the mean. The first part is used to compensate for the additional offset between this section and the overall average deviation caused by the nonlinear deformation of the substrate; the second part is the local distortion compensation amount. This is used to compensate for pattern misalignment caused by variations in substrate width or errors in previous coating processes. For the middle width partition, since the nonlinear deformation of the substrate typically manifests as significantly greater deformation in the edge areas than in the middle areas, the lateral offset of the middle width partition has been... The overall correction has been basically compensated, so no additional layering is needed. For this item, only the local distortion compensation amount needs to be retained. That's all.

[0084] It should be noted that, The value can be positive or negative, depending on the direction of the actual lateral offset of the partition relative to the average of all width partitions. When it is positive, the fine-tuning amount of this partition is... On the basis of increasing positively; when When the value is negative, the fine-tuning amount for this partition is... It decreases in the negative direction based on the existing structure.

[0085] It should be noted that this embodiment distinguishes the operating state of the substrate into a uniform deviation state and a nonlinear deformation state by comparing the lateral offset difference between the edge partition and the center partition, and formulates differentiated control strategies for the two states. When the substrate is in a nonlinear deformation state, an additional nonlinear compensation amount is applied only to the edge partition where the deformation is most significant, while the compensation amount is not superimposed on the middle partition. This ensures both the accuracy and specificity of the control, while avoiding control oscillations caused by overcompensation, thus achieving a balance between global stability and local accuracy.

[0086] S3.7: The preset stored value T (i.e. the set value) is a judgment benchmark value obtained by pre-calibration through static repeatability test, which is the critical threshold for distinguishing between the uniform deviation state and the nonlinear deformation state of the substrate.

[0087] The specific method for static repeatability testing is as follows: within a preset duration of continuous and stable operation of the equipment, multiple sets of sample data are collected, representing the absolute values ​​of the lateral offset differences between the center and edge width partitions. For example, under normal operating conditions, M sets of data (M≥100) are continuously collected at a fixed sampling period. Each set of data includes the lateral offset of each width partition, and the absolute value of the difference between the edge and center partitions in each set of data is calculated to form a sample set.

[0088] The preset storage value T is determined based on the statistical distribution characteristics of the sample set. In a preferred embodiment, the mean and standard deviation of the sample set are calculated, and the preset storage value T is set as the mean of the sample set plus three times the standard deviation, i.e., T = μ + 3σ, where μ is the mean of the sample set and σ is the standard deviation of the sample set. The above value selection method is based on the statistical control principle, and the deviation judgment value at a certain moment... When the value exceeds μ+3σ, it indicates that the deformation state of the substrate at that moment is a low-probability abnormal event in a statistical sense, and should be judged as a nonlinear deformation state and the corresponding control strategy should be initiated.

[0089] S3.8: Fine-tuning amount for each width partition in the above two control strategy branches. All measurements are in pixels, and their sign definitions are consistent with the overall horizontal offset. Maintain consistency. That is, when the fine-tuning amount is positive, it means that fine-tuning needs to be performed in the positive direction of the image width; when the fine-tuning amount is negative, it means that fine-tuning needs to be performed in the negative direction of the image width.

[0090] Overall correction amount and the fine-tuning amount of each width zone (i=1,2,...,N) are aggregated to form a lateral correction control instruction set. The control instruction set is arranged in partition index order, and each data entry contains at least the partition index i and the corresponding fine-tuning amount for that partition. and the overall global correction amount The control command set is transmitted to the reverse coating control unit, which is used to drive the main correction actuator and the fine-tuning actuators corresponding to each zone in subsequent steps.

[0091] S4: For reverse coating, based on the leading edge and trailing edge positions of each width partition in the longitudinal boundary, the preset fixed travel distance from the camera to the coating head, and the encoder mileage, the nozzle opening trigger position and nozzle closing trigger position of each width partition are determined respectively. When the encoder's real-time accumulated mileage reaches the nozzle opening trigger position, the coating nozzle of the corresponding width partition is opened, and when it reaches the nozzle closing trigger position, the coating nozzle of the corresponding width partition is closed. At the same time, based on the overall correction control amount and the fine-tuning control amount of each width partition, the overall correction actuator and the fine-tuning actuator corresponding to each width partition are controlled to perform lateral displacement respectively.

[0092] S4.1: Before executing reverse coating control, the control system predetermines and stores the following system parameters: First, a fixed physical distance along the substrate conveying direction is pre-determined between the image acquisition position of the front vision camera and the coating execution position of the back coating head, denoted as . Fixed physical spacing refers to the actual physical distance along the substrate conveying direction between the center line of the photosensitive surface of the front vision camera and the spray point of the nozzle of the reverse coating head. This distance is determined by the mechanical structure of the equipment and is obtained through actual measurement during the equipment installation and commissioning phase, with the unit being millimeters (mm). Since both the front vision camera and the reverse coating head are fixedly mounted on the equipment frame, their relative positions remain unchanged during equipment operation; therefore, the fixed physical spacing... It is a constant value.

[0093] Secondly, the vertical pixel equivalent of the image acquisition system is pre-calibrated. The longitudinal pixel equivalent represents the actual physical distance of each pixel row in the image along the physical transport direction of the substrate, measured in millimeters per pixel (mm / pixel). Specifically, under standard conditions, two marker points with a known physical distance are set on the substrate surface along the transport direction. An image containing the two marker points is acquired using a front-view camera. The difference in pixel row coordinates between the two marker points in the image is calculated. Dividing the known physical distance by the pixel row coordinate difference yields the longitudinal pixel equivalent. .

[0094] Pre-calibrate the lateral pixel equivalent of the image acquisition system The lateral pixel equivalent represents the actual physical distance of each pixel column in an image along the width of the substrate, measured in millimeters per pixel (mm / pixel). The calibration method for the lateral pixel equivalent is as follows: Under standard conditions, two marker points with a known physical distance are set on the substrate surface along the width direction. An image containing the two marker points is acquired using a front-view camera. The difference in pixel column coordinates between the two marker points in the image is calculated. The lateral pixel equivalent is obtained by dividing the known physical distance by the difference in pixel column coordinates. .

[0095] The aforementioned system parameters are pre-stored in the storage unit of the control system and can be retrieved and used by the control system at any time during equipment operation.

[0096] S4.2: The purpose of reverse coating longitudinal start and stop control is to precisely align the leading and trailing edges of the reverse coating pattern in the substrate conveying direction with the leading and trailing edges of the front coating pattern. To achieve this objective, this embodiment employs a synchronous triggering method based on encoder mileage.

[0097] Specifically, when the front vision camera acquires an image of the substrate, the encoder synchronously latches the reference position of the substrate segment corresponding to that frame on the transport path. Since there is a fixed physical distance between the front vision camera and the reverse coating head, when the substrate segment moves along the transport path to the position of the reverse coating head, the control system calculates the nozzle opening and closing positions corresponding to each width partition based on the reference position, the fixed physical distance, and the pixel row coordinates of the longitudinal boundary of the front pattern in the image. When the encoder's accumulated mileage reaches the above positions, the control system triggers the coating nozzles of the corresponding width partition to perform opening or closing actions.

[0098] S4.3: In this embodiment, a conveying encoder is provided on the substrate conveying path. The conveying encoder is coaxially connected to the substrate conveying drive roller or follower roller. As the substrate is continuously conveyed, the conveying encoder outputs a pulse signal proportional to the conveying distance.

[0099] The control system reads the cumulative pulse value of the conveyor encoder in real time, and multiplies the cumulative pulse value by the encoder pulse equivalent (unit: mm / pulse) to convert it into the physical cumulative travel distance L(t) of the substrate, where t represents the current time.

[0100] The encoder pulse equivalent is pre-calibrated during the equipment commissioning phase. The driving substrate is transported a certain physical distance, and the total number of pulses output by the encoder during this process is recorded. The encoder pulse equivalent is obtained by dividing the physical distance by the total number of pulses.

[0101] The control system continuously updates the cumulative travel distance L(t) at a fixed sampling period to ensure the accurate conveying position of the substrate at any given time. The sampling period is determined based on the equipment's operating speed and is typically no greater than 1 millisecond to guarantee real-time control.

[0102] S4.4: The control system retrieves the longitudinal tracking reference data obtained in step S1 from the storage unit. The longitudinal tracking reference data includes the row coordinates of the leading edge pixels of each width partition. and trailing edge pixel row coordinates and the encoder reference odometer value corresponding to that frame image. .

[0103] Among them, encoder reference mileage value This refers to the reference travel distance obtained by synchronous latching by the encoder at the start of exposure of the substrate image acquired by the front vision camera. The synchronous latching is implemented as follows: at the start of exposure of the front vision camera, the control system sends a latch trigger signal to the encoder counting unit. The encoder counting unit, in response to the latch trigger signal, immediately locks the current cumulative pulse value and converts the latched cumulative pulse value into the physical travel distance, which is used as the reference mileage value for that frame of image. Through a synchronous latching mechanism, a precise mileage correspondence is established between each frame of the front image and the substrate transport position.

[0104] The control system uses the leading edge pixel line coordinates of each width partition. Combined with vertical pixel equivalent Calculate the image domain distance in the transport direction relative to the reference image acquisition position within each width partition of the front coating pattern. The image domain distance is equal to... Similarly, based on the row coordinates of the trailing pixel... Calculate the image domain distance along the boundary of the front coating pattern within each width partition relative to the reference image acquisition position in the transport direction. The image domain distance is equal to... .

[0105] The image domain distance mentioned above represents the longitudinal physical distance of the boundary of the front-coated pattern relative to the starting position of the image acquisition area in the image coordinate system. It should be noted that... and All measurements are taken using a fixed reference row of the image (such as the first row at the top of the image) as the origin of the coordinate system. and All represent the longitudinal physical distance of the boundary relative to the reference line. The reference line corresponds to the starting position of the field of view of the frontal vision camera, and is predetermined and fixed during the equipment debugging phase; therefore, the entire width partition... and All measurements are taken in the same coordinate system and are comparable to each other.

[0106] The control system will fix the physical spacing Superimposed with the image domain distance, and then combined with the encoder's reference odometer value. Calculate the physical positions of the nozzles when they are open and closed for each width zone.

[0107] Specifically, for the i-th width partition, the physical position of the nozzle opening is:

[0108] Nozzle closing physical position:

[0109] It should be noted that when a specific point on the substrate moves from the field of view of the front vision camera to the nozzle position of the back coating head, the physical distance it needs to traverse is a fixed physical spacing. Therefore, when the leading edge of the front pattern moves from the field of view of the front vision camera to the nozzle position of the back coating head, the encoder's cumulative travel distance should reach the reference mileage value. Plus In addition, consider the image domain distance of the leading edge boundary relative to the reference row. Upon reaching this position, the coating nozzle of this section should open to align the starting boundary of the reverse coating with the leading edge boundary of the front pattern. Similarly, when the cumulative travel distance reaches... + + At this time, the coating nozzle of the zone should be closed so that the termination boundary of the reverse coating is aligned with the trailing edge boundary of the front pattern.

[0110] For example, let the row coordinates of the leading edge pixels of a certain partition be... =500 rows, vertical pixel equivalent =0.05mm / pixel, then the image domain distance of the leading edge boundary relative to the reference row is 500×0.05=25mm. If the reference mileage value... =1000mm, fixed physical spacing =800mm, then the physical position of the nozzle opening =1000+800+25=1825mm. When the cumulative travel distance of the substrate conveying encoder reaches 1825mm, the coating nozzle of this zone opens.

[0111] S4.5: The control system continuously monitors the cumulative travel distance L(t) of the substrate conveyor during the reverse coating process and correlates L(t) with the physical position of the nozzle opening in each width zone. and nozzle closing physical position Perform real-time comparisons.

[0112] When the cumulative travel distance L(t) reaches the nozzle opening physical position of a certain wide section When the control system sends an opening command to the coating nozzle corresponding to the width zone, the coating nozzle immediately opens in response to the opening command and begins to coat the reverse coating material onto the substrate surface.

[0113] When the cumulative travel distance L(t) reaches the nozzle closing physical position of a certain wide section When the control system sends a shut-off command to the coating nozzle corresponding to the width zone, the coating nozzle immediately shuts off in response to the shut-off command, stopping the application of the reverse coating material to the substrate surface.

[0114] In the above process, the coating nozzles of each width zone are controlled independently and do not affect each other. This is because the leading pixel row coordinates of each width zone... and trailing edge pixel row coordinates The difference may occur due to variations in the longitudinal boundary position of the front coating pattern in each section. For example, the leading edge of the front coating pattern may not be perfectly straight in the width direction, so the physical positions of the nozzle opening and closing may also differ slightly in each section. By controlling each section independently, the longitudinal boundary of the reverse coating in each width section is precisely aligned with the longitudinal boundary of the front coating pattern in each section.

[0115] S4.6: In this embodiment, the lateral correction execution system includes a main correction actuator and N fine-tuning actuators, where N is equal to the total number of width partitions.

[0116] The main alignment actuator is located on the substrate conveying path, upstream of the reverse coating head, and is used to apply a lateral displacement force to the entire substrate to compensate for any lateral deviation. The main alignment actuator can be an active alignment roller, a swing roller, or a lateral displacement platform, or any other actuator capable of driving the entire substrate to move in the width direction.

[0117] Each fine-tuning actuator is set up in a one-to-one correspondence with each width zone. Each fine-tuning actuator is located at the reverse coating head and corresponds to a local area of ​​the substrate within each width zone. It is used to apply an independent local lateral displacement force to the local area of ​​the substrate within the corresponding width zone to compensate for local deformation or pattern misalignment within that zone.

[0118] In a preferred embodiment, the fine-tuning actuator is a miniature piezoelectric guide wheel, which is disposed at the reverse coating head and arranged one-to-one with each width partition. Each miniature piezoelectric guide wheel contacts the corresponding width partition of the substrate before the substrate enters the reverse coating head. Upon receiving a control signal, it applies a reverse lateral correction force to the corresponding area of ​​the substrate according to the direction and magnitude of the fine-tuning amount of the corresponding partition. This causes a slight lateral displacement of the substrate in that area before it enters the coating head, thereby achieving lateral following of the reverse coating relative to the front pattern within that partition.

[0119] Miniature piezoelectric guide wheels are characterized by fast response speed and high displacement accuracy, enabling rapid compensation for high-frequency, small-amplitude lateral deviations.

[0120] S4.7: In step S3, the control system has determined the overall correction amount and the fine-tuning amount for each width zone. All the above quantities are in pixels. Before performing lateral correction, the control system converts the correction quantities in the pixel domain into displacement quantities in the physical domain.

[0121] Specifically, the overall correction amount Multiply by the horizontal pixel equivalent ρ_x to obtain the physical horizontal overall correction displacement. = The unit is millimeters (mm). Physical lateral overall correction displacement. This indicates the physical distance of the lateral displacement that the overall correction actuator needs to perform.

[0122] Fine-tuning the width of each section Multiply by the horizontal pixel equivalent respectively The physical lateral local fine-tuning displacement corresponding to each width partition is obtained. = The unit is millimeters (mm). Physical lateral local fine-tuning displacement. This represents the physical distance of the local lateral displacement that the fine-tuning actuator corresponding to the i-th width partition needs to perform.

[0123] In the above physical conversion, the sign of each correction value remains unchanged; a positive value represents a positive displacement in the width direction, and a negative value represents a negative displacement in the width direction. The control system pre-stores the displacement direction definitions of each actuator to ensure that the actuator's direction of motion is consistent with the sign of the correction value.

[0124] S4.8: The control system will adjust the overall lateral displacement. The command is sent to the drive unit of the overall alignment actuator. Based on the received displacement command, the drive unit of the overall alignment actuator drives the actuator to perform a lateral overall displacement, the magnitude of which is equal to... The absolute value, the direction is from The sign is determined to compensate for the overall lateral deviation of the substrate.

[0125] The control system simultaneously adjusts the physical lateral local fine-tuning displacement of each width zone. The commands are sent to the drive units of the fine-tuning actuators corresponding to each width partition. Each drive unit of the fine-tuning actuator independently drives its corresponding fine-tuning actuator to perform a local lateral displacement based on the displacement command it receives. The magnitude of the local lateral displacement is equal to the corresponding... The absolute value, the direction is determined by the corresponding The sign is determined.

[0126] Each fine-tuning actuator operates independently and is not coupled. The action of each actuator only affects a local area of ​​the substrate within its corresponding width zone and does not affect other width zones. Through this independent control, precise lateral correction of each width zone is achieved.

[0127] During the correction process, the main correction actuator and each fine-tuning actuator operate simultaneously to jointly complete the lateral correction. The main correction actuator eliminates the overall deviation of the substrate at the macro level, while each fine-tuning actuator eliminates the local deviation of each zone at the micro level. The two work together to align the lateral edges of the reverse coating within each width zone with the lateral edges of the front coating pattern.

[0128] Through the synergistic effect of the aforementioned longitudinal trigger control and lateral correction control, in this embodiment, in the substrate transport direction (longitudinal), the leading edge boundary of the reverse coating in each width section is aligned with the leading edge boundary of the front coating pattern in each width section, and the trailing edge boundary of the reverse coating in each width section is aligned with the trailing edge boundary of the front coating pattern in each width section. That is, the starting and ending positions of the reverse coating in the longitudinal direction are consistent with the front coating pattern. In the substrate width direction (lateral), the left edge of the reverse coating in each width section is aligned with the left edge of the front coating pattern in each width section, and the right edge of the reverse coating in each width section is aligned with the right edge of the front coating pattern in each width section. That is, the left and right edge positions of the reverse coating in the lateral direction are consistent with the edge positions of the front coating pattern.

[0129] Combining the above two dimensions, both longitudinal and transverse, the overall outline of the reverse coating on the substrate surface achieves high-precision overlap and alignment with the overall outline of the front coating pattern.

[0130] In summary, this embodiment achieves precise alignment between the longitudinal boundary of the reverse coating and the longitudinal boundary of the front coating pattern through the steps of triggering the start and end of the longitudinal coating of each width zone based on the longitudinal boundary, the preset fixed stroke from the camera to the coating head, and the encoder mileage, and driving the actuators based on the overall correction amount and the fine adjustment amount of each width zone. Through the encoder mileage synchronous triggering mechanism and fixed physical spacing compensation, the longitudinal boundary of the reverse coating is precisely aligned with the longitudinal boundary of the front coating pattern. Through the hierarchical collaborative control of the overall correction actuator and the fine adjustment actuators of each zone, the lateral edge of the reverse coating is precisely followed by the lateral edge of the front coating pattern throughout the entire width range.

[0131] On the other hand, the present invention also provides a machine vision-based double-sided coating alignment error compensation system, comprising: a front reference extraction module, used to acquire the longitudinal boundary and lateral edge data of the front coating pattern as a following reference for the reverse coating; a partition distortion calculation module, used to detect the lateral offset of the substrate within each width partition, and calculate the local distortion control amount of each width partition according to the position of the front pattern edge and the substrate edge; and an error diversion decision module, used to calculate the mean of the lateral offset of all width partitions, and compare the offset difference between the edge partition and the center partition: if it is less than a set value, the overall correction control amount is taken as the mean value and the fine-tuning control amount of each width partition is taken as the distortion control amount; otherwise, the overall correction control amount is still taken as the mean value, and the fine-tuning control amount of the edge partition is superimposed. The difference between the lateral offset and the mean of the corresponding width partition is added, and the fine-tuning control amount of the remaining width partitions is taken as the distortion control amount; the follow-up coating execution module is used for reverse coating. Based on the leading edge boundary position and trailing edge boundary position of each width partition in the longitudinal boundary, the preset fixed stroke from the camera to the coating head and the encoder mileage, the nozzle opening trigger position and nozzle closing trigger position of each width partition are determined respectively. When the encoder's real-time accumulated mileage reaches the nozzle opening trigger position, the coating nozzle of the corresponding width partition is opened, and when it reaches the nozzle closing trigger position, the coating nozzle of the corresponding width partition is closed; at the same time, based on the overall correction control amount and the fine-tuning control amount of each width partition, the overall correction actuator and the fine-tuning actuator corresponding to each width partition are controlled to perform lateral displacement respectively.

[0132] The system also includes one or more processors and memory.

[0133] The memory is used to store operable instructions that, when executed by the one or more processors, cause the one or more processors to perform operations, including the flow of a machine vision-based double-sided coating alignment error compensation method according to the foregoing embodiments, particularly... Figure 1 The flowchart of the method is shown.

[0134] Other aspects disclosed in the embodiments of the present invention also propose a computer-readable medium for storing software including instructions executable by one or more computers, which, upon execution, cause the one or more computers to perform operations including the flow of a machine vision-based double-sided coating alignment error compensation method of the foregoing embodiments, particularly... Figure 1 The flowchart of the method is shown.

[0135] It should be recognized that embodiments of the present invention may be implemented or carried out by computer hardware, a combination of hardware and software, or by computer instructions stored in a non-transitory computer-readable storage medium.

[0136] The method can be implemented using standard programming techniques, including a non-transitory computer-readable storage medium configured with a computer program in the computer program, wherein the storage medium is configured such that the computer operates in a specific and predefined manner.

[0137] Each program can be implemented in a high-level procedural or object-oriented programming language to communicate with the computer system; however, if required, the program can be implemented in assembly or machine language.

[0138] In any case, the language can be either compiled or interpreted.

[0139] Furthermore, for this purpose, the program can run on programmed application-specific integrated circuits.

[0140] The processes described herein (or variations and / or combinations thereof) can be executed under the control of one or more computer systems configured with executable instructions, and can be implemented by hardware or a combination thereof as code (e.g., executable instructions, one or more computer programs, or one or more applications) that commonly executes on one or more processors. The computer program includes a plurality of instructions executable by one or more processors.

[0141] Furthermore, the method can be implemented in any suitable type of computing platform, including but not limited to personal computers, minicomputers, mainframes, workstations, networked or distributed computing environments, standalone or integrated computer platforms, or in communication with charged particle tools or other imaging devices.

[0142] Various aspects of the present invention can be implemented in machine-readable code stored on a non-transitory storage medium or device, whether portable or integrated into a computing platform, such as a hard disk, optical read and / or write storage medium, RAM, ROM, etc., such that it can be read by a programmable computer, and when the storage medium or device is read by the computer, it can be used to configure and operate the computer to perform the processes described herein.

[0143] Furthermore, machine-readable code, or parts thereof, can be transmitted via wired or wireless networks.

[0144] When such media includes instructions or programs that combine with a microprocessor or other data processor to implement the steps described above, the invention described herein includes these and other different types of non-transitory computer-readable storage media.

[0145] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A machine vision-based method for compensating alignment errors in double-sided coating, characterized in that: include: Obtain the longitudinal boundary and lateral edge data of the front coating pattern as a reference for the reverse coating. The lateral offset of the substrate within each width zone is detected, and the local distortion control amount of each width zone is calculated based on the position of the front pattern edge and the substrate edge. Calculate the average value of the lateral offset of all width zones, and compare the offset difference between the edge zone and the center zone: if it is less than the set value, the overall correction control amount is taken as the average value and the fine adjustment control amount of each width zone is taken as the distortion control amount; otherwise, the overall correction control amount is still taken as the average value, the fine adjustment control amount of the edge zone is added to the difference between the lateral offset of the corresponding width zone and the average value, and the fine adjustment control amount of the remaining width zones is taken as the distortion control amount. The reverse coating process determines the nozzle opening trigger position and nozzle closing trigger position for each width partition based on the leading edge and trailing edge boundary positions of each width partition in the longitudinal boundary, the preset fixed travel distance from the camera to the coating head, and the encoder mileage. When the encoder's real-time accumulated mileage reaches the nozzle opening trigger position, the coating nozzle of the corresponding width partition is opened; when it reaches the nozzle closing trigger position, the coating nozzle of the corresponding width partition is closed. Simultaneously, based on the overall correction control amount and the fine-tuning control amount of each width partition, the overall correction actuator and the fine-tuning actuator corresponding to each width partition are controlled to perform lateral displacement.

2. The machine vision-based double-sided coating alignment error compensation method as described in claim 1, characterized in that: The acquisition of the longitudinal boundary and lateral edge data of the front coating pattern includes: Images of the substrate with the front coating completed are captured using a front-view camera; The substrate image is divided into multiple width partitions along the width direction of the substrate; Extract the leading edge and trailing edge positions of the front coating pattern in the substrate conveying direction within each wide partition as longitudinal tracking reference data; Extract the left and right edge positions of the front coating pattern in each width partition along the width direction of the substrate, and use them as lateral tracking reference data.

3. The machine vision-based double-sided coating alignment error compensation method as described in claim 1, characterized in that: The detection of the lateral offset of the substrate within each width zone includes: Images of the physical edges of the substrate at each wide section are captured in front of the reverse coating head using a reverse vision camera. Detect the current pixel position of the physical edge of the substrate in the image; The current pixel position is compared with the theoretical pixel position under the pre-calibrated substrate without offset state to obtain the overall lateral offset of the substrate within each width partition.

4. The machine vision-based double-sided coating alignment error compensation method as described in claim 1, characterized in that: The calculation of local distortion control for each width zone based on the position of the front pattern edge and the substrate edge includes: Pre-determine the fixed design margin between the edge of the front coating pattern and the physical edge of the substrate within each width zone; Based on the current actual position of the physical edge of the substrate within each width partition and the fixed design margin, determine the theoretical expected position of the target edge for reverse coating at the current substrate position. By comparing the actual edge position of the front coating pattern in each width partition with the theoretically expected position, the local distortion control amount introduced by the substrate width change or the previous coating error in each width partition is obtained.

5. The machine vision-based double-sided coating alignment error compensation method as described in claim 1, characterized in that: Based on the comparison results, the overall correction control amount and the fine-tuning control amounts for each width zone are determined as follows: Calculate the arithmetic mean of the overall lateral offset of the substrate within all width zones, and use the arithmetic mean as the overall correction control value; Determine the left edge partition, right edge partition, and center partition along the width direction of the substrate; Calculate the first difference in the lateral offset between the left edge partition and the center partition, and the second difference in the lateral offset between the right edge partition and the center partition, respectively. When both the first difference and the second difference are less than the set value, the fine-tuning control amount of each width partition is set to the local distortion control amount corresponding to each width partition. When either the first difference or the second difference is greater than or equal to the set value, the fine-tuning control amount of the left edge partition is set to the difference between the horizontal offset of the left edge partition and the arithmetic mean, plus the local distortion control amount corresponding to the left edge partition; the fine-tuning control amount of the right edge partition is set to the difference between the horizontal offset of the right edge partition and the arithmetic mean, plus the local distortion control amount corresponding to the right edge partition; and the fine-tuning control amount of all intermediate width partitions is set to the local distortion control amount corresponding to each intermediate width partition.

6. The machine vision-based double-sided coating alignment error compensation method as described in claim 5, characterized in that: The set value is a judgment threshold obtained by pre-calibrating through static repeatability testing. The static repeatability testing involves collecting multiple sets of sample data on the absolute value of the difference in lateral offset between the central partition and the edge partition within a preset period of continuous and stable operation of the device, and determining the judgment threshold based on the statistical distribution characteristics of the multiple sets of sample data.

7. The machine vision-based double-sided coating alignment error compensation method as described in claim 6, characterized in that: The process of determining the nozzle opening trigger position and nozzle closing trigger position for each width zone includes: Predetermine the fixed physical distance along the substrate conveying direction between the image acquisition position of the front vision camera and the coating execution position of the back coating head; Real-time acquisition of the cumulative travel distance of the substrate conveyor; Based on the positions of the leading edge and trailing edge boundaries within each width partition, the image domain distance corresponding to the longitudinal boundary within each width partition is determined. The fixed physical spacing is superimposed with the image domain distance, and combined with the reference travel distance corresponding to the moment when the front vision camera acquires the substrate image, the nozzle opening trigger position and nozzle closing trigger position of the reverse coating in each width partition are determined respectively. When the cumulative travel distance reaches the nozzle opening trigger position, the coating nozzle of the corresponding width zone is opened; when the cumulative travel distance reaches the nozzle closing trigger position, the coating nozzle of the corresponding width zone is closed.

8. The machine vision-based double-sided coating alignment error compensation method as described in claim 7, characterized in that: The reference travel distance is obtained by synchronous latching of the encoder; The synchronous latching refers to synchronously reading and locking the current cumulative pulse value of the transport encoder at the start of exposure of the substrate image acquired by the front vision camera, and converting the latched current cumulative pulse value into a physical travel distance as the reference travel distance.

9. The machine vision-based double-sided coating alignment error compensation method as described in claim 8, characterized in that: The step of controlling the main correction actuator and the fine-tuning actuators corresponding to each partition to perform lateral displacement includes: The overall correction control quantity is converted into a physical lateral overall displacement quantity, and the overall correction actuator is controlled to perform the lateral overall displacement to compensate for the overall lateral deviation of the substrate. The fine-tuning control quantities of each width zone are converted into physical lateral local displacement quantities corresponding to each width zone, and the fine-tuning actuators corresponding to each width zone are controlled to perform independent local lateral displacements. The fine-tuning actuators are set in a one-to-one correspondence with the width zones. By combining the overall displacement of the main correction actuator and the local displacement of the fine-tuning actuators in each width zone, the longitudinal boundary of the reverse coating in each width zone is aligned with the longitudinal boundary of the front coating pattern, while the lateral edge of the reverse coating in each width zone follows the lateral edge position of the front coating pattern.

10. A machine vision-based double-sided coating alignment error compensation system, based on the machine vision-based double-sided coating alignment error compensation method according to any one of claims 1 to 9, characterized in that: Also includes: The front reference extraction module is used to obtain the longitudinal boundary and lateral edge data of the front coating pattern as a reference for the reverse coating. The partition distortion calculation module is used to detect the lateral offset of the substrate within each width partition and calculate the local distortion control amount of each width partition based on the position of the front pattern edge and the substrate edge. The error distribution decision module is used to calculate the mean of the lateral offset of all width partitions and compare the offset difference between the edge partition and the center partition: if it is less than the set value, the overall correction control amount is taken as the mean and the fine-tuning control amount of each width partition is taken as the distortion control amount; otherwise, the overall correction control amount is still taken as the mean, the fine-tuning control amount of the edge partition is added to the difference between the lateral offset of the corresponding width partition and the mean, and the fine-tuning control amount of the remaining width partitions is taken as the distortion control amount. The follow-up coating execution module is used for reverse coating. Based on the leading edge and trailing edge positions of each width partition in the longitudinal boundary, the preset fixed stroke from the camera to the coating head, and the encoder mileage, it determines the nozzle opening trigger position and nozzle closing trigger position of each width partition. When the encoder's real-time accumulated mileage reaches the nozzle opening trigger position, the coating nozzle of the corresponding width partition is opened; when it reaches the nozzle closing trigger position, the coating nozzle of the corresponding width partition is closed. At the same time, based on the overall correction control amount and the fine-tuning control amount of each width partition, it controls the overall correction actuator and the fine-tuning actuator corresponding to each width partition to perform lateral displacement.

Images