A silk positioning and printing closed-loop compensation method and compensation system based on spectral feature point array and tension decoupling control

Abstract

This invention relates to the field of digital printing technology for textile fabrics, specifically to a closed-loop compensation method and system for silk positioning printing based on spectral feature dot matrix and tension decoupling control. A spectral feature dot matrix is ​​deployed on the surface of the silk fabric. This dot matrix consists of at least two types of functional inks, including infrared absorbing ink and upconversion fluorescent ink. The infrared absorbing ink provides a planar positioning reference, while the upconversion fluorescent ink is used in conjunction with structured light to measure the Z-axis height of the fabric surface. This invention uses functional inks such as infrared absorbing ink, upconversion fluorescent ink, and thermochromic ink to actively deploy a spectral feature dot matrix on the surface of the silk fabric. This dot matrix is ​​transparent under visible light, without interfering with the aesthetics of the designed pattern; it can be accurately identified under infrared and specific excitation light, providing a high-density positioning reference that is not limited by the inherent characteristics of the fabric for deformation measurement.

Description

Technical Field

[0001] This invention relates to the field of digital printing technology for textile fabrics, specifically to a closed-loop compensation method and system for silk positioning printing based on spectral feature matrix and tension decoupling control. Background Technology

[0002] Digital printing technology, with its advantages of short production cycle, rich colors, and environmental friendliness and energy saving, has been widely used in the textile printing and dyeing industry. Silk, as a high-end textile fabric, is widely used in the production of cheongsams, scarves, and high-end clothing due to its soft luster, smooth feel, and good breathability. However, silk fabric is composed of protein fibers with an extremely low elastic modulus (approximately 1 / 3 to 1 / 2 that of cotton fibers) and significant anisotropy. During the digital printing process, it is highly susceptible to nonlinear elastic deformation due to factors such as tension fluctuations and belt transmission, leading to positioning deviations in the printed pattern and seriously affecting product quality.

[0003] Currently, existing technologies mainly address the printing positioning deviation problem caused by textile fabric deformation by focusing on the following aspects: I. Mechanical weft straightening technology Traditionally, weft straightening is performed on the fabric before printing. This involves using a weft straightening mechanism to adjust the relative speeds of the warp yarns, causing the corresponding sections of the weft yarns that are bent to be "ahead" or "lag behind," thus ensuring that the warp and weft threads are as perpendicular as possible. However, as is well known in the art, after weft straightening, the fabric still needs to undergo processes such as sizing, drying, and laying it flat on a platform. These processes can easily lead to fabric deformation again. Therefore, relying solely on pre-printing weft straightening cannot completely eliminate the problem of warping and distortion in the printed pattern.

[0004] II. Deformation Compensation Technology Based on Image Recognition In recent years, with the development of machine vision technology, deformation compensation methods based on image recognition have emerged. For example, Chinese patent CN114905864A (application number 202210418612.7, patentee Hangzhou Honghua Digital Technology Co., Ltd.) discloses an "Adaptive Precision Positioning Printing Method and Device for Textile Fabric Deformation". This technical solution acquires fabric images, extracts characteristic yarns of the fabric (specifically, weft yarn curves), calculates fabric deformation parameters, and then performs image interpolation deformation processing on the original print image to match the fabric deformation state before printing. This technical solution characterizes fabric deformation by detecting the curvature of the weft yarns and pre-deforms the print image based on this, thus overcoming the problem of printed pattern distortion caused by fabric deformation to a certain extent.

[0005] However, this technical solution has the following limitations: It relies on the inherent yarn texture of the fabric as the deformation detection benchmark. For fabrics with smooth surfaces and indistinct yarn textures, such as silk, it is difficult to accurately extract the characteristic yarns. It only detects the two-dimensional planar deformation of the fabric (i.e., the bending of the weft yarns) and fails to detect the warping deformation of the fabric in the thickness direction (Z-axis). In fact, silk fabrics are prone to local arching during the conveyor belt process due to uneven tension distribution, and the warping height can reach 0.5mm-1.2mm. This Z-axis deformation is completely invisible in traditional two-dimensional visual alignment, but it will directly cause changes in the distance between the printhead and the fabric, resulting in micron-level misregistration. It only uses image interpolation deformation as a single compensation method, without actively suppressing deformation at the mechanical level, and it is an open-loop control, so it cannot provide feedback correction based on the actual effect after printing.

[0006] III. Overprint Compensation Technique Based on Feature Point Matching Another existing technology, Chinese patent CN107901622B (application number 201711248314.3, patentee Hangzhou Honghua Digital Technology Co., Ltd.), discloses a "method and fabric for precise digital printing on fabric using a digital printing machine". This technical solution uses a floral pattern on the fabric as the original template, selects feature points, and deforms and adjusts the target pattern template to achieve precise overprinting of the secondary design with the original pattern. This technology is mainly applicable to secondary design scenarios for fabrics with existing patterns such as lace, jacquard, and embroidery. By matching feature points between the original template and the fabric to be printed, the deformation and alignment of the target pattern are achieved.

[0007] However, this technical solution also has the following technical limitations: It relies on the inherent pattern of the fabric as a positioning reference. For blank fabrics (such as natural silk) or fabrics with indistinct pattern features, it is impossible to extract effective feature points. It also only involves deformation compensation in a two-dimensional plane and does not consider the effect of fabric Z-axis warping; Its compensation method is also a single image deformation, without mechanical tension control, and it is an open-loop control architecture.

[0008] In summary, while existing technologies have made some progress in deformation compensation for textile fabrics, the following technological gaps still exist for digital positioning printing on silk, a special substrate (protein fiber, extremely low elastic modulus, and significant anisotropy): Reliability issues of deformation reference: The surface of silk is smooth and the inherent yarn texture or pattern features are not obvious. It is difficult to implement the method of relying on the inherent features of the fabric as the positioning reference on silk. In addition, traditional physical targets (such as crosshairs) will deform synchronously with the fabric, resulting in "it looks like it is aligned, but it is not actually aligned". The problem of missing perception of three-dimensional deformation: Existing technology only focuses on two-dimensional planar deformation and fails to perceive and compensate for the offset of the landing point caused by the Z-axis warping of the fabric. This error can reach more than 0.5mm, far exceeding the micron-level overprinting accuracy requirements. The problem of the single compensation method: the existing technology only uses image pre-deformation as a compensation method, without actively suppressing deformation at the mechanical level, and lacks the division of labor and cooperation of "physical coarse adjustment + digital fine adjustment"; Open-loop problem of control architecture: Existing technologies are all open-loop control, lacking a real-time feedback correction mechanism, and cannot form an adaptive closed loop of "perception-decision-execution-re-perception".

[0009] Especially for specific scenarios such as matching patterns on cheongsam collars, positioning silk scarves at the corners, and seamlessly integrating gold thread overlocking with digital printing, higher requirements are placed on the accuracy of registration. In traditional solutions, the limit of registration accuracy for multi-color groups (including spot colors, white ink, and varnish) is about ±0.1mm. However, for high-end silk products, customers often require clear edges, no white gaps, and no color deviation when observed under a magnifying glass, which approaches or even exceeds the limits of traditional solutions.

[0010] Therefore, how to establish a positioning printing compensation mechanism on silk, a low-modulus, anisotropic substrate, that can simultaneously sense planar deformation and Z-axis warping, achieve coordinated mechanical and digital compensation, and provide closed-loop feedback throughout the entire process, has become a technical challenge that urgently needs to be solved in this field. Summary of the Invention

[0011] The present invention aims to overcome at least some of the defects of the prior art and provide a positioning printing compensation method and system for silk fabrics that integrates "active layout of spectral feature dot matrix - multispectral three-dimensional perception - multi-zone tension decoupling control - inkjet pixel-level reconstruction - full-process closed-loop feedback" to solve the problem of micron-level overprinting deviation caused by nonlinear elastic deformation and Z-axis warping in digital positioning printing of silk.

[0012] To achieve the above objectives, the present invention provides the following technical solution: A closed-loop compensation method for silk positioning printing based on spectral feature lattice and tension decoupling control includes the following steps: Step S1: Marking and Layout Steps A spectral feature dot matrix is ​​arranged on the surface of a silk fabric. The spectral feature dot matrix is ​​composed of at least two functional inks, including an infrared absorbing ink and an upconversion fluorescent ink. The infrared absorbing ink is used to provide a planar positioning reference, and the upconversion fluorescent ink is used to measure the Z-axis height of the fabric surface in conjunction with structured light.

[0013] Preferably, the spectral feature matrix further includes a thermosensitive color-changing ink, used to trace the thermal deformation history of the fabric in the subsequent color-fixing process.

[0014] Preferably, the spectral feature dot matrix adopts a non-uniform dot distribution strategy: the dot density in the fabric edge region, arc region and corner region is higher than the dot density in the flat region of the fabric, wherein the dot spacing in the edge region, arc region and corner region is 5mm-10mm, and the dot spacing in the flat region is 15mm-25mm.

[0015] This invention introduces for the first time an actively implanted spectral feature dot matrix, solving the technical problem of the lack of deformation reference on the silk surface. Unlike CN114905864A and CN107901622B, which rely on the inherent yarn or pattern features of the fabric, this invention uses functional inks such as infrared absorption ink, upconversion fluorescent ink, and thermosensitive color-changing ink to actively deploy a spectral feature dot matrix on the silk fabric surface. This dot matrix is ​​transparent under visible light, without interfering with the aesthetics of the design pattern; it can be accurately identified under infrared and specific excitation light, providing a high-density positioning reference for deformation measurement that is not limited by the inherent characteristics of the fabric.

[0016] More importantly, the combined use of the three functional inks produced a synergistic effect: the infrared absorbing ink provided a planar positioning reference, the upconversion fluorescent ink, in conjunction with structured light, enabled Z-axis height measurement, and the thermochromic ink recorded the thermal deformation history of the subsequent color fixing process—a single marker point simultaneously carries planar position, height information, and thermal history, achieving a "one code, multiple uses" information density and providing a data foundation for subsequent multi-physics decoupling. The non-uniform dot distribution strategy further densified the dot distribution in areas with large deformation gradients, improving the spatial resolution of the deformation field function (up to 1mm×1mm), creating conditions for subsequent compensation.

[0017] Step S2: Multispectral Sensing Step The three-dimensional spatial coordinates of the spectral feature array are captured in real time by a multispectral imaging system, which includes an infrared camera and a structured light projection module. A nonlinear deformation field function of the fabric surface is constructed based on the captured coordinates. The nonlinear deformation field function describes both the planar deformation and Z-axis warping of the fabric.

[0018] Preferably, the nonlinear deformation field function is constructed using a thin plate spline interpolation algorithm. The measured coordinates of each captured marker point are compared with the theoretical coordinates in the pre-press file, and the continuous deformation field function of the entire fabric surface is obtained through interpolation calculation.

[0019] This invention marks the first time that online measurement and compensation of fabric Z-axis warpage has been achieved in the field of digital printing, filling a technological gap in this area. Traditional 2D visual alignment completely ignores Z-axis deformation, and the industry has long believed that as long as the pressure rollers are sufficient, Z-axis errors can be ignored. However, this invention, through the combination of a structured light projection module and upconversion fluorescent ink, uses the triangulation principle to achieve online measurement of the Z-axis height of the fabric surface. Actual measurements revealed that silk fabrics are prone to localized arching during conveyor belt transport due to uneven tension distribution, with warpage heights reaching 0.5mm-1.2mm, resulting in a landing point offset of over 0.3mm—a finding that was itself unexpected.

[0020] This step works synergistically with step S1: the infrared marker provides the X and Y coordinates, the upconversion fluorescent ink + structured light provides the Z-axis height, and the multispectral camera (infrared + visible light) and functional ink achieve "spectral matching," making the invisible marker clearly identifiable in specific wavelengths without affecting the aesthetics of the pattern, while ensuring recognition accuracy. The nonlinear deformation field function constructed through the thin-plate spline interpolation algorithm simultaneously describes planar deformation and Z-axis warping, achieving deep fusion of "one code, multiple uses" marker information and multispectral sensing, laying the foundation for subsequent layered compensation.

[0021] To facilitate subsequent layered control, the nonlinear deformation field function Δ(x,y,z) includes a planar deformation component Δ_xy(x,y) and a Z-axis warping component Δ_z(x,y). Specifically, the tension decoupling control in step S3 performs macroscopic deformation suppression based on the complete Δ(x,y,z) function, while the inkjet compensation in step S4 performs micrometer-level compensation based on the residual planar deformation field Δ_res(x,y) remaining after suppression in step S3.

[0022] Step S3: Tension decoupling control step Based on the nonlinear deformation field function, the multi-zone independently controlled tension roller group set on the guide belt is decoupled and controlled. By adjusting the output torque of each tension zone, the macroscopic nonlinear deformation of the fabric is suppressed, so that the fabric deformation is reduced to below the preset threshold.

[0023] Preferably, the multi-zone independent control tension roller group is divided into 8-16 independent control zones along the width direction of the guide belt, and each control zone is equipped with an independent servo motor and force sensor; the decoupling control adopts a model predictive control algorithm, which calculates the target tension distribution in reverse according to the nonlinear deformation field function, and independently adjusts the output torque of each control zone.

[0024] This invention is the first to introduce multi-zone tension decoupling control into the field of digital silk printing, achieving a leap from "passive adaptation" to "active suppression." Traditional solutions only compensate for deformation through image distortion, failing to actively suppress the deformation source at the physical level. This invention, through 8-16 zone independent control of tension roller groups and a model predictive control algorithm, calculates the target tension distribution inversely based on the deformation field function, suppressing the macroscopic nonlinear deformation of silk fabric to below 0.2%.

[0025] This step, together with step S2, creates a synergistic effect: multispectral sensing acquires the deformation field, and tension decoupling control actively adjusts the tension distribution based on the deformation field, forming a preliminary closed loop of "sensing-decision-execution". More importantly, this step, together with the subsequent step S4, forms a two-layer compensation architecture of "coarse adjustment + fine adjustment". Tension control is responsible for suppressing "large deformation" to the range of "micro deformation" (target <0.2%), with a response speed of 50ms and a macroscopic range of action (cm level); inkjet compensation is responsible for correcting the remaining micron-level residual deformation, with a response speed of microseconds and a microscopic range of action (μm level). The two complement each other in the temporal and spatial domains, creating conditions for the final breakthrough in accuracy.

[0026] Step S4: Inkjet compensation step Based on the residual deformation field suppressed in step S3, the image data to be printed is subjected to reverse pre-deformation processing, and the ignition pulse parameters of the printhead are dynamically reconstructed so that the printhead reduces the ignition frequency in the stretching region and increases the ignition frequency in the shrinking region to compensate for the micron-level residual deformation.

[0027] Preferably, the reverse pre-deformation processing specifically involves: assuming the original design image is I(u,v) and the residual deformation field function on the current fabric surface is Δ(x,y), then the pixel value P(x,y) ejected by the nozzle at the physical coordinates (x,y) satisfies P(x,y)=I(u,v), where (u,v)=(x,y)+Δreverse(x,y), and Δreverse is a compensation field opposite to the direction of the measured deformation field.

[0028] Preferably, when dynamically reconstructing the ignition pulse parameters of the printhead, an ink volume conservation constraint is simultaneously executed: ensuring that the total mass of dye per unit area remains constant before and after compensation, which is achieved by adjusting the ink droplet volume or the number of ink droplets, in order to avoid color deviation caused by deformation compensation.

[0029] This invention constructs a dual-domain compensation mechanism of "image domain + ink droplet domain", achieving a dual-objective synergy of deformation compensation and color consistency. Traditional solutions only compensate through image interpolation deformation, without considering changes in ink droplet distribution, resulting in a decrease in ink volume per unit area in stretched areas (lighter color) and an increase in ink volume in contracted areas (darker color), producing color difference.

[0030] The innovation of this invention lies in three aspects: First, adjusting the macroscopic pattern position through reverse pre-deformation of the image; second, adjusting the microscopic ink droplet distribution through ignition pulse reconstruction, reducing the ignition frequency and increasing the ink droplet spacing in the stretched area, and increasing the ignition frequency and compressing the ink droplet spacing in the contracted area; third, adjusting the ink droplet volume in real time through ink volume conservation constraints to ensure that the total mass of dye per unit area remains constant before and after compensation. These three aspects work together to achieve the dual goals of "deformation compensation and color consistency," resolving the contradiction that cannot be achieved by a single method.

[0031] This step, together with step S3, forms a collaborative framework of "coarse adjustment + fine adjustment": tension adjustment suppresses macroscopic deformation to below 0.2%, creating a "comfort zone" for inkjet compensation, so that inkjet compensation does not need to deal with large deformation and avoids image distortion caused by over-correction; inkjet compensation corrects micron-level residual deformation that tension adjustment cannot eliminate. The two work together to enable the final registration accuracy to break through the physical limits of the equipment (reaching ±0.03mm, which is better than the traditional ±0.1mm).

[0032] Step S5: Closed-loop feedback step The fabric printed in step S4 is then subjected to the multispectral sensing system in step S2 to obtain the actual placement deviation, and the deviation is fed back to steps S3 and S4 to form a closed-loop control for the entire process.

[0033] This invention establishes a closed-loop feedback mechanism throughout the entire process, achieving adaptive control. Traditional solutions are all open-loop controls, which cannot be verified after a single compensation and cannot cope with batch-to-batch differences and cumulative errors.

[0034] The innovations of this invention are as follows: First, through a dual-camera layout (upstream and downstream), the upstream camera captures input deformation, while the downstream camera provides feedback on the actual placement deviation after printing, forming a real-time closed loop of "perception-decision-execution-re-perception" with a control cycle of 50ms. Second, by using thermochromic ink to record the thermal deformation history of the subsequent color-fixing process, a cross-process closed-loop feedback is achieved, feeding downstream data forward to the upstream compensation model, continuously optimizing the accuracy of subsequent batches. This dual feedback architecture of intra-process closed loop + inter-process closed loop enables the system to have adaptive evolution capabilities, automatically correcting batch-to-batch differences and cumulative errors without manual intervention.

[0035] This invention also provides a closed-loop compensation system for silk positioning printing based on spectral feature lattice and tension decoupling control, for implementing the above method, including: A marking unit is used to lay out a spectral feature dot matrix on the surface of silk fabric. The marking unit includes at least one set of digital inkjet printheads or roller coating units for printing infrared absorbing ink and upconversion fluorescent ink. A multispectral sensing unit includes at least two sets of multispectral line scan cameras respectively disposed upstream and downstream of the nozzle along the transmission direction of the guide band, and a set of structured light projection modules. The multispectral line scan cameras are equipped with dual-channel filters for infrared and visible light bands to capture the three-dimensional spatial coordinates of the spectral feature array in real time. A force decoupling control unit includes a tension roller group consisting of multiple independent control zones arranged along the width direction of the guide belt, and a model prediction controller connecting each control zone, used to adjust the output torque of each control zone according to the nonlinear deformation field function; An inkjet compensation unit includes an image pre-deformation module and a printhead ignition pulse reconstruction module. The image pre-deformation module is used to perform reverse pre-deformation processing on the image data to be printed, and the printhead ignition pulse reconstruction module is used to dynamically adjust the ignition frequency and waveform parameters of the printhead. A central controller is connected to the multispectral sensing unit, the tension decoupling control unit, and the inkjet compensation unit respectively. It is used to receive data from the sensing unit, calculate the nonlinear deformation field function, and coordinate the control of the tension decoupling control unit and the inkjet compensation unit.

[0036] Preferably, the multispectral sensing unit further includes a temperature-sensitive imaging module disposed downstream of the printhead. The temperature-sensitive imaging module is used to capture the color change of the temperature-sensitive color-changing ink after color fixation and feed the change data back to the central controller for correcting the deformation field function model of subsequent batches.

[0037] Preferably, the inkjet compensation unit further includes an ink volume conservation controller, which is linked with the printhead ignition pulse reconstruction module to calculate the dye mass per unit area in real time based on the deformation compensation amount and adjust the ink droplet volume parameters to maintain a constant total dye mass.

[0038] The closed-loop compensation system provided by this invention solidifies the technical features of the above methods into hardware modules, forming a complete industrial solution. The synergistic effect between the modules is reflected in: First, the spectral matching between the marking unit and the multispectral sensing unit: the functional ink printed by the marking unit is precisely matched with the filter band of the multispectral sensing unit, so that the invisible mark is clearly distinguishable under specific bands, achieving a perfect balance between transparency under visible light and high contrast during detection.

[0039] Second, the multispectral sensing unit, tension decoupling control unit, and inkjet compensation unit form a closed-loop linkage: upstream sensing data drives tension adjustment and inkjet compensation, while downstream sensing data provides feedback to correct control parameters, forming a closed loop throughout the entire process.

[0040] Third, cross-process feedback between the temperature-sensitive imaging module and the central controller: thermal deformation data from the subsequent color-fixing process is fed forward to the preceding compensation model, enabling the system to have cross-process adaptive capabilities.

[0041] Fourth, the linkage between the ink volume conservation controller and the ignition pulse reconstruction module: while adjusting the ink droplet spacing, the ink droplet volume is adjusted in real time to ensure that the total mass of dye per unit area remains constant before and after compensation, thus achieving the dual objectives of deformation compensation and color consistency.

[0042] Compared with the prior art, the present invention has the following beneficial effects: This invention constructs a dual-layer compensation architecture of "coarse adjustment + fine adjustment" through "multi-zone tension decoupling control" and "inkjet pixel-level reconstruction": tension control suppresses macroscopic deformation to below 0.2%, and inkjet compensation corrects micron-level residual errors. The two complement each other in the temporal domain and coordinate in the spatial domain, so that the registration accuracy reaches ±0.03mm, which exceeds the traditional physical limit of ±0.1mm.

[0043] The edge roughness of the compensated image obtained by this invention is better than the theoretical value without deformation. The high-density spectral feature lattice enables a spatial resolution of 1mm×1mm for the deformation field. Anti-aliasing optimization is naturally generated during the reverse pre-deformation process, and the edge roughness Ra is 8μm, which is better than the theoretical value of 25μm without deformation, achieving an unexpected effect of "gain compensation".

[0044] The shrinkage consistency of this invention is significantly improved. Multi-zone tension decoupling control pre-stretches and shapes the silk while suppressing deformation, releasing internal stress in advance, thus improving the shrinkage consistency of subsequent washing from ±3% to ±0.5%, generating synergistic benefits across processes.

[0045] This invention, through the combination of structured light and upconversion fluorescent ink, achieves for the first time online measurement and compensation of Z-axis warpage (0.5-1.2mm) of silk, filling a technological gap in this field. Attached Figure Description

[0046] Figure 1 This is a schematic flowchart of the closed-loop compensation method in an embodiment of the present invention; Figure 2 This is a schematic diagram of the non-uniform distribution strategy of the spectral feature matrix in an embodiment of the present invention; Figure 3 This is a schematic diagram of the inkjet compensation principle in an embodiment of the present invention; Detailed Implementation The following is in conjunction with the appendix Figure 1-3 The present invention will be further described in detail below with specific embodiments. The following embodiments are for illustrative purposes only and do not constitute any limitation on the invention. Those skilled in the art, based on their understanding of the technical solutions of the present invention, may make equivalent substitutions or modifications to some of the technical features, all of which should fall within the protection scope of the present invention. Example 1 This embodiment provides a closed-loop compensation method for silk positioning printing based on spectral feature lattice and tension decoupling control, specifically applied to the scene of matching patterns on the collar of a silk cheongsam. For example... Figure 1 As shown, the method mainly includes five steps: marker placement, multispectral sensing, tension decoupling control, inkjet compensation, and closed-loop feedback.

[0047] 1. Marking and placement The silk cheongsam fabric pieces to be printed are made of 12 momme plain crepe satin, 1.2m wide, with an arc-shaped neckline and a left-right symmetry accuracy requirement of ±0.05mm. Before entering the digital printing process, spectral feature dot matrix is ​​laid out on the surface of the fabric pieces using marking units.

[0048] The marking unit uses a digital inkjet printhead to print three functional inks onto the silk surface in a non-contact manner: Infrared absorbing ink: The main component is phthalocyanine infrared absorbing dye, which has strong absorption characteristics in the 950nm band and is completely transparent under visible light. It is used to provide a planar sub-pixel level positioning reference. Upconversion fluorescent ink: The main component is NaYF4:Yb,Er nanoparticles, which emit 550nm green fluorescence under 980nm laser excitation and are used to measure the Z-axis height in conjunction with structured light. Thermochromic ink: Its main component is spiropyran-based thermosensitive material. It changes from colorless to blue at temperatures above 40°C and is used to record the thermal deformation history of subsequent color-fixing processes.

[0049] The dot matrix layout employs a non-uniform strategy: denser dots are placed in the neckline curve area and corner areas, with a dot spacing of 5mm; sparser dots are placed in the flat areas of the cut piece, with a dot spacing of 20mm. A one-to-one mapping relationship is established between the theoretical coordinates of each marker point and the positioning coordinates of the neckline pattern in the pre-press document, with a total of approximately 1200 marker points.

[0050] 2. Multispectral three-dimensional sensing Two sets of multispectral line-scanning cameras are positioned above the guide strip along the transmission direction, located upstream and downstream of the nozzle, respectively. Each camera is equipped with dual-channel filters for the infrared band (950nm bandpass) and the visible light band, with a resolution of 2048 pixels and a scanning frequency of 10kHz. A structured light projection module is positioned in front of the nozzle, emitting a 980nm laser with a power of 50mW and a projection angle of 30°.

[0051] When the cut piece is transported to the area below the upstream camera at a speed of 20m / min: An infrared camera captures infrared-absorbing ink dots, and a Gaussian fitting sub-pixel center positioning algorithm is used to obtain the precise planar coordinates (X,Y) of each marker point, with a positioning accuracy of ±0.01mm. The structured light projection module projects a 980nm laser onto the fabric surface, and the upconversion fluorescent ink is stimulated to emit 550nm green light. The Z-axis height of each marker point is calculated using the triangulation principle, with a measurement accuracy of ±0.02mm. The central controller compares the measured coordinates with the theoretical coordinates and uses a thin plate spline interpolation algorithm to construct the nonlinear deformation field function Δ(x,y,z) of the entire cut surface.

[0052] Measurements showed that during the conveyor belt transport process, the difference in warp stretch rate of the silk fabric piece in the neckline arc area reached 0.5%, the difference in weft shrinkage rate reached 0.3%, and the edge arching height was approximately 0.8 mm.

[0053] 3. Tension decoupling control The guide belt is equipped with multi-zone independently controlled tension roller groups, which are divided into 12 independent control zones along the width direction. Each control zone is 100mm wide and is equipped with an independent servo motor (rated torque 2Nm) and force sensor (range 0-200N, accuracy ±0.5%).

[0054] The central controller uses a model predictive control algorithm to inversely calculate the target tension distribution based on the deformation field function Δ(x,y,z). The control cycle is 50ms, and the prediction time domain is 10 cycles. The specific control logic is as follows: If the warp stretch rate of a certain area is detected to be greater than 0.2%, the torque of the tension roller corresponding to that area is reduced to release the stretch. If the weft shrinkage rate is detected to be greater than 0.1%, the clamping force of the tension roller in the edge area is increased to suppress shrinkage. If the detected Z-axis warpage height is greater than 0.2mm, the tension difference between adjacent areas is adjusted according to the warpage direction to flatten the fabric.

[0055] After tension decoupling control, the maximum stretch rate of the cut piece is reduced to 0.15%, the maximum weft shrinkage rate is reduced to 0.08%, and the Z-axis warpage is reduced to below 0.2mm.

[0056] 4. Inkjet compensation The residual deformation field Δ_res(x,y) after tension adjustment still exists and requires inkjet compensation.

[0057] (1) Image reverse pre-deformation Let the original neckline floral design image be I(u,v) with a resolution of 600 dpi. The central controller calculates the compensation field Δ_reverse(x,y) based on the residual deformation field, ensuring that the pixel value ejected by the nozzle at the physical coordinates (x,y) satisfies: P(x,y) = I(u,v), where (u,v) = (x,y) + Δ_reverse(x,y); This formula ensures a one-to-one correspondence between the inkjet droplets and the pixels of the original design image. Image resampling is performed using a bicubic interpolation algorithm, with a computation time of approximately 0.5 seconds per square meter.

[0058] (2) Ignition pulse reconstruction The nozzle is an industrial-grade piezoelectric nozzle with a native ignition frequency of 30kHz, corresponding to a resolution of 600dpi. The nozzle control unit dynamically adjusts the ignition pulse parameters according to the compensation field. In the stretching region (deformation compensation factor > 1.02), the ignition frequency is reduced to 28-29kHz to increase the physical spacing between ink droplets; In the shrinkage region (deformation compensation factor <0.98), the ignition frequency is increased to 31-32kHz to compress the ink droplet spacing; The frequency adjustment step size is 0.1kHz, and the adjustment accuracy is ±0.05kHz.

[0059] Meanwhile, the ink volume conservation controller calculates the dye mass per unit area in real time and adjusts the ink droplet volume parameters (by changing the amplitude of the piezoelectric waveform, with an adjustment range of ±10%) to ensure that the total dye mass per unit area remains constant before and after compensation.

[0060] 5. Closed-loop feedback After printing, the cut piece continues to be transported to the downstream multispectral camera. The downstream camera again captures the three-dimensional coordinates of the spectral feature matrix, and the central controller compares the actual placement deviation with the theoretical value.

[0061] In this embodiment, the actual registration deviation is +0.02mm (target is 0), which is within the allowable range (preset threshold ±0.03mm). The deviation value is recorded and fed back to the parameter correction module of the tension decoupling control unit and the inkjet compensation unit for fine-tuning in the next batch or the next cycle.

[0062] 6. Post-process traceability and effect verification After the cut pieces are printed, they enter the color-fixing process at 102℃ for 8 minutes. During the color-fixing process, the thermochromic ink records the history of thermal deformation. The subsequent imaging module captures the color changes of the thermochromic ink and feeds the thermal deformation data back to the central controller to correct the deformation field function model for subsequent batches.

[0063] Tests showed that the overprinting accuracy of this embodiment in the cheongsam collar matching print reached ±0.025mm (n=50), and the patterns on the left and right sides of the collar were completely symmetrical; the edge roughness Ra was 7.8μm (measurement length 5mm); and the shrinkage rate after subsequent washing was consistent at ±0.4% (warp) and ±0.3% (weft).

[0064] Example 2 This embodiment provides a closed-loop compensation method for silk positioning printing based on spectral feature matrix and tension decoupling control, which is applied to the scenario of four-corner positioning patterns and gold thread overlock printing on silk scarves.

[0065] The scarf measures 90cm x 90cm and is made of 16 momme plain crepe satin. The four corners are designed with a meander pattern for positioning. The edges are then mechanically overlocked with gold thread, and the overlocking path must be perfectly aligned with the printed pattern.

[0066] Parameter adjustment: A dense array of spectral feature dots is placed in the four corner areas of the scarf, with a dot spacing of 3mm, and approximately 100 marker dots are placed in each of the four corner areas. The tension roller assembly is divided into 8 independent control zones, each with a width of 112.5 mm; The deformation field was constructed using a thin-plate spline interpolation algorithm, achieving a spatial resolution of 0.5mm × 0.5mm.

[0067] Effect verification: Testing showed that the alignment accuracy between the gold thread overlock and the printed pattern in this embodiment reached ±0.03mm (n=30), increasing the product qualification rate from 85% in the traditional solution to 98%. Thermal deformation data recorded by the thermochromic ink showed that the shrinkage rate difference at the four corners of the scarf during the color-fixing process was 0.2%~0.4%. This data was fed back to the deformation field model for feedforward compensation in subsequent batches.

[0068] Example 3 This embodiment provides a closed-loop compensation method for silk positioning printing based on spectral feature matrix and tension decoupling control, which is applied to multi-color overprinting scenarios of light-colored patterns on dark silk fabrics.

[0069] The fabric is black crepe silk (14 momme), which requires first spraying white ink as a base, then overprinting colorful patterns, and finally spraying varnish to increase its shine. The three processes are carried out in sequence, with the fabric moving back and forth on the conveyor belt multiple times.

[0070] Parameter adjustment: The spectral feature matrix is ​​implanted during the first printing, and all subsequent processes use the same matrix as a reference. Upstream and downstream cameras perform deformation sensing before each process, and the accumulated deformation data is transmitted step by step; The inkjet compensation module adjusts the compensation amount based on the cumulative deformation data.

[0071] Effect verification: Testing in this embodiment showed that the registration accuracy of the white ink, color, and varnish processes were ±0.028mm, ±0.025mm, and ±0.030mm (n=40), respectively, with no "white showing" phenomenon and a color difference ΔE < 1.5 (CIE Lab color difference formula). Data recorded by the ink volume conservation controller showed that the dye mass deviation per unit area before and after compensation was less than ±1.2%.

[0072] Comparative Example 1 This comparative example uses the method disclosed in CN114905864A to print the same silk cheongsam fabric piece as in Example 1, but does not perform the marking layout, tension decoupling control and closed-loop feedback of the present invention. Instead, it only performs image interpolation deformation compensation by collecting the inherent yarn texture of the fabric.

[0073] Test results: Registration accuracy: ±0.12mm (n=50), left and right pattern deviation at the collar: 0.08-0.15mm; Edge roughness: Ra 24.5μm (measurement length 5mm); Shrinkage consistency: ±3.2% (warp); Scrap rate: 18%.

[0074] Comparative Example 2 This comparative example uses the method disclosed in CN107901622B to perform gold thread overprinting on the same silk scarf as in Example 2, and uses the inherent floral pattern feature points of the fabric for image deformation compensation.

[0075] Test results: Alignment accuracy between gold thread overlocking and printed pattern: ±0.48mm (n=30); Due to the lack of obvious floral patterns on the silk surface, the feature point extraction failure rate is approximately 30%. Product qualification rate: 72%.

[0076] Comparison of experimental data The test results of Examples 1-3 and Comparative Examples 1-2 are summarized below:

[0077] The above data shows that the embodiments of the present invention are superior to the comparative examples in terms of overprinting accuracy, edge quality, deformation compensation, dimensional stability and product qualification rate.

[0078] Working principle The closed-loop compensation mechanism of this invention is based on the following principle: (1) The synergistic principle of active labeling and 3D perception Silk surfaces lack distinct yarn textures or patterns, making it difficult to establish reliable deformation benchmarks using traditional methods. This invention solves this benchmark deficiency problem by actively embedding functional ink markers. The infrared-absorbing ink is transparent under visible light, preserving the aesthetic appeal of the pattern, while exhibiting high absorption in the 950nm band, allowing for precise identification by an infrared camera. The upconversion fluorescent ink emits 550nm visible light upon 980nm excitation; using structured light triangulation, the fabric surface height can be calculated based on the offset of the fluorescent spot. The combination of these two inks enables a single marker to simultaneously provide both planar coordinates and height information.

[0079] (2) Synergistic principle of deformation field construction and decoupling control Silk exhibits nonlinear and anisotropic elastic deformation, with a coupling effect between warp stretching and weft contraction. This invention uses a thin-plate spline interpolation algorithm to interpolate discrete marker point data into a continuous deformation field function, which simultaneously includes planar deformation components and Z-axis warping components. Based on this, a model predictive control algorithm inversely calculates the tension distribution. By decoupling the servo motor torques of 8-16 independent control zones, warp tension and weft contraction are decoupled, thus suppressing the deformation source at a physical level. Based on an established deformation prediction model, this algorithm aims to minimize the difference between the measured deformation and the target deformation in each control zone, while also considering the energy consumption of tension adjustment. It solves for the optimal torque control sequence within the prediction time domain (e.g., 10 control cycles).

[0080] (3) The principle of breakthrough in accuracy of double-layer compensation Single compensation methods have physical limitations: tension control has a limited response speed (50ms level) and cannot correct micron-level high-frequency deformation; inkjet compensation has a limited correction range, and image distortion easily occurs when the deformation exceeds 0.5%. This invention adopts a "coarse adjustment + fine adjustment" dual-layer architecture: tension control suppresses macroscopic deformation to below 0.2%, creating a linear working range for inkjet compensation; inkjet compensation corrects residual micron-level deformation on this basis. Specifically, the printhead control unit dynamically reconstructs the ignition pulse according to the compensation field: by establishing a proportional relationship between the deformation compensation factor and the ignition frequency adjustment (for example, for every 1% change in the compensation factor, the ignition frequency is adjusted accordingly by 0.5-1kHz), and combined with the fine adjustment of the piezoelectric waveform amplitude, precise control of the ink droplet landing point position and ink volume is achieved. The two complement each other in the time domain (slow response of tension control, fast response of inkjet compensation) and the spatial domain (macroscopic of tension control, microscopic of inkjet compensation), enabling the final registration accuracy to break through the physical limits of their respective individual use.

[0081] (4) Adaptive principle of closed-loop feedback Open-loop control cannot handle batch-to-batch variations and cumulative errors. This invention uses an upstream camera to capture input deformation and a downstream camera to provide feedback on post-printing deviations, forming a real-time closed loop with a control cycle of 50ms. Simultaneously, the thermochromic ink records the thermal deformation history of the subsequent curing process, forming a cross-process closed loop that feeds downstream data forward to the upstream model. This dual closed loop enables the system to adaptively evolve and automatically correct batch-to-batch variations.

[0082] Industrial application verification 1. Equipment adaptability verification The technical solution of this invention can be implemented by modifying existing digital printing equipment. The main modifications include: A multispectral camera and a structured light projection module are installed above the guide belt, and the relative position accuracy between the installation position and the nozzle is required to be ±1mm. The single-zone tension roller group was transformed into an 8-16 zone independently controlled tension roller group, with a servo motor and force sensor added to each control zone; The control system software has been upgraded to include modules for deformation field construction, model predictive control, image pre-deformation, and ignition pulse reconstruction.

[0083] Verification showed that the production speed of the modified equipment remained at 20-30 m / min, which was comparable to that before the modification, and had no negative impact on production efficiency.

[0084] 2. Production stability verification The stability of the scheme in Example 1 was tested under continuous production conditions (8 hours per day, 30 days of continuous operation). Ten samples were randomly selected daily to measure the overprinting accuracy, and the results are as follows: Average registration accuracy: ±0.026mm Standard deviation: 0.004 mm Range: 0.018mm (maximum ±0.035mm, minimum ±0.021mm) Number of equipment downtimes: 0 The above data shows that the solution of the present invention has good production stability.

[0085] 3. Cost-benefit analysis Based on an annual production of 1 million pieces of silk cheongsam collar printed products: The traditional solution has a scrap rate of 15%, resulting in 150,000 scrapped items per year. The scrap rate of this invention is 3%, with 30,000 scrapped items per year; The number of waste products is reduced by 120,000 pieces per year, which, based on a cost of 50 yuan per piece, results in annual cost savings of 6 million yuan.

[0086] Functional ink cost: Each product consumes approximately 0.2 mL of functional ink, costing about 0.5 yuan per unit, resulting in an annual cost increase of 500,000 yuan. Equipment upgrade investment is approximately 300,000 yuan. Taking all factors into account, the annual net benefit is approximately 5.2 million yuan, with an investment payback period of approximately 2 months.

[0087] 4. Scope of application verification The solution of this invention has been verified on the following silk products, and good results have been achieved in all of them: 12-16 Momme Crepe Satin: Cheongsam, Scarf 8-10 momme power textiles: ties, ribbons 14-20 momme crepe de chine: garment pattern pieces Twill silk: scarves, handkerchiefs Heavy crepe (30 momme and above): Formal wear fabric The above verification results show that the present invention has good substrate adaptability.

[0088] In summary, this invention effectively solves the micron-level overprinting error caused by nonlinear elastic deformation and Z-axis warping in digital positioning printing of silk by actively deploying spectral feature dot matrix, multispectral three-dimensional sensing, multi-zone tension decoupling control, inkjet pixel-level reconstruction, and closed-loop feedback throughout the entire process. It has significant industrial application value.

Claims

1. A closed-loop compensation method for silk positioning printing based on spectral feature lattice and tension decoupling control, characterized in that, Includes the following steps: Step S1: Marking and Layout Steps A spectral feature dot matrix is ​​arranged on the surface of a silk fabric. The spectral feature dot matrix is ​​composed of at least two functional inks, including infrared absorption ink and upconversion fluorescent ink. The infrared absorption ink is used to provide a planar positioning reference, and the upconversion fluorescent ink is used to measure the Z-axis height of the fabric surface in conjunction with structured light. Step S2: Multispectral Sensing Step The three-dimensional spatial coordinates of the spectral feature array are captured in real time by a multispectral imaging system, which includes an infrared camera and a structured light projection module. A nonlinear deformation field function of the fabric surface is constructed based on the captured coordinates. The nonlinear deformation field function describes both the planar deformation and Z-axis warping of the fabric. Step S3: Tension decoupling control step According to the nonlinear deformation field function, the multi-zone independent control tension roller group set on the guide belt is decoupled and controlled. By adjusting the output torque of each tension zone, the macroscopic nonlinear deformation of the fabric is suppressed, so that the fabric deformation is reduced to below the preset threshold. Step S4: Inkjet compensation step Based on the residual deformation field suppressed in step S3, the image data to be printed is subjected to reverse pre-deformation processing, and the ignition pulse parameters of the printhead are dynamically reconstructed so that the printhead reduces the ignition frequency in the stretching region and increases the ignition frequency in the shrinking region to compensate for micron-level residual deformation. Step S5: Closed-loop feedback step The fabric printed in step S4 is then subjected to the multispectral sensing system in step S2 to obtain the actual placement deviation, and the deviation is fed back to steps S3 and S4 to form a closed-loop control for the entire process.

2. The method according to claim 1, characterized in that, The spectral feature matrix also includes a thermochromic ink, which is used to trace the thermal deformation history of the fabric in the subsequent color fixing process and to feed the thermal deformation data back to the closed-loop feedback step S5 to correct the deformation field function model.

3. The method according to claim 1, characterized in that, The spectral feature dot matrix described in step S1 adopts a non-uniform dot distribution strategy: the dot density in the fabric edge area, arc area and corner area is higher than the dot density in the flat area of ​​the fabric. The dot spacing in the edge area, arc area and corner area is 5mm-10mm, and the dot spacing in the flat area is 15mm-25mm.

4. The method according to claim 1, characterized in that, In step S2, the nonlinear deformation field function is constructed using a thin plate spline interpolation algorithm. The measured coordinates of each captured marker point are compared with the theoretical coordinates in the pre-press file, and the continuous deformation field function of the entire fabric surface is obtained through interpolation calculation.

5. The method according to claim 3, characterized in that, The spacing between the dots in the edge area, arc area, and corner area is 3mm-5mm, and the spacing between the dots in the flat area is 15mm-20mm.

6. The method according to claim 1, characterized in that, The reverse pre-deformation process described in step S4 is as follows: Let the original design image be I(u,v) and the residual deformation field function on the current fabric surface be Δ(x,y). Then the pixel value P(x,y) sprayed by the nozzle at the physical coordinates (x,y) satisfies P(x,y)=I(u,v), where (u,v)=(x,y)+Δreverse(x,y), and Δreverse is a compensation field opposite to the direction of the measured deformation field.

7. The method according to claim 1, characterized in that, In step S4, when dynamically reconstructing the ignition pulse parameters of the printhead, an ink volume conservation constraint is simultaneously executed: ensuring that the total mass of dye per unit area remains constant before and after compensation. This is achieved by adjusting the ink droplet volume or the number of ink droplets to avoid color deviation caused by deformation compensation.

8. A closed-loop compensation system for silk positioning printing based on spectral feature lattice and tension decoupling control, used to implement the method according to any one of claims 1-7, characterized in that, include: A marking unit is used to lay out a spectral feature dot matrix on the surface of silk fabric. The marking unit includes at least one set of digital inkjet printheads or roller coating units for printing infrared absorbing ink and upconversion fluorescent ink. A multispectral sensing unit includes at least two sets of multispectral line scan cameras respectively disposed upstream and downstream of the nozzle along the transmission direction of the guide band, and a set of structured light projection modules. The multispectral line scan cameras are equipped with dual-channel filters for infrared and visible light bands to capture the three-dimensional spatial coordinates of the spectral feature array in real time. A force decoupling control unit includes a tension roller group consisting of multiple independent control zones arranged along the width direction of the guide belt, and a model prediction controller connecting each control zone, used to adjust the output torque of each control zone according to the nonlinear deformation field function; An inkjet compensation unit includes an image pre-deformation module and a printhead ignition pulse reconstruction module. The image pre-deformation module is used to perform reverse pre-deformation processing on the image data to be printed, and the printhead ignition pulse reconstruction module is used to dynamically adjust the ignition frequency and waveform parameters of the printhead. A central controller is connected to the multispectral sensing unit, the tension decoupling control unit, and the inkjet compensation unit respectively. It is used to receive data from the sensing unit, calculate the nonlinear deformation field function, and coordinate the control of the tension decoupling control unit and the inkjet compensation unit.

9. The system according to claim 8, characterized in that, The multispectral sensing unit also includes a temperature-sensitive imaging module located downstream of the printhead. The temperature-sensitive imaging module is used to capture the color change of the temperature-sensitive color-changing ink after color fixation and feed the change data back to the central controller to correct the deformation field function model of subsequent batches.

10. The system according to claim 8, characterized in that, The inkjet compensation unit also includes an ink volume conservation controller, which is linked with the printhead ignition pulse reconstruction module to calculate the dye mass per unit area in real time based on the deformation compensation amount and adjust the ink droplet volume parameters to maintain a constant total dye mass.

Images