A die cutting machine local pressure online feedback compensation system and method

Abstract

The application discloses a die-cutting machine local pressure online feedback compensation system and method, and belongs to the technical field of automatic control of printing and packaging machinery. The system comprises an online detection module, a distributed pressure adjustment module and a control unit. The online detection module collects images after each die-cutting cycle, identifies die-cutting defects and position coordinates. The distributed pressure adjustment module divides the working plane into multiple independent adjustment units, and selectively applies additional pressure to the local area of the die. The control unit receives the defect coordinates, determines the corresponding adjustment unit according to the preset mapping relationship, and drives it to perform the additional pressure action in the next die-cutting cycle to realize single-cycle point compensation. The application can automatically repair the local die-cutting non-penetrating defects in the flat-pressing flat-die-cutting without stopping the machine, greatly reduces the waste rate, and improves the comprehensive efficiency of the equipment.

Description

Technical Field

[0001] This invention relates to the field of printing and packaging technology, and in particular to an online feedback compensation system and method for local pressure in a die-cutting machine. Background Technology

[0002] In the die-cutting of thin materials such as paper and cardboard, flatbed die-cutting machines are widely used due to their high die-cutting pressure and wide applicability. However, during operation, the cutting lines on the die plate rub against the paper due to long-term cyclical high pressure, easily leading to localized wear. Simultaneously, the elastic pads on the back of the die plate are repeatedly compressed, resulting in uneven thickness and localized collapse. These factors combined cause insufficient protrusion of the cutting lines in certain areas of the die plate, even with sufficient overall die-cutting pressure, resulting in the defect of "incomplete die-cutting," where the paper is not completely cut, leaving residual fine fibers connecting the layers. This defect seriously affects subsequent waste removal processes and product quality, and has long been a pain point for the industry.

[0003] Currently, the common method for addressing partial die-cutting failure is to stop the machine and have the operator attach thin padding paper to the corresponding area on the back of the die plate, relying on experience to repeatedly test cut and adjust. This method is heavily reliant on manual labor, generates a large number of defective products during the debugging period, and cannot detect and repair sudden pressure drops during the production process in real time.

[0004] To reduce manual intervention, some automatic die-cutting pressure adjustment technologies have emerged. For example, patent document CN115431346A discloses an automatic die-cutting mechanism and equipment. This solution uses a visual recognition unit to detect cutting defects and automatically adjusts the pressure of the die-cutting roller through the cooperation of a pressure adjustment unit and a pressure sensing unit to achieve a set pressure range and realize automatic operation. However, this prior art is aimed at rotary die-cutting machines, and its pressure adjustment unit is configured to adjust the overall pressure of each independent rotary die-cutting unit. For flatbed die-cutting machines, a large-area blade completes the die-cutting of the entire width simultaneously. Simply raising or lowering the overall pressure will result in excessive pressure in areas that were already within acceptable limits, causing the blade to wear out too quickly or even crush the paper. This fails to solve the core contradiction of "sufficient overall pressure but insufficient local pressure". Furthermore, the existing technology does not provide effective guidance on the specific hardware configuration of visual recognition, the algorithm logic for defect judgment, the real-time synchronization architecture of the control system, and the fixed-point compensation for local defects. Its control unit is only generally described as a controller that receives signals and performs pressure regulation, lacking circuit-level implementation means to complete the detection, mapping, and compensation closed loop within a single die-cutting cycle under high-speed continuous production conditions.

[0005] Therefore, how to implement differentiated and quantifiable real-time local pressure compensation for different areas of the same die plate of a flatbed die-cutting machine without interrupting production, and to achieve a rapid closed loop within a die-cutting cycle from defect detection to pressure execution, has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0006] This invention addresses the problems of existing automatic die-cutting adjustment schemes being unable to adapt to the local pressure differential compensation requirements of flatbed die-cutting machines, and lacking the hardware architecture and signal processing logic for achieving single-cycle closed-loop control at high production speeds. It proposes an online feedback compensation system and method for local pressure in die-cutting machines, which can automatically detect local die-cutting defects without stopping the machine or reducing speed, and accurately apply additional pressure to the corresponding position of the defect in the next die-cutting cycle, thereby significantly reducing the continuous scrap rate and significantly improving the overall efficiency of the equipment.

[0007] To achieve the above objectives, the following technical solution is proposed: A local pressure online feedback compensation system for a die-cutting machine, comprising: The online detection module is coupled to the paper output side of the die-cutting machine and is configured to acquire images of a preset detection area of ​​the die-cut product after each die-cutting cycle, and identify the die-cutting defects and their position coordinates on the die-cut product based on the pre-stored reference image data. A distributed pressure adjustment module is installed in the upper or lower die base of the die-cutting machine. Its effective working plane is divided into multiple independent adjustment units corresponding to different areas of the die plate. Each independent adjustment unit is configured to selectively apply additional pressure to a corresponding local area on the die plate on top of a base pressure. The control unit is communicatively connected to the online detection module and the distributed pressure regulation module, respectively. The control unit is configured to: receive the position coordinates of the die-cutting defect output by the online detection module, determine the independent adjustment unit corresponding to the position coordinates according to a preset mapping relationship, and drive the determined independent adjustment unit to perform an additional pressure action in the next die-cutting cycle to actively compensate for the die-cutting defect.

[0008] Through the above technical solutions, by setting up a distributed pressure adjustment module on the die-cutting machine that can independently apply pressure to different areas of the die plate, and establishing a coordinate mapping and synchronous triggering mechanism between the online detection module and the control unit, the system can accurately map the defect location in the image space to the specific pressure unit in the physical space, and complete the fixed-point compensation in the next die-cutting cycle following the detection cycle. This achieves a single-cycle closed loop from defect detection to repair, completely eliminating the continuous generation of defects.

[0009] Preferably, the online detection module includes: a linear image sensor installed across the paper's movement direction, configured to acquire images of the preset detection area line by line; a bar light source illuminating the preset detection area at a preset low angle to the plane of the die-cut product; and an image processing unit coupled to the linear image sensor, configured to compare the real-time acquired image grayscale signal with the reference image data point by point to identify the die-cutting defects and their location coordinates.

[0010] By employing the above technical solutions, the use of a linear array image sensor in conjunction with a low-angle grazing strip light source results in significant shadow contrast on the uncut fibers under grazing light illumination. Real-time grayscale comparison is then performed by the image processing unit, greatly improving the detection sensitivity and reliability of micro-connected fibers and effectively avoiding missed detections and false detections.

[0011] Preferably, the distributed pressure adjustment module includes: a base pressure plate coupled to the main pressure mechanism of the die-cutting machine for applying the base pressure to the die plate; and a matrix of piezoelectric ceramic actuators fixed between the base pressure plate and the die plate, each piezoelectric ceramic actuator constituting an independent adjustment unit and contacting a corresponding local grid area on the die plate; wherein the piezoelectric ceramic actuator driven by the control unit applies the additional pressure by outputting an additional displacement to compress the elastic pad of the corresponding local area of ​​the die plate.

[0012] Through the above technical solution, by using a piezoelectric ceramic actuator array built between the base pressure plate and the blade plate, the base pressure is kept uniformly applied as a whole, while each actuator can selectively generate micron-level additional displacement, thus achieving precise pressure application to local areas of the blade plate while maintaining the main pressure path unchanged. The structure is compact, the response speed is fast, and the original frame and main drive layout are not changed, resulting in low modification costs.

[0013] Preferably, the piezoelectric ceramic actuator incorporates a strain gauge displacement sensor, and the control unit is further configured to perform closed-loop control of the additional displacement based on feedback from the strain gauge displacement sensor; wherein, the size of the local grid region is configured to match the size of a typical wear area on the cutting plate, and the closed-loop control range of the additional displacement is configured to cover the amount of cutting edge height loss caused by wear on the cutting plate.

[0014] Through the above technical solutions, since the additional displacement adopts closed-loop control based on strain gauge displacement sensors, and the grid size and displacement range are set according to the actual wear characteristics of the blade, the pressure compensation can accurately cover typical local wear areas while avoiding interference with adjacent normal areas. At the same time, the displacement closed loop ensures high repeatability and long-term stability of the compensation amount.

[0015] Preferably, the control unit includes an image preprocessing subunit, a logic control subunit, and a main control subunit, which are connected via a real-time industrial Ethernet bus. The logic control subunit is configured to receive the spindle phase signal of the die-cutting machine to identify the phase of the die-cutting cycle. The main control subunit is configured to, after receiving the defect coordinates from the online detection module, trigger the additional pressure action within the pressure-combining phase of the next die-cutting cycle using the phase signal provided by the logic control subunit, thereby meeting the real-time requirement of completing the active pressure compensation within a single die-cutting cycle.

[0016] By dividing the control unit into three sub-units—image preprocessing, logic control, and main control—and coordinating them through a real-time bus, image processing, phase synchronization, and compensation decisions can be executed in parallel. This ensures that the process from defect image acquisition to triggering additional pressure action is completed within an extremely short time window, meeting the stringent real-time requirements of single-cycle compensation in high-speed die-cutting production lines.

[0017] Preferably, the online detection module is further configured to identify the die-cutting defect by: performing point-by-point difference calculation between the real-time acquired detection line grayscale curve and the pre-stored reference grayscale curve; when the difference at a certain point exceeds a preset grayscale threshold, and the length of consecutive pixels with the difference exceeding the grayscale threshold is greater than a preset length threshold, it is determined that the die-cutting defect exists at the position corresponding to that point.

[0018] By employing the above technical solutions, the defect identification logic based on a combination of gray-scale difference threshold and continuous abnormal length can not only filter out single-point gray-scale fluctuations caused by random noise, but also effectively capture continuous gray-scale abnormal areas formed by real uncut defects, thus significantly improving the accuracy and anti-interference ability of defect judgment.

[0019] Preferably, the preset mapping relationship in the control unit is generated by the following calibration method: the control unit sequentially drives each of the independent adjustment units to generate an identifiable indentation on the test material under no die-cutting pressure; the online detection module collects the image coordinates of the indentation, and the control unit binds the identifier of each of the independent adjustment units with the collected image coordinates to generate a mapping table.

[0020] By using the above technical solutions, the mapping relationship between the actuator and the image coordinates is generated by the in-situ automatic calibration method, which eliminates the influence of mechanical installation errors and optical distortion on positioning accuracy. This ensures that the defect coordinates can be accurately mapped to the corresponding independent adjustment unit during the production process, laying a key foundation for precise point compensation.

[0021] Preferably, the control unit is also configured to perform a compensation amount self-learning step: when the additional pressure action is performed on the same independent adjustment unit for N consecutive die-cutting cycles, and the online detection module still identifies a die-cutting defect at the position corresponding to the independent adjustment unit, the additional pressure value for the independent adjustment unit is automatically increased, where N is an integer greater than or equal to 2.

[0022] Through the above technical solutions, the system can automatically adapt to the continuous wear trend of the die plate and dynamically optimize the additional pressure value by introducing a compensation amount self-learning mechanism. It can maintain stable die-cutting quality for a long time without repeated manual adjustments and can actively prompt the replacement of the die plate when the wear exceeds the limit.

[0023] A method for online feedback compensation of local pressure in a die-cutting machine, applied to any of the systems described above, includes: Reference acquisition steps: Acquire and store reference image data of qualified die-cut products; Online inspection steps: After each die-cutting cycle, images of the preset inspection area of ​​the die-cut product are acquired in real time and compared with the reference image data to identify die-cutting defects and their location coordinates; Mapping compensation step: According to the preset mapping relationship, the position coordinates are converted into the identifier of the corresponding independent adjustment unit, and in the next die-cutting cycle, the determined independent adjustment unit is driven to perform additional pressure action to actively compensate for the die-cutting defects.

[0024] Through the above technical solutions, this method connects benchmark acquisition, online detection, and mapping compensation into an automated process that can be completed within a single die-cutting cycle. This reduces the time delay from the discovery of local die-cutting defects to their repair to the extreme, fundamentally changing the traditional downtime debugging operation mode and achieving true online real-time quality control.

[0025] Preferably, the mapping compensation step further includes: a phase synchronization sub-step: receiving the phase signal of the die-cutting machine spindle and identifying the phase of the current die-cutting cycle; and a triggering sub-step: when the pressure phase of the next die-cutting cycle is identified, triggering the additional pressure action and maintaining the additional pressure action within the pressure phase.

[0026] By precisely synchronizing the timing of the compensation action with the pressure phase of the die-cutting machine, the additional pressure is ensured to be applied only at the moment the die plate contacts the paper. This avoids both ineffective energy consumption and additional impact on the mechanism, while also ensuring the effectiveness and consistency of the localized deepening cutting effect.

[0027] Therefore, the present invention has at least the following beneficial effects: By dividing the effective area of ​​the die-cutting plate into multiple independently controllable local pressure units through a distributed pressure regulation module, and in conjunction with an online detection module to obtain defect coordinates, the system achieves fixed-point and quantitative online compensation for local die-cutting defects in flatbed die-cutting, completely solving the problem of "sufficient overall pressure, but insufficient local pressure" that traditional solutions cannot handle.

[0028] Through the architecture of the image preprocessing subunit, logic control subunit and main control subunit working together via real-time bus in the control unit, and the single-cycle synchronous triggering strategy based on the spindle phase signal, the entire detection-mapping-compensation closed loop is completed within a single die-cutting cycle, which meets the stringent real-time requirements of high-speed continuous production and avoids the continuous discharge of defective paper.

[0029] By employing a defect identification algorithm based on grayscale difference threshold and continuous abnormal length as dual criteria, and combining it with an optical system consisting of a low-angle grazing light source and a linear array image sensor, the detection rate and anti-interference capability of fine uncut fibers are significantly improved, providing reliable defect information for high-precision compensation. Attached Figure Description

[0030] Figure 1 This is a block diagram of the overall system structure in an embodiment of the present invention.

[0031] Figure 2 This is a schematic diagram of the installation structure of the online detection module.

[0032] Figure 3 This is a cross-sectional schematic diagram of the distributed pressure regulation module.

[0033] Figure 4 This is a schematic diagram of the hardware architecture of the control unit.

[0034] Figure 5 This is a flowchart of the online detection and compensation control method.

[0035] Figure 6 This is a schematic diagram of the actuator-image coordinate mapping calibration process. Detailed Implementation

[0036] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that the following embodiments are only for explaining the present invention and are not intended to limit the scope of protection of the present invention. Any simple modifications, equivalent substitutions, and improvements made by those skilled in the art to the following embodiments without departing from the concept of the present invention should be included within the scope of protection of the present invention. Example

[0037] This embodiment uses a cigarette pack hard box die-cutting production line as an application scenario. The die-cutting machine involved is an automatic flatbed die-cutting machine with a maximum paper size of 740mm × 1050mm, a maximum die-cutting speed of 5500 sheets / hour, and a main pressure of 200 tons. In this embodiment, the system is modified and upgraded to have online local pressure feedback compensation capability.

[0038] like Figure 1 As shown, the system in this embodiment consists of three main parts: an online detection module, a distributed pressure adjustment module, and a control unit. The online detection module is installed on the paper output side of the die-cutting machine and is used to acquire images of the preset detection line positions on the die-cut paper and identify defects after each die-cutting cycle. The distributed pressure adjustment module is installed inside the upper die holder of the die-cutting machine, and its working surface is divided into multiple independent adjustment units. Each unit can apply controllable additional pressure to a local area on the die plate. The control unit is communicatively connected to both the online detection module and the distributed pressure adjustment module, and is responsible for receiving defect coordinates, querying mapping relationships, and driving the corresponding adjustment unit to perform compensation actions in the next cycle.

[0039] The specific installation structure of the online detection module is as follows: Figure 2 As shown, a high-resolution linear CMOS camera is mounted above the chain conveyor belt on the paper output side of the die-cutting machine, spanning the width of the paper. This camera boasts a resolution of 8192 pixels, a pixel size of 7μm, and a line frequency of up to 30kHz. With a 25mm focal length lens, it covers a 740mm wide sheet of paper at a working distance of approximately 300mm, achieving an object-side resolution of approximately 90μm / pixel, reliably distinguishing connecting fibers larger than 0.2mm. A red LED high-brightness strip light source is positioned on the lower surface of the paper, directly below the camera's imaging line. The long axis of its light outlet is parallel to the camera's scanning line, and the light is projected at a low grazing angle of approximately 15° to the paper plane. When the die-cut is incomplete, residual fibers or rough edges form tiny raised shadows under the grazing light, appearing as localized dark areas or bright spots in the image. Conversely, the completely cut area exhibits a smooth and uniform background grayscale, resulting in high contrast.

[0040] Image signals are transmitted in real-time via a CameraLink interface to an FPGA-based image preprocessing subunit integrated within the control cabinet. This subunit is configured for hardware-accelerated image stitching, grayscale extraction of detection lines, and real-time defect identification. Specific identification algorithms will be combined with… Figure 5 This will be discussed in detail below.

[0041] The structural cross-section of the distributed pressure regulation module is as follows Figure 3As shown. While maintaining the original frame and main pressure cylinder, the upper mold base blade clamping area was modified into a three-layer structure consisting of a base pressure plate, an actuator grid pad, and the blade itself. The base pressure plate is a 30mm thick high-quality steel plate, bolted to the main pressure cylinder piston, with its lower surface precision ground. The actuator grid pad consists of a high-strength aluminum alloy frame and 315 independent piezoelectric ceramic actuators embedded within it. The frame evenly divides the effective working area of ​​750mm × 1050mm into a matrix of 15 columns horizontally and 21 rows vertically, with each grid approximately 50mm × 50mm. This grid size matches the typical defect area size on the blade caused by localized wear and elastic pad collapse, ensuring sufficient precision to accurately cover individual wear points without increasing system complexity and cost due to excessive fragmentation. Each piezoelectric ceramic actuator has a maximum stroke of 80μm, a maximum thrust of 800N, and a built-in strain gauge displacement sensor for outputting displacement feedback signals. Each actuator has a 50mm×50mm×5mm hardened steel pad bonded to its top, with the lower surface of the pad directly contacting the back plate of the cutter. Under normal conditions, the lower surfaces of all pads are strictly coplanar, and the base pressure applied by the main pressure cylinder is evenly transmitted to the cutter through the base pressure plate and all pads. When an actuator is driven, its pad is slightly pushed downwards, compressing the elastic pad on the back of the cutter within that grid area, increasing the protrusion of the corresponding cut line and thus locally deepening the cutting depth. The closed-loop control range for the additional displacement is 0-80μm, determined based on the typical cut line height loss due to wear (usually 20-60μm), which fully covers the compensation requirements. The drive power module for each actuator is mounted on the side of the upper die base and connected to the control unit via a shielded cable.

[0042] As an alternative implementation, the aforementioned distributed pressure regulating module can also be installed in the lower die holder of the die-cutting machine. This involves placing a grid-like actuator array below the anvil of the lower die holder, with each actuator pushing upwards to locally increase the support stiffness of the corresponding area of ​​the anvil, thus achieving the same effect of locally increasing the cutting pressure. The control logic and mapping calibration method of this implementation are the same as those in the above embodiment and will not be repeated here.

[0043] The hardware architecture of the control unit is as follows Figure 4As shown, the control unit consists of an image preprocessing subunit, a logic control subunit, and a main control subunit, which exchange data via an EtherCAT real-time industrial Ethernet bus. The image preprocessing subunit, implemented using an FPGA board, is specifically responsible for online processing of the high-speed image stream output from the line scan camera, including grayscale benchmark comparison and defect coordinate extraction. The logic control subunit, implemented using a PLC, has its digital input ports connected to the spindle rotary encoder and pressure proximity switch of the die-cutting machine. This allows for real-time acquisition of the spindle angle and pressure / disengagement status signals, thereby accurately identifying the phase of each die-cutting cycle. The main control subunit, implemented using an industrial computer running a Linux real-time operating system, handles human-machine interface interaction, actuator-image coordinate mapping table storage and retrieval, compensation decision-making and self-learning logic execution, and sending control commands to the distributed pressure regulation module. A touchscreen is installed on the operator's side, displaying a graphical die-cutting grid interface. Operators can view the compensation status of each grid in real time and perform manual fine-tuning.

[0044] The three sub-units work together via a real-time bus, enabling image acquisition and preprocessing, logical phase recognition, and compensation value calculation to be executed in parallel and in a streamlined manner. This effectively ensures the real-time requirements of defect detection and compensation execution within a single die-cutting cycle.

[0045] The online feedback compensation method flow in this embodiment is as follows: Figure 5 As shown, it mainly includes the benchmark acquisition and calibration stage and the online detection and compensation stage.

[0046] Baseline Acquisition and Calibration Phase: After the system is powered on for the first time or the die plate is changed, baseline image acquisition is performed first. The operator takes a die-cut sample sheet that has been confirmed to be fully qualified through offline quality inspection, manually feeds it into the paper output inspection station, and starts the acquisition program. The line scan camera scans the entire sample sheet, and the main control subunit extracts the preset inspection line position (such as the grayscale change band of key contours such as the edge of the cigarette pack flap and tear line) along the machine movement direction (Y direction) from the image, generates the grayscale baseline curve Gref(y) of the inspection line, and stores it. This baseline curve reflects the relationship between the pixel grayscale value of each point on the inspection line of the qualified product and the Y coordinate. Each Y coordinate value corresponds to an 8-bit grayscale value (0-255).

[0047] Then, actuator-image coordinate mapping calibration is performed. (Reference) Figure 6Under no master pressure, the main control subunit sequentially applies a 10ms pulse signal with a 30μm displacement to each piezoelectric ceramic actuator, causing it to leave a slight indentation on the blank cardboard being transported. Because this indentation forms a strip-shaped shadow with a significant grayscale difference from the surrounding planar area under low-angle grazing light illumination, the image preprocessing subunit can accurately extract the image coordinates of the indentation by detecting the center position of this grayscale anomaly band. Experimental results show that a 30μm indentation depth can produce a grayscale response spanning multiple pixels on an image with a 90μm / pixel resolution, and can be reliably identified. A linear scan camera continuously scans and captures images line by line. The image preprocessing subunit detects the horizontal image coordinate X of the indentation and the vertical Y coordinate corresponding to the trigger moment, and sends these coordinates to the main control subunit via the EtherCAT bus. The main control subunit establishes a one-to-one correspondence between the actuator number (i-th column, j-th row) and the image coordinates (Xi, Yj), generating a mapping table covering all 315 actuators. The mapping table is specifically a two-dimensional lookup table. The input is the image coordinates of the defect (X_def, Y_def), and the output is the corresponding actuator number. When the defect coordinates are near two grid boundaries, the nearest neighbor principle is used to select the actuator corresponding to the grid center closest to the defect coordinates. The calibration of 315 points takes approximately 3 minutes. The completed mapping table is stored in the non-volatile memory of an industrial computer for real-time lookup during production.

[0048] Online detection and compensation phase: During normal production, the die-cutting speed is set to 5000 sheets / hour, corresponding to a paper linear speed of approximately 1.0 m / s. The line scan camera's line frequency is set to 15 kHz to ensure a resolution of approximately 67 μm / pixel along the motion direction.

[0049] As each die-cut sheet of paper passes through the paper output detection station, the photoelectric sensor located at the die-cutting exit detects the leading edge of the paper and sends a trigger signal to the image preprocessing subunit, initiating a line scan of the preset detection area of ​​the paper. The image preprocessing subunit extracts only a narrow strip of image data within ±2mm of the detection line for real-time stitching and processing to reduce computational load. For the stitched grayscale image of the detection line, the current grayscale value Gcur(y) is obtained line by line, and a point-by-point difference calculation is performed with the stored reference grayscale curve Gref(y), i.e.: Diff(y) = |Gcur(y) - Gref(y)|; Where Diff(y) is the absolute value of the grayscale difference at Y coordinate position y; Gcur(y) is the measured grayscale value of the current paper at y; and Gref(y) is the grayscale value of the reference grayscale curve at y. The image preprocessing subunit internally presets a grayscale threshold T (40 in this embodiment, for an 8-bit grayscale range of 0-255) and a continuous length threshold L (0.3mm in this embodiment, corresponding to approximately 5 pixel rows). When Diff(y0) > T at a certain Y coordinate point y0, and the length of pixels with Diff(y) > T continuously appearing along the Y direction from that point exceeds L, it is determined that a "die-cutting not through" defect exists near the vertical position y0. The image preprocessing subunit calculates the vertical center coordinate Y_def of this continuous abnormal pixel segment, and combines it with the corresponding horizontal position X_def to form the defect coordinates (X_def, Y_def). Then, this coordinate is sent to the main control subunit via the EtherCAT bus in an interrupt manner, and simultaneously a sound and light alarm trigger command is sent to the touchscreen.

[0050] Upon receiving the defect coordinates, the main control subunit immediately queries a pre-stored mapping table to obtain the actuator number corresponding to those coordinates. For example, if the defect coordinates (X_def, Y_def) fall within the image coordinate range of the actuator in column 7, row 12 of the mapping table, the target actuator number is determined to be (7, 12). Afterward, the main control subunit enters a waiting state, continuously reading the spindle phase information uploaded by the logic control subunit via EtherCAT. The logic control subunit monitors the crankshaft angle in real time using an encoder and proximity switch. When it detects that the crankshaft angle has reached the preset trigger angle within the closing phase of the next die-cutting cycle (e.g., 185°, the critical angle at which the die plate begins to contact the paper), it immediately sends a compensation trigger flag to the main control subunit. Upon receiving the trigger flag, the main control subunit immediately applies an analog voltage value corresponding to a displacement of 40 μm to the piezoelectric ceramic actuator driver numbered (7, 12) through the analog output channel. The actuator's built-in strain gauge displacement sensor provides real-time feedback on the actual displacement. The PID controller inside the actuator performs closed-loop regulation based on this feedback, enabling the actuator to rapidly and stably apply a 40μm displacement above the base pressure and maintain this displacement throughout the entire pressure application period (approximately 0.1 seconds). After the pressure application ends, the actuator quickly returns to a zero-displacement state. Thus, on the next sheet of paper immediately following the defect, the localized area is successfully deepened, and the defect is eliminated.

[0051] To address the continuous wear of the die-cutting plate, the control unit also performs a self-learning compensation step. The main control subunit maintains a historical compensation counter for each actuator. If, for the same actuator number (i,j), additional pressure is applied in every single one of the three consecutive die-cutting cycles, and the online detection module still identifies a die-cutting failure in the corresponding area of ​​that actuator in subsequent detections, the main control subunit automatically increases the default additional displacement of that actuator from 40μm to 60μm. If a defect still occurs after increasing to 60μm three times consecutively, the displacement is increased to the upper limit of 80μm. When a defect still occurs after reaching the upper limit, the main control subunit highlights the grid on the touchscreen interface and prompts "Die-cutting plate is severely worn, replacement recommended." Simultaneously, the operator can manually set the permanent compensation value for this grid on the touchscreen. This value is stored as a parameter for the die-cutting plate recipe and can be directly retrieved the next time the same die-cutting plate is replaced.

[0052] This invention also provides a manual fine-compensation mode. Operators can view the current additional displacement value of each grid in real time through a gridded interface displayed on a touchscreen. When an accidental defect is detected or fine-tuning is desired, the operator can directly click on the corresponding grid, drag the slider on the pop-up control bar, or input a value to set any compensation amount within the range of 0-80μm. This setting is sent to the driver of the corresponding actuator via the EtherCAT bus and takes effect immediately in the next die-cutting cycle. The production line can run at full speed during this process without any downtime. It should be noted that when an operator sets a compensation value for a grid in manual mode, the control unit pauses the automatic compensation self-learning logic for that grid and outputs compensation based on the manually set value until the operator selects to resume automatic mode. After resuming automatic mode, the control unit restarts the historical compensation counting for that grid.

[0053] The key signal flow of the entire system is as follows: image data from the line scan camera flows into the image preprocessing subunit (FPGA) via the Cameralink interface. After the FPGA completes defect identification, it sends the defect coordinates to the main control subunit (industrial computer) via the EtherCAT bus. The spindle phase and voltage trigger signals acquired by the logic control subunit (PLC) are also periodically or event-triggered and uploaded to the main control subunit via the EtherCAT bus. After the main control subunit executes mapping queries and compensation decisions, it outputs control commands to the piezoelectric ceramic actuator driver through the analog output channel or the EtherCAT distributed I / O module. The driver drives the actuator to move according to the commands and displacement feedback in a closed loop. All communication is based on the EtherCAT real-time Ethernet protocol, ensuring that the delay from defect coordinate output to trigger compensation is controlled in the microsecond to millisecond range, fully meeting the real-time compensation requirements of a single cycle on a production line of 5000 sheets / hour.

[0054] In the above embodiments, the image sensor of the online detection module can be a linear CMOS camera, or a linear CCD camera or an area array camera combined with a global shutter, depending on the accuracy requirements. The light source can also be a laser light source or a high-brightness LED array. In addition to piezoelectric ceramics, the actuators in the distributed pressure regulation module can also be magnetostrictive actuators, miniature electromagnetic push rods, or miniature pneumatic / hydraulic bladders, etc., which are driving elements capable of generating micro-displacement. The grid division is not limited to a uniform 50mm square; it can be designed as a non-uniform grid size according to the density of the pattern on the die plate and the distribution of common wear areas, to provide finer resolution in easily worn areas. The sub-unit division of the control unit can have different physical carriers, such as integrating image preprocessing functions into a smart camera, or merging logic control and main control functions into a high-performance embedded controller. Any hardware architecture capable of real-time collaborative processing falls within the protection scope of this invention.

Claims

1. A local pressure online feedback compensation system for a die-cutting machine, characterized in that, include: The online detection module is coupled to the paper output side of the die-cutting machine and is configured to acquire images of a preset detection area of ​​the die-cut product after each die-cutting cycle, and identify the die-cutting defects and their position coordinates on the die-cut product based on the pre-stored reference image data. A distributed pressure adjustment module is installed in the upper or lower die base of the die-cutting machine. Its effective working plane is divided into multiple independent adjustment units corresponding to different areas of the die plate. Each independent adjustment unit is configured to selectively apply additional pressure to a corresponding local area on the die plate on top of a base pressure. The control unit is communicatively connected to the online detection module and the distributed pressure regulation module, respectively. The control unit is configured to: receive the position coordinates of the die-cutting defect output by the online detection module, determine the independent adjustment unit corresponding to the position coordinates according to a preset mapping relationship, and drive the determined independent adjustment unit to perform an additional pressure action in the next die-cutting cycle to actively compensate for the die-cutting defect.

2. The online feedback compensation system for local pressure of a die-cutting machine according to claim 1, characterized in that, The online detection module includes: A linear image sensor is mounted across the direction of paper movement and configured to acquire images of the preset detection area line by line. A strip light source illuminates the preset detection area at a preset low angle to the plane of the die-cut product. The image processing unit, coupled to the linear image sensor, is configured to compare the real-time acquired image grayscale signal with the reference image data point by point to identify the die-cutting defect and its position coordinates.

3. The online feedback compensation system for local pressure of a die-cutting machine according to claim 1, characterized in that, The distributed pressure regulation module includes: A base pressure plate, coupled to the main pressure mechanism of the die-cutting machine, is used to apply the base pressure to the die plate; Multiple piezoelectric ceramic actuators arranged in a matrix are fixed between the base pressure plate and the blade plate. Each piezoelectric ceramic actuator constitutes an independent adjustment unit and contacts a corresponding local grid area on the blade plate. The piezoelectric ceramic actuator driven by the control unit applies additional pressure by outputting an additional displacement to compress the elastic pad of the corresponding local area of ​​the blade.

4. The online feedback compensation system for local pressure of a die-cutting machine according to claim 3, characterized in that, The piezoelectric ceramic actuator incorporates a strain gauge displacement sensor, and the control unit is further configured to perform closed-loop control of the additional displacement based on feedback from the strain gauge displacement sensor; wherein, the size of the local grid region is configured to match the size of a typical wear area on the cutting plate, and the closed-loop control range of the additional displacement is configured to cover the amount of cutting line height loss caused by wear on the cutting plate.

5. The online feedback compensation system for local pressure of a die-cutting machine according to claim 1, characterized in that, The control unit includes an image preprocessing subunit, a logic control subunit, and a main control subunit, which are connected via a real-time industrial Ethernet bus. The logic control subunit is configured to receive the spindle phase signal of the die-cutting machine to identify the phase of the die-cutting cycle. The main control subunit is configured to, after receiving the defect coordinates from the online detection module, trigger the additional pressure action within the pressure-combining phase of the next die-cutting cycle using the phase signal provided by the logic control subunit, so as to meet the real-time requirement of the active pressure compensation being completed within a single die-cutting cycle.

6. The online feedback compensation system for local pressure of a die-cutting machine according to claim 1, characterized in that, The online detection module is further configured to identify the die-cutting defects in the following manner: The grayscale curve of the detection line acquired in real time is compared with the pre-stored reference grayscale curve by point-by-point difference calculation. When the difference at a certain point exceeds a preset grayscale threshold, and the length of consecutive pixels whose difference exceeds the grayscale threshold is greater than a preset length threshold, it is determined that the die-cutting defect exists at the position corresponding to that point.

7. The online feedback compensation system for local pressure of a die-cutting machine according to claim 1, characterized in that, The preset mapping relationship in the control unit is generated through the following calibration method: The control unit sequentially drives each of the independent adjustment units to generate a recognizable indentation on the test material without die-cutting pressure. The online detection module acquires the image coordinates of the indentation, and the control unit binds the identifier of each independent adjustment unit with the acquired image coordinates to generate a mapping table.

8. The online feedback compensation system for local pressure of a die-cutting machine according to claim 1, characterized in that, The control unit is also configured to perform a compensation amount self-learning step: when the additional pressure action is performed on the same independent adjustment unit for N consecutive die-cutting cycles, and the online detection module still identifies a die-cutting defect at the position corresponding to the independent adjustment unit, the additional pressure value for the independent adjustment unit is automatically increased, where N is an integer greater than or equal to 2.

9. A method for online feedback compensation of local pressure in a die-cutting machine, applied to an online feedback compensation system for local pressure in a die-cutting machine as described in any one of claims 1 to 8, characterized in that, include: Reference acquisition steps: Acquire and store reference image data of qualified die-cut products; Online inspection steps: After each die-cutting cycle, images of the preset inspection area of ​​the die-cut product are acquired in real time and compared with the reference image data to identify die-cutting defects and their location coordinates; Mapping compensation step: According to the preset mapping relationship, the position coordinates are converted into the identifier of the corresponding independent adjustment unit, and in the next die-cutting cycle, the determined independent adjustment unit is driven to perform additional pressure action to actively compensate for the die-cutting defects.

10. A method for online feedback compensation of local pressure in a die-cutting machine according to claim 9, characterized in that, The mapping compensation step further includes: Phase synchronization sub-step: Receive the phase signal of the die-cutting machine spindle and identify the phase of the current die-cutting cycle; Triggering sub-step: When the closing pressure phase of the next die-cutting cycle is detected, the additional pressure action is triggered, and the additional pressure action is maintained within the closing pressure phase.

Images