Parameter adjustment method in PCB punching process, computing device, and storage medium

Abstract

The application discloses a parameter adjustment method in a PCB punching process, a computing device and a storage medium. The method comprises the following steps: acquiring double-sided offset data output in a PCB punching alignment process; determining process data of the current punching operation according to the double-sided offset data; storing the process data and sending the process data to a backend detection device; receiving actual detection data fed back by the detection device, and adjusting screening parameters of domain reduction iteration operation and / or compensation coefficients of differential compensation according to a comparison result of the actual detection data and the process data. Thus, data barriers between a punching process and a detection process are broken, and the effect of adaptive iterative optimization of process parameters is realized.

Description

Technical Field

[0001] This application relates to the field of printed circuit board (PCB) processing technology, and more specifically, to a parameter adjustment method, computing device and computer storage medium in PCB punching process. Background Technology

[0002] In the inner layer core board processing of printed circuit boards (PCBs), punching machines are used to process multilayer board lamination positioning holes and rivet holes on the inner layer core board. The punching accuracy directly affects the alignment accuracy of the conductor patterns between the subsequent multilayer boards. Currently, the mainstream punching alignment solution in the industry is based on a CCD vision positioning architecture. That is, a CCD camera captures the alignment target on the PCB board, and after image processing and coordinate calculation, the motion platform is driven to complete the correction compensation before performing the punching operation.

[0003] Current PCB punching alignment technology generally adopts a single-sided imaging architecture, which involves placing two or four CCD cameras on one side (usually the top) of the PCB board to acquire target images. The equipment software only acquires the target coordinate data on the upper surface of the board and uses algorithms such as least squares fitting to solve for the board's positional offset (translation and rotation angle). The platform then performs correction compensation before punching. However, because this type of solution cannot acquire target data on the reverse side of the board, it cannot detect the layer offset of the front and back patterns, and it is difficult to adapt to the different deformations of the front and back sides of the board for separate compensation.

[0004] More importantly, the process parameters of existing punching alignment systems, including fitting parameters, convergence conditions, and compensation coefficients in the positioning algorithm, are entirely determined by manual experience. The debugging process is not only time-consuming and labor-intensive, but also highly dependent on the skill level of the debugging personnel. Furthermore, the parameters debugged by different personnel are inconsistent, making it difficult to respond quickly when the characteristics of incoming material batches or the status of equipment fluctuate.

[0005] Therefore, how to achieve adaptive optimization of process parameters has become an urgent technical problem to be solved. Summary of the Invention

[0006] The main purpose of this application is to provide a parameter adjustment method, computing device and storage medium in PCB punching process, so as to solve the technical problem that the process parameters of the existing punching alignment system rely entirely on manual experience for debugging and cannot be adaptively optimized based on actual processing results, so as to break down the data barrier between the punching process and the inspection process and achieve the effect of adaptive iterative optimization of process parameters.

[0007] To achieve the above objectives, the first aspect of this application proposes a parameter adjustment method for PCB punching processes, comprising the following steps: acquiring double-sided offset data output during the PCB punching alignment process, wherein the double-sided offset data is the front and back offset data corresponding to the optimal alignment posture selected through step-by-step shrinking iteration calculation based on the target coordinate data at the four corner positions of the front and back sides of the PCB board; determining the process data for this punching operation based on the double-sided offset data, wherein the process data is the process data corresponding to the punching operation performed based on the differential compensation of each punching position on the front and back sides of the PCB board according to the double-sided offset data and the compensation positioning; storing the process data and sending it to the back-end inspection equipment; receiving the actual inspection data fed back by the inspection equipment, and adjusting the screening parameters of the shrinking iteration calculation and / or the compensation coefficient of the differential compensation according to the comparison results between the actual inspection data and the process data.

[0008] The above scheme involves iterative calculations with progressively narrowing domains, including: obtaining the four corner coordinates of the front and back sides of a pre-set standard PCB board as theoretical reference points; using the straight-line distance between the actual coordinates of each target and the corresponding theoretical reference point under the current posture as the target offset distance, and constructing a circular screening domain with the theoretical reference point as the screening center and the target offset distance as the radius; taking the four corner points of the inscribed square parallel to the axis of the circular screening domain as candidate points, calculating the cumulative total of the four-point offsets on each candidate point, where the cumulative total of the four-point offsets on each side is the sum of the offset distances of the four targets on the same board surface; and using the minimum cumulative total offset value as the minimum value of the cumulative offset. The corresponding candidate point is taken as the optimal target point in the current screening domain. If the distance between the theoretical reference point and the optimal target point is less than the preset accuracy threshold, the posture corresponding to the current optimal target point is taken as the optimal alignment posture. Otherwise, the length of the line connecting the theoretical reference point and the current optimal target point is taken as the diameter of the new circular screening domain, and the midpoint of the line is taken as the new screening center. The above steps of selecting candidate points and determining optimal target points are repeated until the accuracy requirements are met or the mechanical convergence limit is reached. The mechanical convergence limit is: the distance between the optimal target points obtained by two adjacent iterations is less than the minimum executable step distance of the three-axis servo attitude adjustment platform.

[0009] In the above scheme, the cumulative value of the four-point offset on a single surface is calculated according to the following formula:

[0010] in, L i The target offset distance at the lower corner position in the current attitude. L sum This is the cumulative total of the four-point offset on a single surface. x i , y i ) is the first i The actual coordinates of each target, ( x0i , y 0i ) is the first i The coordinates of a theoretical reference point.

[0011] In the above scheme, the differentiated compensation includes: determining the translation and rotation amounts of the entire board on the front and back sides based on the double-sided offset data, and generating the full board compensation amount for the front side and the full board compensation amount for the back side; determining the precise alignment compensation amount for the front side and the precise alignment compensation amount for the back side corresponding to the product specification information of the current PCB board; superimposing the full board compensation amount for the front side and the precise alignment compensation amount for the front side to generate the final compensation amount for the front side; and superimposing the full board compensation amount for the back side and the precise alignment compensation amount for the back side to generate the final compensation amount for the back side.

[0012] In the above scheme, the double-sided offset data includes the coordinates of each target on the upper plate surface ( x u,i , yu,i ) and the coordinates of each target on the lower plate ( x d,i , y d,i Before the differential compensation, the method further includes: calculating the layer offset of the homologous target based on the double-sided offset data. Δx bias,i , Δy bias,i ): .

[0013] If the layer deviation exceeds the preset acceptable tolerance, the PCB board is determined to be defective and a diversion and sorting process is performed.

[0014] In the above scheme, obtaining the double-sided offset data output during the PCB punching alignment process includes: simultaneously acquiring full-area images of the front and back sides using four sets of CCD cameras positioned above and below the PCB board conveying path; sequentially performing grayscale conversion, Gaussian filtering for noise reduction, and edge extraction on the acquired double-sided full-area images, and eliminating invalid noise contours based on contour area and roundness to output Mark contour images; identifying the pixel coordinates of each Mark target in its corresponding camera pixel coordinate system based on the Mark contour images; and transforming each pixel coordinate to a unified device world coordinate system using a multi-camera unified world coordinate calibration method to obtain target coordinate data. The multi-camera unified world coordinate calibration method includes: manipulating eight CCD cameras (four sets above and four below) to photograph a standard checkerboard calibration board, completing lens distortion correction and multi-camera spatial pose calculation, and generating and storing the transformation matrix from each camera pixel coordinate system to the device unified world coordinate system.

[0015] In the above scheme, the sieving parameters and / or compensation coefficients of the narrowing iteration calculation are adjusted according to the comparison results of actual test data and process data. This includes: statistically removing outliers from the test data and taking the test data that does not exceed the preset confidence area as valid feedback data; calculating the process deviation trend of the current batch based on the valid feedback data, and gradually adjusting the sieving parameters and / or compensation coefficients according to the preset upper limit of parameter change rate.

[0016] The above scheme further includes: calculating the mean and standard deviation of the detection data; for each detection data, if the difference between the detection data and the mean is less than or equal to three times the standard deviation, then the detection data is determined to fall into the preset confidence region.

[0017] To achieve the above objectives, a second aspect of this application proposes a computing device, including a processor and a memory storing a computer program, wherein when the processor runs the computer program, the steps of the parameter adjustment method in any of the above-described PCB punching processes are implemented.

[0018] To achieve the above objectives, a third aspect of this application proposes a computer storage medium storing a computer program, which, when executed by a processor, implements the steps of the parameter adjustment method in any of the above-described PCB punching processes.

[0019] The technical solutions provided by the embodiments of this application may include the following beneficial effects: In this application, by acquiring double-sided offset data and corresponding process data generated based on double-sided target data during the punching alignment process, storing and sending them to the back-end testing equipment, and then using the comparison results of the actual testing data fed back by the testing equipment with the process data to drive the adaptive adjustment of screening parameters and / or compensation coefficients, a complete closed loop is formed from punching data acquisition to process parameter determination, then to testing feedback acquisition, and finally to parameter iterative update. This solves the technical problem in the prior art that process parameters rely entirely on manual experience for debugging and cannot be continuously optimized based on actual processing results. It achieves the technical effect of breaking down the data barrier between the punching process and the back-end testing process and realizing adaptive iterative optimization of process parameters. Attached Figure Description

[0020] The accompanying drawings, which form part of this application, are used to provide a further understanding of the application and to make other features, objects, and advantages of the application more apparent. The illustrative embodiments and descriptions of this application are used to explain the application and do not constitute an undue limitation of the application. In the drawings: Figure 1 A flowchart illustrating the parameter adjustment method in the PCB punching process provided in this application embodiment; Figure 2A schematic diagram of the method for step-by-step shrinkage iteration calculation in the PCB punching process provided in this application embodiment; Figure 3 This is a schematic diagram of the method for differential compensation in PCB punching process provided in the embodiments of this application; Figure 4 A schematic diagram of the structure of a computing device provided in an embodiment of this application. Detailed Implementation

[0021] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0022] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0023] In this application, the terms "upper," "lower," "left," "right," "front," "rear," "top," "bottom," "inner," "outer," "middle," "vertical," "horizontal," "lateral," and "longitudinal" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are primarily for the purpose of better describing this application and its embodiments, and are not intended to limit the indicated device, element, or component to having a specific orientation, or to be constructed and operated in a specific orientation.

[0024] Furthermore, in addition to indicating location or positional relationship, some of the aforementioned terms may also have other meanings. For example, the term "above" may also be used in some cases to indicate a certain dependency or connection relationship. Those skilled in the art can understand the specific meaning of these terms in this application based on the specific circumstances.

[0025] Furthermore, the terms "installation," "setup," "equipped with," "connection," "linked," and "socketing" should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral structure; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, or an internal connection between two devices, components, or parts. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0026] It should be noted that the embodiments of this application can be applied to the field of adaptive optimization of process parameters in the punching process of PCB inner layer core boards. For example, by deploying a double-sided eight-CCD imaging module, a three-axis servo attitude adjustment platform, and an industrial control host on the PCB punching production line, target coordinate data at the four corners of each side of the PCB board are acquired. The optimal alignment posture is then selected through iterative calculations with progressive domain shrinkage, and the corresponding double-sided offset data is output. Based on this double-sided offset data, the industrial control host performs differential compensation on each punching position on the front and back sides of the PCB board and executes the punching operation. The process data of this punching operation is stored and sent to the back-end X-Ray inspection equipment. After completing the layer offset and expansion / contraction detection, the X-Ray inspection equipment feeds back the actual detection data to the industrial control host in the punching process. The industrial control host compares the actual detection data with the process data and adaptively adjusts the screening parameters of the progressive domain shrinkage iterative calculation and / or the compensation coefficients of the differential compensation based on the comparison results. This aims to replace the limitations of traditional processes where process parameters rely entirely on manual experience for adjustment and cannot be continuously optimized based on actual processing results.

[0027] Understandably, existing solutions for optimizing the parameters of the PCB inner layer core board punching process face the following main technical challenges. First, the process parameters of the punching alignment system, including fitting parameters, convergence conditions, and compensation coefficients in the positioning algorithm, rely entirely on manual experience for debugging. This debugging process is not only time-consuming and labor-intensive, but also highly dependent on the skill level of the debugging personnel, and the parameters determined by different personnel are inconsistent. Second, the data between the punching process and the downstream X-Ray layer misalignment detection equipment are isolated, lacking data exchange and closed-loop linkage mechanisms. After punching, layer misalignment and expansion / contraction results can only be detected by X-Ray equipment after the subsequent lamination process. These detection results cannot be fed back to the punching process to optimize alignment parameters. Furthermore, when the characteristics of incoming material batches fluctuate or the equipment status changes, the punching accuracy cannot be adaptively restored because the actual processing results cannot be used for continuous calibration of the process parameters. Manual readjustment is necessary, severely restricting the improvement of production line quality consistency and process intelligence.

[0028] Based on this, the embodiments of this application aim to provide a parameter adjustment method in PCB punching process. By storing the process data during the punching alignment process and sending it to the back-end inspection equipment, the actual inspection data fed back by the inspection equipment is compared with the process data, and the screening parameters and / or compensation coefficients in the punching alignment process are adaptively adjusted accordingly. This achieves closed-loop iterative optimization of punching process parameters, breaks down the data barrier between the punching process and the inspection process, reduces the reliance on manual debugging, and improves the consistency and intelligence level of mass production line processes.

[0029] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0030] Figure 1 This is a flowchart illustrating the parameter adjustment method in the PCB punching process provided in this application embodiment, as shown below. Figure 1 As shown, applied to a computing device, the method includes: Step S1: Obtain the double-sided offset data output during the PCB punching alignment process. The double-sided offset data is the front and back offset data corresponding to the optimal alignment posture selected by step-by-step shrinking iterative calculation based on the target coordinate data at the four corners of the front and back of the PCB board.

[0031] It should be noted that this method is executed by a computing device, which generally refers to an industrial control computer (ICC) with data acquisition, image processing, coordinate calculation, and motion control capabilities. An ICC is a dedicated computer that uses a bus architecture to detect and control production processes and electromechanical equipment. It is typically deployed in the control cabinet of a PCB punching machine and communicates with a CCD camera, a three-axis servo attitude adjustment platform, and back-end detection equipment via Ethernet or a dedicated data interface. The ICC has a built-in image acquisition card, motion control card, and the parameter adjustment program described in this application embodiment, used to perform steps such as target coordinate extraction, step-by-step iterative calculation, process data storage and transmission, detection data reception and comparison, and adaptive adjustment of screening parameters and compensation coefficients. It is understood that the computing device is not limited to an ICC; it can also be an embedded controller, edge computing node, or other types of industrial computer with corresponding computing and communication capabilities.

[0032] A PCB (Printed Circuit Board) serves as the support and electrical connection carrier for electronic components. The inner core of a PCB refers to the substrate inside a multilayer PCB used to form the inner layer circuitry. Its surface is covered with copper foil, which is etched to form the inner conductive patterns. In the manufacturing process of multilayer PCBs, the core is usually used as the starting layer. Prepreg (PP) sheets and outer copper foil are alternately stacked on both sides of the core and then laminated to form a multilayer board structure. A multilayer PCB is made by laminating multiple inner cores, and the conductor patterns on each core layer need to maintain a precise alignment. To achieve interlayer alignment, the inner cores need to be punched before lamination, and these punched holes serve as positioning references for subsequent lamination processes. Therefore, precise positioning of the inner cores during the punching process is a prerequisite for ensuring the interlayer alignment accuracy of the multilayer board.

[0033] Target coordinate data refers to the position coordinates of preset Mark points (also known as optical positioning points or reference points) on the PCB board surface in the unified world coordinate system of the device. Mark points are specific marking graphics designed on the surface of the printed circuit board, typically in geometric shapes such as solid circles, crosses, squares, or irregular shapes. Their function is to provide a unified coordinate reference and positioning reference for automated vision equipment. In the PCB inner core board punching process, Mark points are usually placed at the four corners of the board. The CCD camera acquires images of the Mark points, and after image processing and coordinate transformation, obtains the physical coordinates (usually in millimeters) of each Mark point in the device's world coordinate system. These coordinates are the target coordinate data. It is understood that the specific values ​​of the target coordinate data vary depending on the placement posture, deformation state of the board, and the origin of the device's coordinate system. The target coordinate data described in this embodiment includes both the currently acquired real-time coordinates and the pre-stored theoretical reference coordinates corresponding to a standard PCB board. The theoretical reference coordinates are used as a reference target for alignment operations, and the deviation between the actual coordinates and the theoretical reference coordinates reflects the offset and deformation state of the board.

[0034] like Figure 2 As shown: In some implementations, the step-by-step domain shrinking iteration operation includes: Step S21: Obtain the preset reference coordinates of the four corners of the front and back of the standard PCB board as theoretical reference points; Step S22: Using the straight-line distance between the actual coordinates of each target and the corresponding theoretical reference point under the current posture as the target offset distance, construct a circular screening area with the theoretical reference point as the screening center and the target offset distance as the screening radius; Step S23: Take the four corner points of the inscribed square parallel to the axis of the circular screening area as candidate points, and calculate the cumulative value of the single-sided four-point offset corresponding to each candidate point. The cumulative value of the single-sided four-point offset is the sum of the offset distances of the four targets on the same plate. Step S24: Select the candidate point corresponding to the minimum cumulative offset value as the optimal target point in the current screening domain; Step S251: If the distance between the theoretical reference point and the optimal target point is less than the preset accuracy threshold, then the pose corresponding to the current optimal target point is taken as the optimal alignment pose. Step S252: Otherwise, take the length of the line connecting the theoretical benchmark point and the current optimal target point as the diameter of the new circular screening area and the center of the line as the new screening center, and repeat the above steps of candidate point selection and optimal target point determination until the accuracy requirements are met or the mechanical convergence limit is reached. The mechanical convergence limit is: the distance between the optimal target points obtained by two adjacent iterations is less than the minimum executable step distance of the three-axis servo attitude adjustment platform.

[0035] The theoretical reference point is acquired by a CCD camera and then sent to the industrial control equipment so that the equipment can perform subsequent processing based on the theoretical reference point.

[0036] It is understandable that the preset standard PCB board front and back corner reference coordinates can refer to a pre-selected standard PCB sample without deformation or layer offset, with Mark points (optical positioning points) set at the four corners of its front and back. Eight CCD cameras (four sets each at the top and bottom, for a total of eight channels) simultaneously acquire Mark images of the front and back of this standard sample. After image processing and coordinate transformation, the coordinate values ​​of each Mark point in the device's unified world coordinate system are obtained. These reference coordinates serve as the reference target for subsequent alignment operations and are denoted as (…). x 0i , y 0i ), i =1, 2, 3, 4, which correspond to the theoretical reference points at the four corners of the board. The purpose of the theoretical reference points is to provide a unified reference benchmark for evaluating the actual offset and deformation of the board. That is, the deviation between the actual target coordinates and the theoretical target coordinates reflects the offset and deformation state of the board.

[0037] Among them, the actual coordinates of the target refer to the real-time coordinate values ​​of each Mark point in the unified world coordinate system of the equipment, which are acquired and identified by the CCD camera after the PCB board to be processed has been pre-aligned and coarsely adjusted and placed on the working platform of the punching equipment. This is denoted as ( x i , y i ), i =1, 2, 3, 4.

[0038] The current posture refers to the actual spatial pose of the PCB board to be processed on the equipment's working platform, including the board's position in the X direction, its position in the Y direction, and its rotation along the XY plane. Since there is still a certain positioning deviation after the board is loaded and pre-aligned, and the board itself may be deformed due to previous processes such as etching and lamination, there will inevitably be a deviation between the actual coordinates of the target under the "current posture" and the theoretical reference coordinates.

[0039] Target offset distance refers to the straight-line distance between the actual coordinates of the target at the same angular position and the corresponding theoretical reference point, calculated according to the following formula:

[0040] This distance quantifies the offset of the Mark points at each corner position under the current attitude.

[0041] Among them, the circular sieve area refers to the area based on the theoretical reference point. O 0 is the target offset distance at a certain corner position with the current posture as the center. L i A circular region is constructed with radius . This circular region serves as the search space for the initial screening, limiting the possible attitude compensation range around the theoretical reference point to a circular boundary. Subsequent candidate points are all selected from this circular region.

[0042] In this context, candidate points refer to the four corner points of the axis-parallel inscribed square within the aforementioned circular screening domain as candidate pose points to be evaluated. An "axis-parallel inscribed square" means that the sides of this square are parallel to the X and Y axes of the equipment's world coordinate system, and all four vertices lie on the circumference of the circular screening domain. Each candidate point represents a possible alignment compensation pose, that is, the virtual pose of the plate material when this point is used as the new alignment reference.

[0043] In some implementations, the cumulative value of the four-point offset on a single surface, i.e., the sum of the offset distances, refers to the sum of the offset distances of the actual coordinates of the target at the four corner positions on the same plate surface (front or back) relative to their respective theoretical reference points, and is calculated according to the following formula:

[0044] The cumulative total reflects the degree of overall offset of the four corners of the entire plate from the theoretical baseline under the current candidate posture. The smaller the value, the better the overall alignment performance of the current posture.

[0045] The minimum cumulative offset value refers to the minimum value among the four candidate points in the current domain, calculated one by one, after calculating the cumulative offset of the four points on one side.

[0046] The optimal target point refers to the candidate point corresponding to the minimum cumulative deviation value, denoted as . P 1 (first iteration) or P n (Nth iteration). This point represents the optimal compensation posture that minimizes the overall offset of the four corners of the plate within the current screening domain.

[0047] The preset accuracy threshold refers to the maximum allowable positioning deviation value that is pre-entered by process engineers through a human-machine interface based on the customer's order requirements for PCB punching positioning accuracy. The specific value of this threshold can be customized according to different board material rules and processing requirements; for example, it can be set to 0.3mil (approximately 7.62μm) or other values ​​that meet the process requirements.

[0048] The optimal alignment posture refers to the situation where the distance between the current optimal target point and the theoretical reference point is less than a preset accuracy threshold, and the actual Mark point offset difference at each corner position does not exceed a preset single-point accuracy threshold. In this case, the posture corresponding to the current optimal target point is used as the final alignment compensation reference. At this point, the posture of the sheet metal meets the processing accuracy requirements, and the subsequent differential compensation and punching execution steps can proceed.

[0049] The three-axis servo attitude adjustment platform is an actuator used in PCB punching equipment to perform fine-tuning of the board's attitude. For example, the three-axis servo attitude adjustment platform can adopt a parallel linkage structure of U, V, and W axes. The U-axis is driven independently to achieve translation adjustment in the X-direction of the equipment coordinate system; the V-axis and W-axis are fed synchronously to achieve translation in the Y-direction; and the V-axis and W-axis move in opposite directions at differential speeds to achieve angular rotation adjustment of the platform in the XY plane. This platform can receive attitude adjustment commands from the industrial control host and complete micron-level displacement and rotation movements.

[0050] The minimum executable step distance of a three-axis servo attitude adjustment platform refers to the minimum displacement or rotation that the platform can achieve with a single adjustment command. Due to the physical precision limits of the servo motor drive and mechanical transmission mechanism, the platform cannot perform attitude adjustments smaller than this step distance. This minimum step distance is an inherent property determined by physical factors such as the pulse equivalent of the servo motor, the lead screw, and the clearance of the transmission mechanism.

[0051] The mechanical convergence limit refers to the point where the distance between the optimal target points obtained from two consecutive iterations is less than the minimum executable step size of the three-axis servo attitude adjustment platform. This means that even if the iteration domain is further narrowed, the theoretically calculated better attitude has exceeded the actual physical execution capability of the platform and cannot be achieved through platform actions. At this point, continuing the iteration is meaningless, hence it is used as one of the iteration termination conditions.

[0052] It is understandable that in step S252, if the first round of screening results does not reach the preset accuracy threshold, the candidate point selection and optimal target point determination steps are re-executed with the length of the line connecting the theoretical benchmark point and the current optimal target point as the diameter of the new circular screening area and the midpoint of the line as the new screening center, until the accuracy requirements are met or the mechanical convergence limit is reached.

[0053] Thus, in this embodiment, the optimal compensation posture is approximated within a limited number of iterations through a hierarchical screening convergence mechanism. Unlike traditional single-fit algorithms that solve the compensation coefficients all at once, this embodiment employs a hierarchical convergence strategy that progresses from large-scale coarse screening to small-scale fine screening and then to progressively narrowing the search range. In each iteration, the search range is further reduced within the neighborhood of the optimal target point from the previous iteration, causing the optimal target point to gradually converge toward the theoretical benchmark point. This hierarchical screening mode avoids invalid posture calculations and shortens the convergence time. Furthermore, this embodiment uses the "minimum executable step size of the three-axis servo posture adjustment platform" as one of the iteration termination conditions. The iteration terminates when the distance between the optimal target points of two adjacent iterations is less than the platform's minimum step size, reducing the problem of theoretically calculated compensation exceeding the platform's physical execution range and ensuring that the optimal posture output by the algorithm is truly executable. Furthermore, this implementation method has extremely low overall computational load and fast convergence speed, which can adapt to the needs of high-end PCB mass production. Finally, this implementation method uses the total value of the four-point offset on a single side as a comprehensive evaluation index of posture advantage, rather than relying on single-point offset for judgment. This can effectively reduce the risk of getting stuck in local optima due to local deformation or single-point recognition anomalies, and adapt to non-uniform deformation of the board material caused by uneven etching, copper thickness differences, etc.

[0054] Step S2: Based on the double-sided offset data, determine the process data for this punching operation. The process data is based on the double-sided offset data, which is used to differentiate the punching positions on the front and back of the PCB board, and the corresponding process data for the punching operation is executed based on the compensated positioning.

[0055] In this context, process data refers to the complete set of recordable and traceable parameters involved in the entire punching operation. For example, process data may include: target coordinate data, offset data, screening parameters, compensation coefficients, and quality judgment results, etc.

[0056] The screening parameters include the screening radius, preset accuracy convergence threshold, and mechanical convergence limit in the step-by-step narrowing iteration calculation; the compensation coefficients include the full-board compensation amount on the front and back sides generated during the double-sided differential compensation process, as well as the precise alignment compensation amount on the front and back sides corresponding to the product specification information; the full-board compensation amount and precise alignment compensation amount on the same board are superimposed to form the final compensation amount for that board. The quality judgment results can include pre-defect screening results, layer deviation verification results, and other quality judgment information.

[0057] Please see Figure 3 ,like Figure 3 As shown, in some implementations, differential compensation includes: Step S31: Based on the double-sided offset data, determine the translation and rotation amounts of the entire plate on the front and back sides respectively, and generate the compensation amounts of the entire plate on the front and back sides. Step S32: Based on the current PCB board product specification information, determine the front and back precision alignment compensation amounts corresponding to the product specification information. Step S33: Add the front whole board compensation amount and the front fine alignment compensation amount to generate the front final compensation amount; add the back whole board compensation amount and the back fine alignment compensation amount to generate the back final compensation amount.

[0058] For example, based on the double-sided offset data, the front whole-board compensation amount and the back whole-board compensation amount can be generated separately. The front whole-board compensation amount refers to the translation and rotation compensation applied to the entire front of the PCB board, which includes three components: ΔX_U, ΔY_U, and θ_U, and is used to correct the offset and deflection of the entire front board relative to the theoretical reference position. The back whole-board compensation amount refers to the translation and rotation compensation applied to the entire back of the PCB board, which includes three components: ΔX_D, ΔY_D, and θ_D, and is used to correct the offset and deflection of the entire back board relative to the theoretical reference position.

[0059] It should be noted that the front and back sides of the PCB board undergo independent etching and exposure processes during manufacturing. Due to factors such as uneven distribution of etching solution, exposure alignment deviation, and differences in stress release during pressing, the deformation amplitude and direction of the front and back sides are often inconsistent. Therefore, the compensation amount of the front board and the compensation amount of the back board are calculated independently to adapt to the different deformation states of the front and back sides.

[0060] Product specification information refers to a set of parameters characterizing the characteristics of the current PCB batch, which may include at least one of the following: number of PCB layers, board thickness, core material type, copper thickness, design expansion / contraction coefficient, and customer alignment accuracy level. Different PCB specifications exhibit differences in fixture positioning characteristics, tool wear compensation requirements, and platform dynamic response characteristics during the punching process. These differences introduce additional systematic positioning deviations beyond the overall board compensation amount, requiring correction through "precision alignment compensation."

[0061] The precise alignment compensation amount is a correction amount used to compensate for systematic errors at the product specification level. It differs from the whole-board compensation amount in step S31. The whole-board compensation amount comes from the actual offset detection of the current board material based on the target coordinate data, reflecting the current physical posture deviation of the board material. The precise alignment compensation amount comes from the machine tool error correction parameters obtained by pre-calibration for different product specifications, reflecting the systematic deviation characteristics of the punching equipment when processing this type of board material, including but not limited to: the clamping position deviation caused by different board thicknesses, the target imaging offset caused by different copper thicknesses, and the reference error caused by different material expansion and contraction characteristics.

[0062] For example, the precision alignment compensation amount can be predetermined as follows: During the equipment debugging phase, for each product specification, multiple punching tests are performed using a standard template. The measured positioning deviations after punching are statistically analyzed, and the obtained systematic error is entered into the industrial control host as the precision alignment compensation amount corresponding to that product specification and stored in the product parameter library. When a PCB board of a certain product specification enters the punching process, the industrial control host automatically retrieves the corresponding front and back precision alignment compensation amounts from the parameter library based on the current product specification information. The precision alignment compensation amount and the overall board compensation amount are different types of compensation amounts, and they are stored and retrieved separately.

[0063] It is understandable that the superposition in step S33 refers to vector addition of the overall board compensation amount and the precise alignment compensation amount in the X and Y directions, respectively, and angular addition in the θ direction. Since both the overall board compensation amount and the precise alignment compensation amount include translational components in the X and Y directions and rotational components around the Z axis, the final compensation amount after superposition also includes translational components in the X and Y directions and rotational components around the Z axis. Specifically, the X-direction component of the final compensation amount on the front side is the sum of the X-direction translational amount of the overall board compensation amount on the front side and the X-direction translational amount of the precise alignment compensation amount on the front side, and the Y-direction component is calculated similarly, as is the θ-direction component. The calculation method for the final compensation amount on the back side is the same, but the overall board compensation amount and the precise alignment compensation amount on the back side are superimposed. From a functional perspective, the overall board compensation amount addresses the deviation between the actual placement posture of the PCB board and the theoretical benchmark, which is a feedback compensation based on real-time target detection; the precise alignment compensation amount addresses the systematic positioning deviation of the punching equipment for different specifications of board materials, which is a feedforward compensation based on offline calibration. Feedback compensation ensures the correction of individual plate posture deviations, while feedforward compensation ensures the correction of equipment system deviations. When the two are combined, the final compensation amount takes into account both individual difference correction and system error correction, ensuring that the punching positioning accuracy meets the micron-level processing requirements.

[0064] In some implementations, after the final compensation amount is generated by superposition, it is also necessary to verify whether the final compensation amount of each punch position exceeds the preset global locking boundary. The "global locking boundary" refers to the maximum allowable compensation range preset according to the physical travel limit and process safety requirements of the equipment's three-axis servo attitude adjustment platform. If the final compensation amount of a certain punch position exceeds the locking boundary, the boundary value is used as the actual compensation amount of that punch position to avoid the theoretically calculated compensation amount exceeding the physical execution capability of the equipment.

[0065] Thus, in this implementation method, the compensation amount for the front and back sides of the PCB board is calculated independently based on the target offset data of each side. This accurately adapts to the differential deformation caused by etching, lamination and other previous processes on the front and back sides of the PCB board, avoiding the crude mode of sharing the same set of compensation amounts on both sides in the single-sided imaging scheme. At the same time, it introduces fine alignment compensation based on product specifications, setting compensation amounts for different PCB board product specifications, effectively compensating for the systematic positioning deviation of the machine caused by factors such as board thickness, copper thickness and fixture characteristics, and improving the equipment's adaptability to products of different specifications. The whole board compensation and fine alignment compensation are layered independently and do not interfere with each other. The whole board compensation based on real-time target detection solves the individual posture deviation of each board, while the fine alignment compensation based on offline calibration solves the systematic deviation of the equipment for different specifications of boards. The two are superimposed to correct the positioning errors from two different sources, forming a complete error compensation chain. This takes into account the individual differences of each board in the mass production line and the system adaptation requirements of the equipment for different specifications of products, ensuring that the punching positioning accuracy meets the micron-level processing requirements.

[0066] In some implementations, the two-sided offset data includes the coordinates of each target on the upper plate surface. x u,i , yu,i ) and the coordinates of each target on the lower plate ( x d,i , y d,i Prior to the differential compensation, the method further includes: The layer offset of the homologous target is calculated based on the aforementioned two-sided offset data. Δx bias,i , Δy bias,i ):

[0067] If the layer deviation exceeds the preset acceptable tolerance, the PCB board is determined to be defective and a diversion and sorting process is performed.

[0068] In this context, "homogeneous targets" refer to two Mark points on the front and back of a PCB board that are physically located at the same corner. Specifically, taking an inner layer PCB as an example, there is one Mark point on each of its front and back sides at the same corner (e.g., the upper left corner). These two Mark points completely overlap during the design phase; they originate from the same film data and have the same theoretical coordinate values, only located on opposite sides of the board. These two Mark points are called "homogeneous targets" because they originate from the same design data.

[0069] In this embodiment, firstly, defective products with layer misalignment are intercepted before the punching process to prevent high-priced core boards from flowing into the lamination process and causing mass scrap. In the manufacturing of multilayer PCBs, the front and back patterns of the inner core board are often affected by factors such as etching parameter fluctuations, exposure alignment deviations, and material stress release, resulting in varying degrees of misalignment (i.e., layer misalignment). If the layer misalignment exceeds the standard, the subsequent lamination will inevitably fail to meet the interlayer alignment accuracy requirements, leading to the scrapping of the entire board. Traditional single-sided imaging schemes cannot obtain the target coordinates on the lower board surface, and layer misalignment defects cannot be identified during the punching process. Core boards with layer misalignment defects can only be detected by X-ray after punching and lamination are completed. By this time, the processing time, material costs, and lamination resources already invested are irrecoverable. This embodiment relies on a double-sided synchronous imaging architecture, which can calculate the layer misalignment amount under the optimal alignment posture. Layer misalignment judgment and defective product diversion are completed before the punching operation, fundamentally preventing defective products with layer misalignment from continuing to occupy processing resources. Secondly, the calculation formula uses simple coordinate difference operations, resulting in extremely low computational load. Eight differences at the four corner positions are compared one by one with preset acceptable tolerances. All judgment logic is automatically completed by the industrial control host without manual intervention, and the judgment process is extremely short, without affecting production line cycle time. Thirdly, PCB board layer misalignment may exhibit a complex spatial distribution (severe misalignment only at local corner positions while other corner positions are normal). Traditional X-ray sampling methods easily miss localized layer misalignments. This implementation method independently calculates and judges the layer misalignment at each of the four corner positions. If any position exceeds the standard, a release interception is triggered, ensuring the comprehensiveness and reliability of layer misalignment detection.

[0070] Furthermore, this implementation method treats layer deviation calculation and defect diversion as preliminary steps before the differentiated compensation step. This means that the complete technical solution of this application not only achieves "parameter closed-loop optimization" but also includes the quality control function of "layer deviation pre-verification based on double-sided data". Both pre-defect screening and layer deviation verification are performed after obtaining the double-sided target coordinate data and before entering the dynamic screening operation. That is, pre-defect screening performs initial screening of defective plates based on the single-point offset threshold and the double-sided deformation difference threshold, while layer deviation verification intercepts plates with excessive layer deviation based on the coordinate difference of the same source target. The two work together to complete the comprehensive screening and diversion of three types of defective plates—excessive offset, deformation imbalance, and excessive layer deviation—before the punching alignment operation starts, reducing ineffective processing losses and scrap costs from the processing source. At the same time, the calculation results of the layer deviation data ( Δx bias,i , Δy bias,i This data will also be stored and sent to back-end testing equipment as part of the process data, providing data support for subsequent parameter adjustments and quality traceability, and forming a close collaboration with the subsequent "storing process data and sending it to back-end testing equipment".

[0071] Step S3: Store the process data and send it to the back-end testing equipment.

[0072] Back-end inspection equipment refers to specialized equipment installed after the PCB punching process to perform quality inspection on the punched PCB board. Back-end inspection equipment can be X-ray inspection equipment. X-ray inspection equipment utilizes the penetrating power of X-rays to perform non-destructive inspection of the internal structure of multilayer PCB boards. Because X-rays can penetrate the outer layers of the PCB board and directly irradiate the inner layer reference points (i.e., the targets or positioning marks of each inner layer), it can clearly visualize and inspect the accuracy of inner layer alignment and whether there is any misalignment. This equipment can measure and inspect the expansion and contraction offset of the inner layer reference points of multilayer printed circuit boards during the lamination process, as well as the quality characteristics of lamination. Through internal imaging and measurement, X-ray inspection equipment can display the position and data of misalignment, tilt, elongation, contraction, and distortion between inner layers.

[0073] Here, the back-end testing equipment must have at least the following functions: layer misalignment detection, which detects the interlayer alignment deviation between the inner layers of the PCB board (especially between the front and back patterns); expansion and contraction detection, which detects the degree of expansion and contraction of the inner layers of the PCB board after lamination; and data output, which outputs the test results (i.e., the actual test data) in digital form and interacts with the industrial control host of the punching equipment through communication interfaces such as Ethernet.

[0074] It should be noted that the purpose of sending the process data to the back-end inspection equipment is to provide the X-Ray inspection equipment with a complete parameter background for this punching operation, including the target coordinates, screening parameters, and compensation coefficients used for this punching. After the X-Ray equipment completes the layer misalignment and expansion / contraction detection of the PCB board, it can compare the actual detection results (such as actual layer misalignment and actual expansion / contraction) with the received process data to determine the actual effect of the punching alignment parameters, providing a basis for subsequent parameter adjustments.

[0075] For example, the process data can be sent in real time after the punching operation is completed, or it can be sent in batches after the entire batch of plates has been processed. The data format can be JSON, XML, or other structured data formats to ensure data interoperability and parsing compatibility between the punching equipment and the X-Ray inspection equipment.

[0076] Step S3 stores the full-dimensional process data of this punching operation and actively sends it to the back-end testing equipment. Firstly, it breaks down the data barrier between the punching and testing processes in traditional solutions, enabling the testing stage to obtain complete background information on process parameters and establishing a data channel for subsequent data comparison and closed-loop processing. Secondly, the sent process data includes the screening parameters and compensation coefficients used in this punching operation. When the testing equipment returns the actual layer deviation and expansion / contraction test results, the industrial control host can compare the actual test data with historical process data, accurately identify the source of deviation, and guide targeted adjustments to the screening parameters and / or compensation coefficients, ensuring that the test results truly become parameters for iteration. This serves as the driving force for optimization. Simultaneously, the stored multi-dimensional process data constitutes a complete digital archive of the punching operation for each PCB board. When quality anomalies occur in subsequent processes, the specific parameter settings of the punching stage can be quickly traced back, providing data support for root cause analysis and continuous improvement of production line quality. Furthermore, X-Ray inspection, as a non-destructive testing method, can accurately measure inner layer misalignment and expansion / contraction without damaging the PCB board. Combined with the process data transmission mechanism of this step, the punching process can obtain objective feedback data without increasing additional losses, truly achieving data-driven closed-loop quality control and comprehensively improving production line quality consistency and process intelligence.

[0077] Step S4: Receive the actual test data fed back by the testing equipment, and adjust the screening parameters of the shrinkage iteration calculation and / or the compensation coefficient of the differential compensation based on the comparison results between the actual test data and the process data.

[0078] In some implementations, step S4 involves adjusting the sieving parameters and / or the compensation coefficients for differential compensation in the narrowing iteration calculation based on the comparison results between the actual detection data and the process data. This includes: Statistical outlier removal is performed on the detection data, and the detection data that does not exceed the preset information range is taken as valid feedback data; Based on effective feedback data, the process deviation trend of the current batch is calculated, and the screening parameters and / or compensation coefficients are gradually adjusted according to the preset upper limit of parameter change rate.

[0079] It's important to understand that the actual test data returned by the back-end Ray inspection equipment may contain a small amount of outlier data that deviates from the normal range due to factors such as localized PCB warping, interference from the testing environment, and occasional fluctuations in the equipment's sensors. If all the feedback data is directly used for parameter adjustments, these outliers may mislead the adjustment direction, causing the process parameters to deviate from their optimal values. Therefore, before using the feedback data, it is necessary to statistically remove outliers.

[0080] Statistical outliers, also known as outliers or gross errors, refer to individual data points in a set of test data that differ significantly from other data and deviate markedly from the normal distribution range. These outliers may arise from non-systematic factors such as measurement errors, occasional equipment malfunctions, environmental interference, or local anomalies of the tested object, such as detection deviations caused by local warping of the board surface.

[0081] In some implementations, the method further includes: calculating the mean and standard deviation of the detection data; for each detection data, if the difference between the detection data and the mean is less than or equal to three times the standard deviation, then the detection data is determined to fall into a preset confidence region.

[0082] It should be noted that the difference between the test data and the mean here refers to the absolute value of the difference between the test data and the mean.

[0083] In this embodiment, the 3σ criterion (also known as the Raida criterion) is used for outlier identification and removal. The basic principle of the 3σ criterion is as follows: assuming the test data follows a normal distribution, calculate the mean μ and standard deviation σ of the data set; for each test data point x, if its deviation from the mean exceeds three times the standard deviation, i.e., |x-μ|>3σ, then the data is determined to be an outlier (abnormal data) and removed; if |x-μ|≤3σ, then the data is determined to fall within a pre-set confidence region and is retained as valid feedback data. The statistical basis of the 3σ criterion is that under a normal distribution, approximately 99.7% of the data falls within the range of the mean ± 3σ, and data exceeding this range accounts for only 0.3%, which is a low-probability event and can be considered abnormal.

[0084] Specifically, in this embodiment, after the industrial control host receives a batch of actual test data from the X-Ray device (such as the layer deviation test results of several PCB boards in the same batch), it first calculates the mean μ and standard deviation σ of the batch of data; then it judges each test data one by one, removes outliers that exceed the μ±3σ range, and only uses the valid feedback data that falls within the range for subsequent process deviation trend calculation and parameter adjustment.

[0085] In this embodiment, the confidence region is defined with the mean as the center and three times the standard deviation as the boundary. This not only effectively eliminates a very small percentage of abnormal data based on the statistical law of normal distribution and ensures that the feedback data involved in parameter adjustment is statistically representative, but also avoids the accidental deletion of normal fluctuation data due to an overly narrow threshold or the omission of real anomalies due to an overly wide threshold. This ensures that the adjustment direction of the screening parameters and compensation coefficients is accurate and reliable, and avoids parameter oscillation caused by being misled by single-point abnormal data.

[0086] The process deviation trend refers to the direction and degree of systematic deviation of the actual test results relative to the expected process target, obtained through statistical analysis of effective feedback data from multiple PCBs in the same batch. Specifically, firstly, for each PCB that has been punched and completed X-ray inspection, the difference between its actual test value (such as actual layer offset or actual expansion / contraction) and the corresponding process data (such as the expected layer offset or expansion / contraction value used for this punching) is calculated to obtain the deviation amount for a single board. Then, the deviation amounts of all effective feedback data in the current batch are statistically analyzed, and the consistency of the deviation direction (e.g., whether the layer offset of multiple boards is biased in the same direction) and the cumulative amount (e.g., the average, median, or cumulative sum of the deviation amounts) are analyzed to determine whether there is a systematic shift in the current process parameters. For example, if the layer offset test results of 10 PCBs in a batch all show a positive deviation in the X direction (i.e., the upper layer pattern is uniformly biased to the right relative to the lower layer pattern), and the deviation amount shows a gradually increasing trend, it indicates that the current compensation coefficient has a systematic deficiency in the X direction and needs adjustment.

[0087] The preset parameter change rate limit refers to the maximum range of change allowed for each type of adjustable parameter (such as screening parameters, compensation coefficients, etc.) that is preset by process engineers based on equipment characteristics and process requirements. This limit is usually expressed as a percentage or an absolute value, such as "the compensation coefficient shall not exceed 5% of the current value each time it is adjusted" or "the screening radius shall not exceed 0.1 mil each time it is adjusted".

[0088] It should be noted that the core purpose of setting an upper limit for the rate of change of parameters is to prevent over-modulation of parameters from causing system instability. In an adaptive control system, if the adjustment range is too large in a single step, the parameter may cross the optimal value in one step, causing system oscillation or even collapse. By limiting the rate of change of a single adjustment, the parameter is ensured to gradually approach the optimal value in a smooth and controllable manner, thus ensuring the stability of the system and the reliability of the process.

[0089] In the above implementation, the 3σ criterion is used to statistically remove outliers from the X-Ray feedback data, filtering out abnormal data introduced by occasional factors such as local warping and detection interference. This ensures that the feedback data involved in parameter adjustment is statistically representative and avoids misleading parameter adjustment directions due to single-point outliers. Based on this, the process deviation trend of the current batch is calculated using effective feedback data to accurately identify the systematic shift direction and degree of parameters, providing clear data support for parameter adjustment rather than blind trial and error. Simultaneously, a preset upper limit for the parameter change rate constrains the magnitude of each adjustment, preventing system oscillations or process instability caused by parameter overmodulation due to fluctuations in single-batch data. This ensures that parameters gradually approach the optimal value in a smooth and controllable manner. These three elements work together: outlier removal ensures the reliability of the input data, deviation trend calculation provides the direction for adjustment, and the upper limit constraint on the change rate ensures the safety of the adjustment process. Together, they achieve stable, reliable, and adaptive iterative optimization of the screening parameters and compensation coefficients, effectively solving the technical problem in traditional solutions where process parameters rely entirely on manual experience for adjustment and cannot be continuously calibrated based on actual processing results.

[0090] In some implementations, acquiring the double-sided offset data output during the PCB punching alignment process includes: Four sets of CCD cameras, one above and one below the PCB board, are used to simultaneously acquire full-area images of the front and back sides. The acquired double-sided global image is sequentially converted to grayscale, Gaussian filtered for noise reduction, and edge extraction is performed. Invalid noise contours are removed based on contour area and roundness, and the Mark contour image is output. Based on the Mark contour image, identify the pixel coordinates of each Mark target in its corresponding camera pixel coordinate system; The target coordinate data is obtained by transforming the coordinates of each pixel to a unified device world coordinate system through a multi-camera unified world coordinate calibration method. The multi-camera unified world coordinate calibration method includes: manipulating four sets of CCD cameras at the top and four sets at the bottom, for a total of eight CCD cameras, to photograph a standard chessboard calibration board, completing lens distortion correction and multi-camera spatial pose calculation, and generating and storing the transformation matrix from the pixel coordinate system of each camera to the unified device world coordinate system.

[0091] In this embodiment, the PCB punching and alignment equipment has four sets of industrial CCD cameras positioned above and below the PCB board's transport path, totaling eight imaging units. The four upper CCD cameras are used to acquire target images on the PCB board's surface, while the four lower CCD cameras are used to acquire target images on the lower surface of the board. Each CCD camera is equipped with a high-resolution lens and a ring-shaped supplementary light source. The light source can employ a low-angle ring illumination method to ensure a clear contrast between the mark feature points and the PCB board background. All eight CCD cameras are electrically connected to a synchronous trigger board, which uniformly issues trigger signals to achieve microsecond-level synchronous image capture by the eight cameras.

[0092] For example, after the PCB board is transported to the punching station by the conveyor and pre-alignment coarse adjustment is completed, the industrial control host sends an image acquisition command to the synchronous trigger board. The synchronous trigger board simultaneously outputs trigger pulse signals to eight CCD cameras. At the same time they receive the trigger signals, the eight cameras take pictures of the four corner areas of the upper and lower surfaces of the PCB board, respectively, and acquire full-area images of the local areas containing each Mark feature point.

[0093] It should be noted that the eight-channel CCD camera uses a synchronous rather than time-division acquisition method. The technical significance of this is that it avoids the inconsistency error in acquisition time caused by the slight displacement of the board material that may occur during time-division shooting (such as conveyor belt vibration, platform micro-motion, etc.). It ensures that the front four corner target data and the back four corner target data acquired at the same time are strictly aligned in time, providing original data with consistent time reference for subsequent double-sided layer offset calculation.

[0094] For example, after the eight CCD cameras complete synchronous acquisition, they transmit the acquired global images to the industrial control host in digital signal form. The industrial control host performs the following preprocessing operations on each image channel in sequence: (1) Grayscale processing: If the original image acquired by the CCD camera is a color image, the industrial control host first converts it into a grayscale image. Grayscale processing integrates the information of the R, G, and B channels of the color image into single-channel grayscale information, which greatly reduces the amount of data for subsequent processing and improves the computational efficiency while preserving the contour and position information of the Mark feature points in the image.

[0095] (2) Gaussian filtering noise reduction: Gaussian filtering is applied to the grayscale image to eliminate various noises introduced during image acquisition, including electronic noise from the CCD sensor itself, ambient light interference, artifacts caused by minor scratches or dust particles on the PCB board. Gaussian filtering effectively suppresses noise while preventing the edge information of Mark feature points from being overly smoothed by weighting the grayscale value of each pixel in the image with the grayscale values ​​of its neighboring pixels.

[0096] (3) Edge extraction: Edge detection processing is performed on the denoised image. Specifically, the Sobel operator can be used to extract the contour edges of the Mark feature points. Since the Mark feature points are preferably circular markers, their edges have a closed loop structure, and the edge detection algorithm can accurately extract the loop contour. For the extracted edge points, sub-pixel interpolation methods (such as gray-level centroid method or Gaussian fitting) can be further used to improve the positioning accuracy to the sub-pixel level.

[0097] (4) Invalid noise contour removal: In the edge extraction results, there may be invalid contours that are not Mark feature points caused by PCB board texture, slight scratches or residual stains. The industrial control host screens and filters all extracted contours according to the preset geometric parameters of Mark feature points (such as roundness, area range, diameter range, etc.), removes invalid noise contours that do not conform to the preset geometric features, and retains only the valid contours that match the geometric attributes of Mark feature points, i.e., Mark contour images.

[0098] Further, after image preprocessing, the industrial control host performs center positioning calculations on the retained valid Mark contours. For circular Mark feature points, the centroid of the region enclosed by the contour is calculated as the center position of the Mark feature point, output in pixel coordinates. Subsequently, using the coordinate transformation matrix generated by the pre-completed multi-camera unified world coordinate calibration, the pixel coordinates of each Mark feature point are transformed one by one to the device's unified world coordinate system, obtaining the actual physical coordinates in millimeters. After the above processing, the actual coordinates of the four Mark feature points on the upper plate and the four Mark feature points on the lower plate are finally extracted for subsequent screening and alignment calculations.

[0099] Through the above implementation method, the synchronous acquisition mechanism of eight CCD cameras eliminates the displacement error that may be introduced by time-division acquisition, ensuring strict alignment of the front and back target data in the time dimension. Through a series of preprocessing operations such as grayscale conversion, Gaussian filtering, edge extraction, and invalid contour removal, various noise interferences are effectively filtered out, improving the accuracy and stability of Mark feature point recognition. With the unified world coordinate calibration of multiple cameras, the independent pixel coordinates of each camera are uniformly transformed to the same physical reference, providing original data with a consistent spatial reference for subsequent double-sided layer offset verification and double-sided difference compensation.

[0100] In summary, the above embodiments of this application acquire double-sided offset data and corresponding process data generated based on double-sided target data during the punching alignment process, store and send them to the back-end testing equipment, and then use the comparison results of the actual testing data fed back by the testing equipment with the process data to drive the adaptive adjustment of screening parameters and / or compensation coefficients. This forms a complete closed loop from punching data acquisition to process parameter determination, then to testing feedback acquisition, and finally to parameter iterative update. This solves the technical problem in the prior art that process parameters rely entirely on manual experience for debugging and cannot be continuously optimized based on actual processing results. It achieves the technical effect of breaking down the data barrier between the punching process and the back-end testing process and realizing adaptive iterative optimization of process parameters.

[0101] Based on the same inventive concept as the foregoing embodiments, this application provides a computing device, such as... Figure 4 As shown, the device includes: a processor 310 and a memory 311 storing a computer program; wherein, Figure 4 The processor 310 shown in the diagram does not indicate that there is only one processor 310, but only indicates the positional relationship of the processor 310 relative to other devices. In practical applications, there can be one or more processors 310; similarly, Figure 4 The memory 311 illustrated in the diagram has the same meaning, that is, it is only used to indicate the positional relationship of memory 311 relative to other devices. In practical applications, the number of memories 311 can be one or more. When the processor 310 runs the computer program, it implements the parameter adjustment method in the PCB punching process applied to the above-mentioned device. The device may also include: at least one network interface 312. The various components in the device are coupled together through a bus system 313. It is understood that the bus system 313 is used to realize the connection and communication between these components. In addition to the data bus, the bus system 313 also includes a power bus, a control bus, and a status signal bus. However, for clarity, in Figure 4 The general designated all buses as Bus System 313.

[0102] The memory 311 can be volatile memory or non-volatile memory, or both. Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), ferromagnetic random access memory (FRAM), flash memory, magnetic surface memory, optical disc, or compact disc read-only memory (CD-ROM); magnetic surface memory can be disk storage or magnetic tape storage. Volatile memory can be random access memory (RAM), used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Synchronous Static Random Access Memory (SSRAM), Dynamic Random Access Memory (DRAM), Synchronous Dynamic Random Access Memory (SDRAM), Double Data Rate Synchronous Dynamic Random Access Memory (DDRSDRAM), Enhanced Synchronous Dynamic Random Access Memory (ESDRAM), SyncLink Dynamic Random Access Memory (SLDRAM), and Direct Rambus Random Access Memory (DRRAM).The memory 311 described in the embodiments of this application is intended to include, but is not limited to, these and any other suitable types of memory.

[0103] The memory 311 in this embodiment is used to store various types of data to support the operation of the device. Examples of this data include: any computer programs used to operate on the device, such as operating systems and applications; contact data; phonebook data; messages; pictures; videos, etc. The operating system includes various system programs, such as the framework layer, core library layer, driver layer, etc., used to implement various basic services and handle hardware-based tasks. Applications can include various applications, such as media players, browsers, etc., used to implement various application services. Here, the program implementing the method of this embodiment can be included in the application.

[0104] Based on the same inventive concept as the foregoing embodiments, this embodiment also provides a computer-readable storage medium storing a computer program. The computer-readable storage medium can be a magnetic random access memory (FRAM), a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a flash memory, a magnetic surface memory, an optical disc, or a compact disc read-only memory (CD-ROM), etc.; it can also be various devices including one or any combination of the above-mentioned memories, such as mobile phones, computers, tablet devices, personal digital assistants, etc. When the computer program stored in the computer-readable storage medium is executed by a processor, it implements the parameter adjustment method in the above-mentioned PCB punching process. For the specific steps implemented when the computer program is executed by the processor, please refer to [link to relevant documentation]. Figure 1 , Figure 2 and Figure 3 The description of the illustrated embodiments will not be repeated here.

[0105] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0106] In this document, the terms “comprising,” “including,” or any other variations thereof are intended to cover non-exclusive inclusion, which includes not only the elements listed but also other elements not expressly listed.

[0107] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A parameter adjustment method in PCB punching process, characterized in that, Includes the following steps: The double-sided offset data output during the PCB punching alignment process is obtained. The double-sided offset data is the front and back offset data corresponding to the optimal alignment posture selected by step-by-step shrinking iterative calculation based on the target coordinate data at the four corners of the front and back of the PCB. The step-by-step shrinking iteration calculation includes: obtaining the preset reference coordinates of the four corners of the front and back of a standard PCB board as theoretical reference points; using the straight-line distance between the actual coordinates of each target in the current posture and the corresponding theoretical reference point as the target offset distance, and constructing a circular screening domain with the theoretical reference point as the screening center and the target offset distance as the screening radius; taking the four corner points of the inscribed square parallel to the axis of the circular screening domain as candidate points, calculating the cumulative value of the single-sided four-point offset corresponding to each candidate point, the cumulative value of the single-sided four-point offset is the sum of the offset distances of the four targets on the same board surface; and using the candidate point corresponding to the minimum cumulative offset value as the current screening point. The optimal target point within the domain; if the distance between the theoretical reference point and the optimal target point is less than a preset accuracy threshold, then the posture corresponding to the current optimal target point is taken as the optimal alignment posture; otherwise, the length of the line connecting the theoretical reference point and the current optimal target point is taken as the diameter of the new circular screening domain, and the midpoint of the line is taken as the new screening center. The above steps of candidate point selection and optimal target point determination are repeated until the accuracy requirements are met or the mechanical convergence limit is reached. The mechanical convergence limit is: the distance between the optimal target points obtained in two adjacent iterations is less than the minimum executable step distance of the three-axis servo attitude adjustment platform; the cumulative value of the single-plane four-point offset is calculated according to the following formula: in, L i The target offset distance at the lower corner position in the current attitude. L sum This is the cumulative total of the four-point offset on a single surface. x i , y i ) is the first i The actual coordinates of each target, ( x 0i , y 0i ) is the first i Coordinates of one theoretical benchmark point; Based on the double-sided offset data, the process data for this punching operation is determined. The process data is based on the double-sided offset data, and the punching positions on the front and back sides of the PCB board are differentiated and compensated, and the punching operation is performed based on the compensated positioning. Store the process data and send it to the backend testing equipment; Receiving actual test data fed back by the testing equipment, and adjusting the screening parameters of the narrowing iteration operation and / or the compensation coefficient of the differential compensation based on the comparison results of the actual test data and the process data, includes: statistically removing outliers from the test data, and taking the test data that does not exceed the preset information region as valid feedback data; calculating the process deviation trend of the current batch based on the valid feedback data, and gradually adjusting the screening parameters and / or the compensation coefficient according to the preset upper limit of parameter change rate.

2. The method according to claim 1, characterized in that, The differential compensation includes: Based on the double-sided offset data, determine the translation and rotation of the entire plate on the front and back sides respectively, and generate the compensation amount for the entire plate on the front side and the compensation amount for the entire plate on the back side. Based on the current PCB board product specifications, determine the front and back precision alignment compensation amounts corresponding to the product specifications. The compensation coefficient includes the point-level fine-tuning amount of each punch position generated during the double-sided differential compensation process, as well as the plate-level global compensation amount of the front side and the plate-level global compensation amount of the back side formed by superimposing the point-level fine-tuning amounts; the front whole plate compensation amount is superimposed with the front fine alignment compensation amount to generate the front final compensation amount; the back whole plate compensation amount is superimposed with the back fine alignment compensation amount to generate the back final compensation amount.

3. The method according to claim 1, characterized in that, The dual-plane offset data includes the coordinates of each target on the upper plate ( x u,i , yu,i ) and the coordinates of each target on the lower plate ( x d,i , y d,i ); Prior to the differential compensation, the method further includes: The layer offset of the homologous target is calculated based on the aforementioned two-sided offset data. Δx bias,i , Δy bias,i ): If the layer deviation exceeds the preset acceptable tolerance, the PCB board is determined to be defective and a diversion and sorting process is performed.

4. The method according to claim 1, characterized in that, The acquisition of double-sided offset data output during the PCB punching alignment process includes: Four sets of CCD cameras, one above and one below the PCB board conveying path, are used to simultaneously acquire full-area images of the front and back sides. The acquired double-sided global image is sequentially converted to grayscale, Gaussian filtered for noise reduction, and edge extraction is performed. Invalid noise contours are removed based on contour area and roundness, and the Mark contour image is output. Based on the Mark contour image, identify the pixel coordinates of each Mark target in its corresponding camera pixel coordinate system; The target coordinate data is obtained by transforming the coordinates of each pixel to a unified device world coordinate system using a multi-camera unified world coordinate calibration method. The multi-camera unified world coordinate calibration method includes: manipulating four sets of CCD cameras at the top and four at the bottom, for a total of eight CCD cameras, to photograph a standard checkerboard calibration board, completing lens distortion correction and multi-camera spatial pose calculation, and generating and storing the transformation matrix from the pixel coordinate system of each camera to the unified device world coordinate system.

5. The method according to claim 1, characterized in that, The method further includes: Calculate the mean and standard deviation of the test data; For each detection data point, if the difference between the detection data point and the mean is less than or equal to three times the standard deviation, then the detection data point is determined to fall into the preset confidence region.

6. A computing device, characterized in that, include: A processor and a memory storing a computer program, wherein when the processor runs the computer program, the steps of the parameter adjustment method in the PCB punching process according to any one of claims 1 to 5 are implemented.

7. A computer storage medium, characterized in that, The device contains a computer program that, when executed by a processor, implements the steps of the parameter adjustment method in the PCB punching process according to any one of claims 1 to 5.

Images