Gluing method and device for calibrating paper box by utilizing industrial vision three-dimensional vector and box gluing machine

Abstract

The invention relates to a method and device for calibrating carton gluing through industrial vision three-dimensional vectors and a carton gluing machine. The method comprises the steps that an industrial three-dimensional camera array is used for scanning a carton blank unfolding structure, elastic deformation compensation is conducted on scanned three-dimensional point clouds in combination with paper gram weight and fiber toughness parameters, and a three-dimensional vector coordinate system of carton blanks is constructed; generating a gluing path based on a matching result of the three-dimensional vector coordinate system and a preset model; driving the glue gun to execute a gluing path, dynamically adjusting the parameters of the glue gun according to the real-time stress distribution, and correcting the thickness of a glue layer and the specification of a nozzle in a linkage manner based on the carton region type; by monitoring glue solution permeation and excitation feedback of a three-dimensional vector field, the gluing quality is verified, and a glue supplementing path is generated and executed, so that the purposes of accurate execution of a three-dimensional path, real-time optimization of glue layer quality and self-adaptive efficient production of complex box types in the paper box gluing process are achieved.

Description

Technical Field

[0001] The present invention relates to the field of industrial automation and intelligent manufacturing technology, and in particular to a method and device for calibrating paper box gluing using industrial visual three-dimensional vectors, and a paper box gluing machine. Background Art

[0002] In the field of paper box packaging manufacturing, the gluing process is the core link that determines the structural strength and appearance quality of the product. The current mainstream technology mainly uses two-dimensional visual guidance or mechanical template positioning. Although it can complete basic bonding, it has significant defects when dealing with complex working conditions such as special-shaped boxes and high-weight paper. Most of the existing systems are based on rigid model assumptions and do not consider the elastic deformation characteristics of paper (such as fiber rebound and humidity deformation), which leads to the accumulation of glue seam position deviations; most of the existing glue gun parameters (speed, pressure) are preset to fixed values ​​and cannot be dynamically adjusted according to the real-time stress distribution, resulting in uneven glue layer thickness and local stress concentration; quality inspections mostly use offline sampling mode, which cannot feedback defects in real time during the production process and trigger compensation actions, and the rework rate remains high. With the increase in packaging precision requirements for consumption upgrades (such as the micron-level glue seam tolerance of luxury boxes), traditional technologies have been difficult to meet the needs of intelligent and flexible production. Summary of the invention

[0003] The main purpose of the present invention is to provide a method, device and gluer for paper box gluing using industrial vision three-dimensional vector calibration, so as to achieve the purpose of precise execution of three-dimensional paths in the paper box gluing process, real-time optimization of glue layer quality and adaptive and efficient production of complex box types through layered structural design and intelligent closed-loop control.

[0004] To achieve the above object, the present invention provides a method for calibrating carton gluing using industrial vision three-dimensional vectors, comprising the following steps: The unfolded structure of the carton blank is scanned by an industrial 3D camera array. The 3D point cloud scanned is elastically deformed and compensated for in combination with the paper weight and fiber toughness parameters to construct a 3D vector coordinate system for the carton blank. Generate a gluing path based on the matching result between the three-dimensional vector coordinate system and the preset model; Drive the glue gun to execute the gluing path, dynamically adjust the glue gun parameters according to the real-time stress distribution, and modify the glue layer thickness and nozzle specifications based on the carton area type; By monitoring the glue penetration and the excitation feedback of the three-dimensional vector field, the adhesive quality is verified, the glue filling path is generated and executed.

[0005] Furthermore, the steps of scanning the unfolded structure of the carton blank by an industrial three-dimensional camera array, performing elastic deformation compensation on the scanned three-dimensional point cloud in combination with the paper weight and fiber toughness parameters, and constructing a three-dimensional vector coordinate system of the carton blank include: The carton blank is scanned from multiple perspectives by an industrial 3D camera array to generate initial 3D point cloud data; Synchronously collect paper physical parameters, including paper weight parameters and fiber toughness parameters; Performing elastic deformation compensation on the initial three-dimensional point cloud data based on physical parameters of the paper; The compensated 3D point cloud data is registered with the preset carton model through the ICP algorithm to construct a 3D vector coordinate system.

[0006] Furthermore, the step of performing elastic deformation compensation on the initial three-dimensional point cloud data based on the physical parameters of the paper includes: Calculate the elastic recovery factor based on the fiber toughness parameter; A nonlinear coordinate correction is performed on the concave or warped area detected in the initial three-dimensional point cloud data, and the correction amount decays exponentially as the deformation amount increases.

[0007] Furthermore, the step of registering the compensated three-dimensional point cloud data with the preset paper box model by using the ICP algorithm includes: Adjust the iterative error weight according to the paper weight parameter. The larger the paper weight, the lower the weight distribution of the registration residual. The construction of the three-dimensional vector coordinate system is completed when the registration residual is less than 0.1 mm.

[0008] Furthermore, based on the matching result between the three-dimensional vector coordinate system and the preset model, the step of generating the gluing path includes: According to the crease characteristics in the three-dimensional vector coordinate system of the paper box blank, the gluing area is divided into a folding corner area, a plane area and a curved surface area; According to the adhesive division area, a spiral path, a bidirectional parallel path and a curvature adaptive path are generated.

[0009] Further, the step of driving the glue gun to execute the gluing path and dynamically adjusting the glue gun parameters according to the real-time stress distribution includes: When driving the glue gun to execute the gluing path, the stress distribution of the carton blank is predicted in real time through finite element analysis; When the stress in the corner area exceeds the threshold, the glue gun moving speed is reduced and the glue discharge pressure is increased. When the stress in the plane area exceeds the limit, the speed and pressure are reduced simultaneously. In the curved area, the density of the glue gun moving track is adjusted according to the curvature radius.

[0010] Further, the step of linking and correcting the glue layer thickness and the nozzle specification based on the carton area type includes: The micro eddy current sensor deployed in the folding area detects the thickness fluctuation of the cardboard in real time. When the thickness fluctuation value exceeds ±5% of the nominal thickness, the thickness of the glue layer is corrected by 20% of the thickness fluctuation value. The flat area uses an infrared thickness gauge to dynamically monitor the uniformity of the glue layer in real time. When the uniformity deviation exceeds 5%, it switches to a wide-width nozzle and increases the glue discharge pressure by 10%-15%; The curved surface area uses a laser displacement sensor to track the deformation of the cardboard surface, and the nozzle specifications and glue discharging parameters are adjusted in conjunction with the detection data. When the deformation is >0.3mm, the nozzle angle is adjusted until it coincides with the normal direction of the curved surface, and the glue discharging pressure is increased to 1.2 times the baseline value; The detection and correction operations of the corner area, the plane area and the curved surface area are performed according to the region priority order, and the priority order is corner area>curved surface area>plane area.

[0011] Furthermore, by monitoring the glue penetration and the excitation feedback of the three-dimensional vector field, the steps of verifying the adhesive quality, generating the glue filling path and executing the steps include: The penetration depth of the glue in the paper fiber is monitored in real time through the polarized light sensor, and the glue gun parameter adjustment is triggered when the penetration depth is lower than the preset value; Apply three-dimensional vector field excitation to the glue layer, collect the resonance frequency of the glue layer through the acceleration sensor, and calculate the frequency offset; If the frequency offset exceeds 5 Hz, a glue filling path is generated based on the paper fiber orientation.

[0012] The present invention also provides a paper box gluing device using industrial vision three-dimensional vector calibration, comprising: The scanning registration unit is used to scan the unfolded structure of the carton blank through an industrial 3D camera array, and to perform elastic deformation compensation on the scanned 3D point cloud in combination with the paper weight and fiber toughness parameters to construct a 3D vector coordinate system of the carton blank; A path generation unit, used for generating a gluing path based on a matching result between the three-dimensional vector coordinate system and a preset model; Dynamic execution unit, used to drive the glue gun to execute the gluing path, dynamically adjust the glue gun parameters according to the real-time stress distribution, and link the glue layer thickness and nozzle specifications based on the carton area type; The quality feedback unit is used to verify the gluing quality and generate the glue filling path by monitoring the glue penetration and the excitation feedback of the three-dimensional vector field.

[0013] The present invention also provides a box gluing machine, comprising: The conveying and positioning layer includes a conveyor belt with a three-dimensional vector coordinate system linkage positioning mark on the surface, which is used for directional conveying of carton blanks; The adhesive execution layer is equipped with a multi-degree-of-freedom robotic arm and a glue gun nozzle group, which is configured to execute a spiral path, a bidirectional parallel path, and a curvature adaptive path; The glue circulation layer integrates a recovery pipeline connected to the glue gun and a viscosity control unit to realize glue supply and residual material recovery; The control layer deploys the various units of the adhesive device to coordinate the operations of each layer and process the three-dimensional vector field feedback data.

[0014] The box gluing machine executes any one of the above-mentioned methods for calibrating box gluing using industrial vision three-dimensional vectors.

[0015] The paper box gluing method, device and paper box gluing machine using industrial vision three-dimensional vector calibration provided by the present invention have the following beneficial effects: the present invention breaks through the limitations of traditional rigid models and improves the glue seam alignment accuracy of complex structures by combining paper physical property modeling with three-dimensional vision calibration; autonomously adjusts gluing parameters based on real-time mechanical feedback to reduce material loss while ensuring bonding strength, taking into account both quality and cost; and constructs a closed-loop control link from defect detection to intelligent glue repair, thereby improving process stability and the adaptability of equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a schematic flow chart of a method for calibrating carton gluing by using industrial vision three-dimensional vectors in one embodiment of the present invention; Figure 2 It is a structural block diagram of a paper box gluing device calibrated by using industrial vision three-dimensional vector in one embodiment of the present invention; Figure 3 It is a schematic block diagram of the structure of a box gluing machine in one embodiment of the present invention.

[0017] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0019] Reference Figure 1 , which is a flow chart of a method for calibrating carton gluing using industrial vision three-dimensional vectors proposed by the present invention, comprising the following steps: S1, scan the unfolded structure of the carton blank through an industrial 3D camera array, combine the paper weight and fiber toughness parameters, perform elastic deformation compensation on the scanned 3D point cloud, and construct a 3D vector coordinate system for the carton blank; S2, generating a gluing path based on the matching result between the three-dimensional vector coordinate system and the preset model; S3, drives the glue gun to execute the gluing path, dynamically adjusts the glue gun parameters according to the real-time stress distribution, and links and corrects the glue layer thickness and nozzle specifications based on the carton area type; S4. Verify the adhesive quality by monitoring the penetration of the adhesive and the excitation feedback of the three-dimensional vector field, generate a path for supplementary adhesive application, and execute it. In one embodiment, for step S1, The step of scanning the unfolded structure of the carton blank through an industrial three-dimensional camera array, compensating for the elastic deformation of the scanned three-dimensional point cloud in combination with the paper grammage and fiber toughness parameters, and constructing a three-dimensional vector coordinate system of the carton blank includes: Performing multi-view scanning of the carton blank through an industrial three-dimensional camera array to generate initial three-dimensional point cloud data; Synchronously collecting paper physical parameters, including paper grammage parameters and fiber toughness parameters; Compensating for the elastic deformation of the initial three-dimensional point cloud data based on the paper physical parameters; Registering the compensated three-dimensional point cloud data with a preset carton model through the ICP algorithm to construct a three-dimensional vector coordinate system.

[0020] In a specific implementation, 6 groups of industrial structured light cameras distributed in a ring are used to perform multi-view scanning of the carton blank on the conveyor belt through a synchronous triggering mechanism. The overlap rate of the adjacent camera fields of view is ≥40%. Point cloud stitching is achieved through an improved SLAM algorithm (fusing IMU inertial data) to eliminate motion blur errors. Bilateral filtering and statistical outlier removal algorithms are used for the initial point cloud, and the filtering formula is: where, is the spatial weight, is the reflection intensity weight. It is measured in real time through a high-precision weighing sensor (accuracy ±0.1 g / m²) built into the conveyor belt and stored in association with the blank ID. Polarized light microscopy imaging (wavelength 532 nm) is used to detect the fiber arrangement density online, and the toughness coefficient is calculated through an empirical formula.

[0021] , where, (experimentally calibrated value, unit N / mm²), representing the ability of the material to resist permanent deformation. For the detected concave / warped area (curvature ), the correction amount Δ decays exponentially with the deformation amount : In the formula, controls the decay rate to ensure the suppression of overcorrection in large deformation areas. Gram weight adaptive weight distribution is performed. In the ICP iteration, a weight related to the gram weight (unit g / m²) is applied to each point cloud residual : , where, . Experiments show that when the gram weight increases by 100 g / m², the weight drops by 50%, giving priority to trusting the rigid characteristics of high-gram weight cardboard. When the mean registration residual or the maximum number of iterations The iteration is terminated. Experimental data shows that after compensation, the registration speed is increased by 40%, and the mean residual error is reduced from 0.35 mm to 0.07 mm (see Table 1): Table 1 Comparison of the elastic deformation compensation effect (N = 100 groups of samples):

[0022] Taking the geometric center of the blank unfolded structure as the origin, the crease direction is extracted by principal component analysis (PCA). The main crease direction is set as the X-axis, the direction perpendicular to the crease is set as the Y-axis, and the thickness direction is set as the Z-axis to construct a three-dimensional vector coordinate system to ensure that the coordinate system is consistent with the subsequent adhesive path planning direction. Step S1 solves the problem of glue seam misalignment caused by paper elasticity in traditional vision calibration through physical parameter-driven deformation compensation.

[0023] In one embodiment, the step of performing elastic deformation compensation on the initial three-dimensional point cloud data based on paper physical parameters includes: Calculating the elastic recovery factor according to the fiber toughness parameter; Performing non-linear coordinate correction on the detected concave or warped areas in the initial three-dimensional point cloud data, and the correction amount decays exponentially as the deformation amount increases.

[0024] Specifically, the ability of the paper to resist permanent deformation is quantified and determined by the fiber toughness parameter F (unit: N / mm²). The calculation formula of the elastic recovery factor , is the adjustment factor to control the steepness of the S-shaped curve (experimentally calibrated value); is the critical toughness value (determined by tensile test, when the material is mainly in elastic deformation). The trigger condition for non-linear coordinate correction is the detected point cloud curvature (warped / sunken area). The correction formula is: , is the deformation amount (the distance between the scanned point and the preset model, unit: mm); is the attenuation coefficient to prevent over-correction in large deformation areas (experimentally optimized value). The larger the deformation amount, the lower the correction weight (e.g., when d = 2 m, the attenuation factor ), avoiding overall deformation distortion caused by local mutations. By integrating the fiber toughness into the correction amount through KK, high-toughness paper (K≈1) is fully corrected, while low-toughness paper (K≈0) suppresses ineffective corrections. By integrating the fiber toughness into the correction amount through K, high-toughness paper (K≈1) is fully corrected, and low-toughness paper (K≈0) suppresses ineffective corrections. Due to the elastic deformation (such as depression and warping) of the paper caused by the fiber structure characteristics, there is a deviation between the three-dimensional scanned point cloud and the real geometric shape. Traditional methods only rely on geometric registration and ignore the physical properties of materials, resulting in glue path deviation. This solution calculates the elastic recovery factor through the fiber toughness parameter and designs a non-linear coordinate correction model to achieve precise deformation compensation.

[0025] In one embodiment, the step of registering the compensated three-dimensional point cloud data with the preset carton model through the ICP algorithm includes: Adjust the iteration error weight according to the paper grammage parameter, and the greater the grammage, the lower the weight allocation of the registration residual; When the registration residual is less than 0.1 mm, the three-dimensional vector coordinate system construction is completed.

[0026] Specifically, the grammage adaptive weight allocation formula , is the paper grammage (unit: g / m²), is the adjustment coefficient (optimized by the gradient descent method). The greater the grammage , the smaller the weight , that is, the influence of the residual in the high-grammage area on the overall registration objective function is reduced. When the average value of the registration residual Based on the requirements of the gluing process, the glue layer thickness tolerance is ±0.2 mm) or the maximum number of iterations (to avoid falling into local optimum) to terminate the iteration. Conduct registration effect verification, and conduct registration tests on 5 types of papers with different grammages (150 - 450 g / m²), with 20 repetitions in each group:

[0027] The grammage adaptive weight reduces the registration residual by 60% - 75% and shortens the time by 25% - 30%.

[0028] The traditional ICP algorithm assumes rigid body registration, but due to the difference in grammage, the deformation stiffness of the paper is different: high-grammage paper (such as 400 g / m²) has less deformation, while low-grammage paper (such as 150 g / m²) has significant deformation. This solution dynamically adjusts the ICP residual weight through the grammage parameter, gives priority to trusting the point cloud matching in the high-grammage area, and improves the registration efficiency and accuracy.

[0029] In one embodiment, for step S2, The step of generating the glue path based on the matching result of the three-dimensional vector coordinate system and the preset model includes: According to the crease characteristics in the three-dimensional vector coordinate system of the paper box blank, the gluing area is divided into a folding corner area, a plane area and a curved surface area; According to the adhesive division area, a spiral path, a bidirectional parallel path and a curvature adaptive path are generated.

[0030] In the specific implementation, the three-dimensional vector coordinate system of the carton blank is divided into three categories by using high-precision point cloud data (residual after compensation ≤ 0.1mm) and crease feature analysis, and using the curvature segmentation algorithm (based on principal component analysis to extract local curvature) combined with k-means clustering (fold angle classification number = 3): corner area, crease intersection area ( ), complex structure, high coverage adhesive is required; flat area, flat area ( ), regular surface is suitable for efficient glue coating; curved surface area, curvature change area ( ), which needs to dynamically adapt to deformation. On this basis, according to the structural characteristics and mechanical requirements of different regions, adaptive gluing paths are dynamically generated. The corner area adopts a spiral path, and its parameterized equation is: ,in , ( is the angle width), ensuring that the glue layer overlap rate is ≥30%. The plane area generates bidirectional parallel paths, and the path spacing is dynamically adjusted according to the viscosity of the glue (s=1.5+0.1η, unit mm), with constant glue gun speed (200 mm / s) and pressure control (P=0.2ηMPa), so that the glue layer thickness variance is ≤0.05mm. The curved area is dynamically controlled through the curvature adaptive path, and the glue gun speed is Adaptive adjustment with curvature radius R ( ), the nozzle angle matches the surface normal in real time, and detects the deformation through the laser displacement sensor (threshold 0.3mm), triggering path re-planning (response delay <50ms), thereby reducing the surface glue seam misalignment rate. After the generated path is verified by finite element analysis and stress distribution, it is converted into a robot arm G code instruction and executed by a multi-degree-of-freedom robot arm (positioning accuracy ±0.05mm). During the execution process, the system monitors the penetration depth of the glue in real time through a polarized light sensor (threshold ), and apply 10-100Hz three-dimensional vector field excitation to the glue layer, and detect the resonance frequency offset through the acceleration sensor (qualified criterion Δf≤5 Hz). If the detection is abnormal, the glue filling path is generated based on the direction of the paper fiber, forming a full closed-loop control of planning-execution-feedback.

[0031] In one embodiment, for step S3, The steps of driving the glue gun to execute the gluing path and dynamically adjusting the glue gun parameters according to the real-time stress distribution include: When driving the glue gun to execute the gluing path, the stress distribution of the carton blank is predicted in real time through finite element analysis; When the stress in the folding corner area exceeds the threshold, reduce the moving speed of the glue gun and increase the glue output pressure. When the stress in the flat area exceeds the limit, synchronously reduce the speed and pressure. For the curved surface area, adjust the density of the moving trajectory of the glue gun according to the radius of curvature.

[0032] In a specific implementation, a high-precision finite element model is constructed based on the three-dimensional vector coordinate system of the carton blank. Real-time stress analysis is continuously carried out at a period of 50 ms. The equivalent stress distribution in each area is calculated through a shell element mesh model (element size ≤ 2 mm). The key material parameters of the shell element mesh model are dynamically determined by the physical properties of the paper: elastic modulus E = 0.12W + 2.5F (W is the grammage, F is the fiber toughness), Poisson's ratio ν = 0.3 - 0.001W, and the stress field is updated every 50 ms (calculation time ≤ 10 ms), and the equivalent stress in the folding corner area, flat area, and curved surface area is output. . In the processing of the folding corner area, when it is detected that the equivalent stress exceeds the threshold of 2.0 MPa, dynamic adjustment is immediately started, and the moving speed of the glue gun decays according to the exponential relationship of ( is the initial speed), and at the same time, the glue output pressure is linearly increased according to ( is the initial pressure). By reducing the speed, the accumulation of glue volume per unit time is reduced, and at the same time, the pressure is increased to compensate for the glue layer thickness and relieve stress concentration. For the flat area, a more conservative adjustment strategy is adopted. When the stress exceeds 1.5 MPa, the speed and pressure are synchronously reduced to 70% and 80% of the reference value. By synchronously reducing the speed and pressure, the over-penetration of glue causing flat warping is avoided. And the parameter adjustment in the flat area adopts a step change to ensure the timeliness of the adjustment response. The processing of the curved surface area is more intelligent. The path point spacing d is dynamically adjusted according to the real-time detected radius of curvature R. When R ≤ 10 mm, d = 0.5R; when R > 10 mm, d = 0.3R + 2. This adaptive algorithm significantly improves the density of path points in the small curvature area and reduces the gap of the curved surface glue seam. At the same time, the nozzle angle of the curved surface area is adjusted in real time through the feedback of the laser displacement sensor to ensure that the deviation from the normal direction of the curved surface is ≤ 2°. The entire adjustment process is realized through a high-speed control bus. The piezoelectric ceramic valve accurately controls the glue output pressure at a response speed of 5 ms, and the manipulator motion control system updates the speed command at a frequency of 200 Hz. When it is detected that the stress exceeds the limit and lasts for more than 200 ms, an emergency gun-lifting mechanism is automatically triggered, and a new optimized path is generated within 30 ms.

[0033] In one embodiment, the steps of jointly correcting the glue layer thickness and nozzle specifications based on the carton area type include: The thickness fluctuation of the cardboard is detected in real time by the micro eddy current sensor deployed in the folding corner area. When the thickness fluctuation value exceeds ±5% of the nominal thickness, the glue layer thickness is corrected by 20% of the thickness fluctuation value; The flat area uses an infrared thickness gauge to dynamically monitor the uniformity of the glue layer in real time. When the uniformity deviation exceeds 5%, it switches to a wide-width nozzle and increases the glue discharge pressure by 10%-15%; The curved surface area uses a laser displacement sensor to track the deformation of the cardboard surface, and the nozzle specifications and glue discharging parameters are adjusted in conjunction with the detection data. When the deformation is >0.3mm, the nozzle angle is adjusted until it coincides with the normal direction of the curved surface, and the glue discharging pressure is increased to 1.2 times the baseline value; The detection and correction operations of the corner area, the plane area and the curved surface area are performed according to the region priority order, and the priority order is corner area>curved surface area>plane area.

[0034] In the specific implementation, differentiated detection methods and control strategies are adopted according to the structural characteristics of different areas of the paper box. In the folding corner area, a micro eddy current sensor is deployed. It has strong anti-electromagnetic interference ability and is suitable for deployment in narrow folding corner spaces. The accuracy is ±0.01mm and the thickness fluctuation of the paperboard is monitored in real time. When the thickness deviation is detected to exceed the nominal value ±5%, the glue layer thickness is automatically corrected by 20% of the deviation value. The correction formula is: , where is the nominal adhesive layer thickness (preset value), thickness fluctuation rate ,when When the glue layer thickness correction The flat area uses an infrared thickness gauge (sampling rate 500Hz) to dynamically monitor the uniformity of the glue layer. When the uniformity deviation is detected to exceed 5%, it will automatically switch to a wide nozzle (5mm) and increase the glue pressure by 10%-15%. The pressure adjustment formula is: , The reference glue pressure (unit: MPa or bar) is the initial pressure value preset according to the glue characteristics (such as viscosity, curing time) and process requirements. The curved area uses a laser displacement sensor (accuracy ±0.005mm) to track deformation in real time. When the deformation exceeds 0.3mm, the nozzle angle is adjusted synchronously to coincide with the normal direction of the curved surface (deviation ≤2°), and the glue pressure is increased to 1.2 times the reference value. A three-level priority scheduling mechanism (angle area>curved area>plane area) is adopted, and multi-task parallel processing is realized through FPGA, and the task scheduling cycle is ≤10ms.

[0035] In one embodiment, for step S4, By monitoring the penetration of glue and the excitation feedback of the three-dimensional vector field, the steps of verifying the adhesive quality, generating the glue filling path and executing it include: The penetration depth of the glue in the paper fiber is monitored in real time through the polarized light sensor, and the glue gun parameter adjustment is triggered when the penetration depth is lower than the preset value; Apply three-dimensional vector field excitation to the glue layer, collect the resonant frequency of the glue layer through the acceleration sensor, and calculate the frequency offset; If the frequency offset exceeds 5 Hz, a glue filling path is generated based on the paper fiber orientation.

[0036] In a specific implementation, the penetration of the glue liquid is monitored in real time by a high-precision polarized light sensor. The working principle is to irradiate the glue layer with polarized light of a wavelength of 632.8 nm, and analyze the change in the polarization state of the reflected light through the Mueller matrix model to calculate the penetration depth of the glue liquid in the paper fiber. (k = 0.82 experimental calibration coefficient, λ = 632.8 nm), where Δϕ is the polarization phase difference. When the detected penetration depth is lower than the preset threshold of 0.2 mm, the parameters of the glue gun are immediately adjusted automatically, and the glue output pressure is increased to 1.2 times the reference value ( ), and at the same time, the moving speed is reduced to 80% of the reference value ( ) to ensure sufficient penetration of the glue liquid.

[0037] A three-axis electromagnetic shaker is used to apply a swept-frequency excitation of 10 - 500 Hz to the glue layer. The excitation signal is: Among them, represents the amplitude (peak force) of the excitation signal, t is the time variable, is the time decay constant, represents the exponentially decaying envelope. The vibration response is collected by an MEMS acceleration sensor array with a spacing of 20 mm. The resonance frequency is extracted through FFT transformation and then the offset from the standard value is calculated ( is the reference frequency of the qualified glue layer). When Δf > 5 Hz, it is determined that there is a defect in the glue layer. The paper fiber texture is captured by a 5-megapixel polarized camera, and the main fiber direction θ is extracted by a Gabor filter bank with a wavelength of 50 μm and a direction resolution of 15°, and a fiber orientation vector field is constructed. Based on the above detection results, a glue filling path is intelligently generated, and a piecewise Bezier curve is planned along the main fiber direction. The control point spacing L is dynamically adjusted according to the penetration depth (L = 2D when D ≥ 0.15 mm, L = D / 0.1 when D < 0.15 mm). The density of the glue filling path is intelligently matched with the frequency offset. The path density is positively correlated with the frequency offset. For every 1 Hz increase in Δf, the path overlap rate is automatically increased by 5%.

[0038] Refer to Figure 2 , which is a structural block diagram of a carton gluing device using industrial vision three-dimensional vector calibration in an embodiment of the present invention, including: A scanning and registration unit for scanning the unfolded structure of the carton blank through an industrial three-dimensional camera array, compensating for the elastic deformation of the scanned three-dimensional point cloud in combination with the paper grammage and fiber toughness parameters, and constructing a three-dimensional vector coordinate system of the carton blank; A path generation unit, used for generating a gluing path based on a matching result between the three-dimensional vector coordinate system and a preset model; Dynamic execution unit, used to drive the glue gun to execute the gluing path, dynamically adjust the glue gun parameters according to the real-time stress distribution, and link the glue layer thickness and nozzle specifications based on the carton area type; The quality feedback unit is used to verify the gluing quality and generate the glue filling path by monitoring the glue penetration and the excitation feedback of the three-dimensional vector field.

[0039] For the specific implementation of each unit in the above device example, please refer to the above method embodiment, which will not be repeated here.

[0040] Reference Figure 3 , a schematic structural diagram of a box gluing machine proposed by the present invention, comprising: The conveying and positioning layer includes a conveyor belt with a three-dimensional vector coordinate system linkage positioning mark on the surface, which is used for directional conveying of carton blanks; The adhesive execution layer is equipped with a multi-degree-of-freedom robotic arm and a glue gun nozzle group, which is configured to execute a spiral path, a bidirectional parallel path, and a curvature adaptive path; The glue circulation layer integrates a recovery pipeline connected to the glue gun and a viscosity control unit to realize glue supply and residual material recovery; The control layer deploys the various units of the adhesive device to coordinate the operations of each layer and process the three-dimensional vector field feedback data.

[0041] The box gluing machine executes any one of the above-mentioned methods for calibrating box gluing using industrial vision three-dimensional vectors.

[0042] In summary, the present invention scans the unfolded structure of the carton blank through an industrial three-dimensional camera array, combines the paper weight and fiber toughness parameters, performs elastic deformation compensation on the scanned three-dimensional point cloud, and constructs a three-dimensional vector coordinate system of the carton blank; generates a gluing path based on the matching result of the three-dimensional vector coordinate system and the preset model; drives the glue gun to execute the gluing path, dynamically adjusts the glue gun parameters according to the real-time stress distribution, and linkage corrects the glue layer thickness and nozzle specifications based on the carton area type; verifies the gluing quality by monitoring the glue penetration and the excitation feedback of the three-dimensional vector field, generates and executes the glue filling path, so as to achieve the purpose of accurate execution of the three-dimensional path in the process of carton gluing, real-time optimization of the glue layer quality and adaptive and efficient production of complex box types.

[0043] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium provided by the present invention and used in the embodiments can include non-volatile and / or volatile memories. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM, etc.

[0044] It should be noted that in this article, the term "including", "comprising", or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that includes a series of elements includes not only those elements but also other elements not expressly listed, or elements that are inherent to such process, apparatus, article, or method. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of additional identical elements in the process, apparatus, article, or method that includes such element.

[0045] The above are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent structure or equivalent process transformation made by using the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present invention.

Claims

1. A method for calibrating carton gluing using industrial vision three-dimensional vectors, characterized in that: The following steps are involved: The unfolded structure of the carton blank is scanned by an industrial 3D camera array. The 3D point cloud scanned is elastically deformed and compensated for in combination with the paper weight and fiber toughness parameters to construct a 3D vector coordinate system for the carton blank. Generate a gluing path based on the matching result between the three-dimensional vector coordinate system and the preset model; Drive the glue gun to execute the gluing path, dynamically adjust the glue gun parameters according to the real-time stress distribution, and modify the glue layer thickness and nozzle specifications based on the carton area type; By monitoring the glue penetration and the excitation feedback of the three-dimensional vector field, the adhesive quality is verified, the glue filling path is generated and executed.

2. The method for calibrating carton gluing using industrial vision three-dimensional vector according to claim 1 is characterized in that: The step of scanning the unfolded structure of the carton blank by an industrial three-dimensional camera array, performing elastic deformation compensation on the scanned three-dimensional point cloud in combination with the paper weight and fiber toughness parameters, and constructing a three-dimensional vector coordinate system of the carton blank includes: The carton blank is scanned from multiple perspectives by an industrial 3D camera array to generate initial 3D point cloud data; Synchronously collect paper physical parameters, including paper weight parameters and fiber toughness parameters; Performing elastic deformation compensation on the initial three-dimensional point cloud data based on physical parameters of the paper; The compensated 3D point cloud data is registered with the preset carton model through the ICP algorithm to construct a 3D vector coordinate system.

3. The method for calibrating carton gluing using industrial vision three-dimensional vector according to claim 2 is characterized in that: The step of performing elastic deformation compensation on the initial three-dimensional point cloud data based on the physical parameters of the paper includes: Calculate the elastic recovery factor based on the fiber toughness parameter; A nonlinear coordinate correction is performed on the concave or warped area detected in the initial three-dimensional point cloud data, and the correction amount decays exponentially as the deformation amount increases.

4. The method for calibrating carton gluing using industrial vision three-dimensional vector according to claim 2 is characterized in that: The step of registering the compensated three-dimensional point cloud data with the preset paper box model by using the ICP algorithm includes: Adjust the iterative error weight according to the paper weight parameter. The larger the paper weight, the lower the weight distribution of the registration residual. The construction of the three-dimensional vector coordinate system is completed when the registration residual is less than 0.1 mm.

5. The method for calibrating carton gluing using industrial vision three-dimensional vector according to claim 1 is characterized in that: The step of generating a gluing path based on the matching result between the three-dimensional vector coordinate system and the preset model comprises: According to the crease characteristics in the three-dimensional vector coordinate system of the paper box blank, the gluing area is divided into a folding corner area, a plane area and a curved surface area; According to the adhesive division area, a spiral path, a bidirectional parallel path and a curvature adaptive path are generated.

6. The method for calibrating carton gluing using industrial vision three-dimensional vector according to claim 1 is characterized in that: The step of driving the glue gun to execute the gluing path and dynamically adjusting the glue gun parameters according to the real-time stress distribution includes: When driving the glue gun to execute the gluing path, the stress distribution of the carton blank is predicted in real time through finite element analysis; When the stress in the corner area exceeds the threshold, the glue gun moving speed is reduced and the glue discharge pressure is increased. When the stress in the plane area exceeds the limit, the speed and pressure are reduced simultaneously. In the curved area, the density of the glue gun moving track is adjusted according to the curvature radius.

7. The method for calibrating carton gluing using industrial vision three-dimensional vector according to claim 1 is characterized in that: The step of linking and correcting the glue layer thickness and the nozzle specification based on the carton area type includes: The micro eddy current sensor deployed in the folding area detects the thickness fluctuation of the cardboard in real time. When the thickness fluctuation value exceeds ±5% of the nominal thickness, the thickness of the glue layer is corrected by 20% of the thickness fluctuation value. The flat area uses an infrared thickness gauge to dynamically monitor the uniformity of the glue layer in real time. When the uniformity deviation exceeds 5%, it switches to a wide-width nozzle and increases the glue discharge pressure by 10%-15%; The curved surface area uses a laser displacement sensor to track the deformation of the cardboard surface, and the nozzle specifications and glue discharging parameters are adjusted in conjunction with the detection data. When the deformation is >0.3mm, the nozzle angle is adjusted until it coincides with the normal direction of the curved surface, and the glue discharging pressure is increased to 1.2 times the baseline value; The detection and correction operations of the corner area, the plane area and the curved surface area are performed according to the region priority order, and the priority order is corner area>curved surface area>plane area.

8. The method for calibrating carton gluing using industrial vision three-dimensional vector according to claim 1 is characterized in that: The step of monitoring the glue penetration and the excitation feedback of the three-dimensional vector field to verify the adhesive quality, generate the glue filling path and execute it includes: The penetration depth of the glue in the paper fiber is monitored in real time through the polarized light sensor, and the glue gun parameter adjustment is triggered when the penetration depth is lower than the preset value; Apply three-dimensional vector field excitation to the glue layer, collect the resonant frequency of the glue layer through the acceleration sensor, and calculate the frequency offset; If the frequency offset exceeds 5 Hz, a glue filling path is generated based on the paper fiber orientation.

9. A paper box gluing device using industrial vision three-dimensional vector calibration, characterized in that: include: The scanning registration unit is used to scan the unfolded structure of the carton blank through an industrial 3D camera array, and to perform elastic deformation compensation on the scanned 3D point cloud in combination with the paper weight and fiber toughness parameters to construct a 3D vector coordinate system of the carton blank; A path generation unit, used for generating a gluing path based on a matching result between the three-dimensional vector coordinate system and a preset model; Dynamic execution unit, used to drive the glue gun to execute the gluing path, dynamically adjust the glue gun parameters according to the real-time stress distribution, and link the glue layer thickness and nozzle specifications based on the carton area type; The quality feedback unit is used to verify the gluing quality and generate the glue filling path by monitoring the glue penetration and the excitation feedback of the three-dimensional vector field.

10. A box gluing machine, characterized in that: include: The conveying and positioning layer includes a conveyor belt with a three-dimensional vector coordinate system linkage positioning mark on the surface, which is used for directional conveying of carton blanks; The adhesive execution layer is equipped with a multi-degree-of-freedom robotic arm and a glue gun nozzle group, which is configured to execute a spiral path, a bidirectional parallel path, and a curvature adaptive path; The glue circulation layer integrates a recovery pipeline connected to the glue gun and a viscosity control unit to realize glue supply and residual material recovery; The control layer deploys the various units of the adhesive device to coordinate the operations of each layer and process the three-dimensional vector field feedback data; The box gluing machine implements the method for calibrating box gluing using industrial vision three-dimensional vectors as described in any one of claims 1-8.

Images