An AI-driven garment rapid feedback processing and manufacturing method, system and intelligent terminal

Abstract

The application relates to an AI-driven garment rapid feedback processing and manufacturing method, system and intelligent terminal, and relates to the technical field of garment manufacturing, which comprises the following steps: S10, collecting order information; S11, obtaining detection fabric based on the order information; S12, collecting detection spectral parameters and transmission detection information of the detection fabric, and forming a detection image according to the detection spectral parameters; S13, obtaining abnormal parameters and material arrangement traction parameters by combining the detection image and the transmission detection information; S14, generating a fabric arrangement scheme by the order information, the detection fabric and the abnormal parameters, and arranging materials according to the fabric arrangement scheme and the material arrangement traction parameters; S15, after the material arrangement is completed, obtaining detection processing parameters based on the detection image, the detection fabric and the fabric arrangement scheme, and processing the fabric according to the detection processing parameters to form a detection garment; and S16, conveying the garment by combining the detection garment and the order information. The application has the effect of improving the yield of garment production.

Description

Technical Field

[0001] This invention relates to the technical field of garment manufacturing, and in particular to an AI-driven fast-response garment manufacturing method, system, and intelligent terminal. Background Technology

[0002] Garment manufacturing is a comprehensive industrial process that transforms raw materials (mainly textiles) into finished garment products.

[0003] Garment manufacturing involves multiple steps, including material selection, cutting, transportation, and sewing. During the material selection process, the fabric is usually selected by visual observation or by taking photos of the fabric using a simple RGB industrial camera. After the fabric selection is completed, it is cut manually and transported to the sewing station. The sewing workers then sew the fabric according to the shape required by the order to form the final garment.

[0004] In the garment manufacturing process, garment orders are characterized by a variety of styles, small order quantities, and short delivery periods. When observing fabrics by human vision or by taking photos with a simple RGB industrial camera, it is not easy to perform real-time quantitative analysis of physical parameters such as elastic modulus, weight, and moisture content. This can easily lead to the selection of fabrics with endogenous physical damage for subsequent sewing operations, resulting in poor quality of finished garments. Summary of the Invention

[0005] To improve the yield rate of garment manufacturing, this invention provides an AI-driven fast-response garment manufacturing method.

[0006] In a first aspect, the present invention provides an AI-driven rapid-response garment manufacturing method, employing the following technical solution: An AI-driven rapid-response garment manufacturing method includes: S10: Collect order information; S11: Obtain the inspected fabric based on order information; S12: Collect the detection spectral parameters and transmission detection information of the fabric to be tested, and form a detection image based on the detection spectral parameters; S13: Combine the detection image with the transmission detection information to obtain the abnormal parameters and the material discharge traction parameters; S14: Generate a fabric layout plan based on order information, inspected fabric, and abnormal parameters; then lay out the fabric using the fabric layout plan and the material laying traction parameters. S15: After completing the material layout, the detection processing parameters are obtained based on the detection image, the detection fabric and the fabric layout scheme, and the fabric is processed with the detection processing parameters to form the detection garment. S16: Combine the inspection of the garments with the order information to transport the garments.

[0007] By adopting the above technical solution, the order information is analyzed to initially select the fabric for testing. Then, based on the detection spectral parameters and transmission detection information of the tested fabric, the elastic modulus, weight, moisture content, and intrinsic physical damage of the fabric can be quantitatively identified before the fabric is laid out. This reduces the situation where fabrics with damage risks are used in the sewing process. Furthermore, a fabric layout plan is generated through abnormal parameters, and the fabric is laid out in combination with the laying traction parameters. This can reduce problems such as misalignment of cut pieces and uneven seam shrinkage caused by physical defects in the fabric. Finally, the detection and processing parameters are obtained based on the detection images, the tested fabric, and the fabric layout plan. This achieves closed-loop parameter adaptive adjustment from fabric detection to sewing processing, thereby effectively improving the yield and efficiency of garment production.

[0008] Optionally, methods for generating fabric layout schemes include: S140: Calculate the detection similarity based on the detected fabric; S141: The joint fabric is obtained by comparing the detection similarity with the preset benchmark similarity. S142: Generate a fabric layout plan by combining the tested fabric, combined fabric, and order information; S143: Obtain the remaining outline of the cut fabric based on the fabric layout plan and the inspected fabric, and generate a layout barcode based on the fabric layout plan and the remaining outline of the cut fabric and record it in the warehouse.

[0009] Optionally, methods for obtaining the detection processing parameters include: S20: Retrieve the layout fabric and the layout weight of the layout fabric from the fabric layout scheme, and extract the elastic modulus parameter of the layout fabric from the detection spectral parameters. S21: Obtain air cushion control parameters based on the weight arrangement; S22: Collect the fabric laying speed, electrostatic potential gradient, and fabric edge image, and calculate the difference between the fabric laying speed and the preset fabric feeding roller speed as the speed deviation value. S23: Update the fabric feeding roller speed based on the speed deviation value; S24: The neutralization and ionization parameters are obtained through the electrostatic potential gradient, and the cutting and resting time is calculated based on the elastic modulus parameter; S25: Identify edge features from fabric edge images; S26: Compare the edge features with the preset baseline features to obtain the contrast, and update the fabric edge image with the contrast. S27: Obtain the neatness based on the fabric edge image and edge features, and compare the neatness with the preset benchmark neatness to obtain the fabric conveying direction; S28: Integrate the air cushion control parameters, fabric feeding roller speed, neutralization and ionization parameters, cutting and resting time, and fabric conveying direction into fabric laying parameters, and combine the fabric laying parameters with the detection image to obtain the detection and processing parameters.

[0010] By adopting the above technical solution, the air cushion control parameters are obtained by arranging the weight, and closed-loop adjustment is performed in combination with the speed deviation value. This enables adaptive tension control during the fabric laying process. By using the neutralization and ionization parameters and combining them with the elastic modulus parameters, the cutting resting time is calculated, reducing the risk of misalignment of cut pieces caused by electrostatic adsorption or elastic recoil. Furthermore, the fabric conveying direction is obtained through the fabric edge image, achieving automatic correction of the fabric laying direction. Finally, the air cushion control parameters, fabric release roller speed, neutralization and ionization parameters, cutting resting time, and fabric conveying direction are integrated into the fabric laying parameters, providing a high-precision cut piece basis for subsequent sewing, reducing quality problems caused by fabric laying errors, and improving the yield and efficiency of garment production.

[0011] Optionally, methods for obtaining the detection processing parameters also include: S30: Collects the air resistance of the fabric and the acoustic emission signals of the cutting knife; S31: Combining fabric air resistance with fabric layout to achieve negative pressure strength; S32: Obtain the initial cutting path based on the fabric layout scheme and the laid-out fabric; S33: Compensate the initial cropping path based on the detected image and abnormal parameters to obtain the detection cropping path; S34: Obtain the bounding box parameters of the cutting blade by detecting the cutting path and the preset cutting blade model; S35: Identify high-frequency characteristic spectra from acoustic emission signals and obtain the wear level based on the high-frequency characteristic spectra and a preset reference spectrum signal; S36: Obtain a correction strategy based on the wear level; S37: Integrate negative pressure strength, detection cutting path, bounding box parameters, and correction strategies into cutting parameters, and use the cutting parameters and fabric laying parameters as detection processing parameters.

[0012] By adopting the above technical solution, combining the air resistance of the fabric surface with the negative pressure strength obtained from the fabric arrangement, the fabric layer can be stably fixed during the cutting process, reducing cutting deviation. By detecting images and compensating for abnormal parameters in the initial cutting path, a detection cutting path is obtained to avoid intrinsic physical damage to the fabric during the cutting process. The wear level is obtained using acoustic emission signals and a correction strategy is generated to achieve real-time monitoring and adaptive adjustment of the cutting blade status. Finally, the negative pressure strength, detection cutting path, bounding box parameters, and correction strategy are integrated into cutting parameters to improve cutting accuracy and blade life, reduce the defect rate of finished products caused by cutting defects, and improve the yield and efficiency of garment production.

[0013] Optionally, other methods for obtaining the clipping parameters include: S40: Obtain the cutting area based on the detected cutting path; S41: Obtain the cut piece weight and reference pressure value based on the cutting area and fabric layout; S42: Calculate the electrostatic power based on the weight of the cut piece; S43: Use electrostatic power to adsorb the cut piece and collect the gripping pressure value; S44: Compare the gripping pressure value with the reference pressure value to calculate the pressure deviation value and obtain the number of detection layers; S45: Compare the number of detection layers with the preset reference number of layers and combine the pressure deviation value to obtain the pulse airflow parameters; S46: Run with pulsed airflow parameters to update the number of detection layers until the number of detection layers matches the reference number, and add electrostatic power and pulsed airflow parameters to the trimming parameters.

[0014] Optionally, methods for obtaining the detection processing parameters also include: S50: After cutting is completed, control the preset transport robot to collect the cut pieces and collect environmental detection information; S51: Define the currently collected transport robot as a marked robot, and treat the remaining transport robots as other robots; S52: Retrieve other operating parameters from other robots; S53: Obtain the detection operation vector based on environmental monitoring information and other operating parameters; S54: Obtain the marked driving strategy by detecting the running vector, and collect the workstation detection values ​​and order pass rate; S55: Combine the workstation inspection values ​​with the order qualification rate to obtain the cutting piece sorting order; S56: Integrate the marking driving strategy and the piece sorting sequence into transportation parameters, and add the transportation parameters to the inspection and processing parameters.

[0015] Optionally, methods for obtaining transportation parameters also include: S60: Collects the battery level and location of the marker robot; S61: Compare the marked power level with the preset baseline power level to collect backup location data; S62: Obtain the alternative path by marking the location and the alternative location; S63: Select the backup robot corresponding to the backup path with the shortest distance as the target robot; S64: Control the target robot to move to the marked position to continue executing the marked driving strategy and the piece sorting sequence, and add the marked position and the target robot to the transportation parameters.

[0016] Optionally, methods for obtaining the detection processing parameters also include: S70: Based on transport parameters, acquire sewing images of the cut pieces after transport and needle bar resistance; S71: Identify abnormal sewing parameters using sewing images; S72: Retrieve the fabric thickness corresponding to the sewing image from the transportation parameters; S73: Predicts thread tension based on fabric thickness, sewing image, needle bar resistance, and preset sewing speed; S74: Compare the predicted thread tension with the preset baseline thread tension and combine the preset baseline operating parameters with the sewing abnormal parameters to obtain the sewing operating parameters; S75: Add the sewing operation parameters to the detection and processing parameters.

[0017] Optionally, methods for obtaining sewing operation parameters include: S80: Collects temperature detection information and presser foot pressure value; S81: Compare the temperature detection information with the preset reference temperature information to obtain the cooling spray parameters and re-collect the temperature detection information until the temperature detection information is less than the reference temperature information. S82: Retrieve the cut piece specifications corresponding to the sewing image from the transportation parameters; S83: Obtain the reference presser foot pressure based on the cut piece specifications; S84: Compare the pressure foot pressure value with the reference pressure foot pressure to calculate the pressure deviation value; S85: Update the sewing operation parameters based on the pressure deviation value and add the cooling spray parameters to the sewing operation parameters.

[0018] Secondly, this application provides an AI-driven rapid-response garment manufacturing system, which adopts the following technical solution: An AI-driven rapid-response garment manufacturing system includes: The fabric multimodal quality inspection module is used to collect detection spectral parameters and transmission detection information, and to generate detection images using the detection spectral parameters. An adaptive flexible fabric laying module is used to receive and execute fabric laying parameters to lay the fabric. The non-rigid cutting module is used to receive and execute cutting parameters to cut the fabric. The collaborative handling module is used to receive and execute transportation parameters to transport the cut pieces formed after the fabric is cut, and after transportation, to receive and execute sewing operation parameters to sew the cut pieces into garments. The production scheduling module is used to collect data uploaded by the fabric multimodal quality inspection module, the adaptive flexible fabric spreading module, the non-rigid cutting module, and the collaborative handling module, and generate fabric laying parameters, transportation parameters, and sewing operation parameters and output them to the corresponding modules.

[0019] Thirdly, this application provides a smart terminal, which adopts the following technical solution: A smart terminal includes a memory and a processor, wherein the memory stores a computer program that can be loaded and executed by the processor to perform an AI-driven rapid-response garment manufacturing method.

[0020] Fourthly, this application provides a digital twin real-time synchronization and predictive maintenance system for a garment fast-response factory, characterized in that it collects total operating parameters and constructs a detection and operation model, and then uses the detection and operation model to conduct real-time production simulation to obtain fault parameters for early warning.

[0021] Fifthly, this application provides an adaptive material layout and surplus material utilization optimization system for fragmented production, characterized in that it executes S140 to S143 to generate a fabric layout scheme and material layout barcode, and records the material layout barcode in the warehouse.

[0022] In a sixth aspect, this application provides a zero-tension automatic fabric laying system based on fluid dynamics feedback, characterized in that it controls a preset set of micro-pore array nozzles to execute the air cushion control parameters in the fabric laying parameters S28.

[0023] In a seventh aspect, this application provides an automated piece-grabbing system based on the fusion of electrostatic field and flexible tactile sensation, characterized in that it is used to execute the electrostatic power and pulse airflow parameters generated in S40 to S46.

[0024] Eighthly, this application provides a physical sensing online control system for sewing stitch patterns, characterized in that it is used to execute S70 to S75 to generate predicted thread tension and execute sewing operation parameters for sewing compensation.

[0025] Ninthly, this application provides a microelectromechanical system based on a proportional-integral control algorithm, characterized in that it is used to control the air cushion control parameters of each nozzle in the micro-orifice array.

[0026] In summary, this application includes at least one of the following beneficial technical effects: 1. By analyzing order information to initially select the fabric for inspection, and then based on the spectral parameters and transmission detection information of the inspected fabric, the elastic modulus, weight, moisture content and intrinsic physical damage of the fabric can be quantitatively identified before the fabric is laid out. This reduces the situation where fabrics with damage risks are put into the sewing process. Furthermore, the fabric layout plan is generated through abnormal parameters and the fabric laying out is carried out in combination with the laying out traction parameters. This can reduce problems such as misalignment of cut pieces and uneven seam shrinkage caused by physical defects of the fabric. Finally, the inspection and processing parameters are obtained based on the inspection images, inspected fabrics and fabric layout plan. This realizes the closed-loop parameter adaptive adjustment from fabric inspection to sewing processing, thereby effectively improving the yield and efficiency of garment production. 2. By obtaining air cushion control parameters through fabric weight and combining them with speed deviation values ​​for closed-loop adjustment, adaptive tension control during fabric laying can be achieved. By utilizing neutralization and ionization parameters and combining them with elastic modulus parameters, the cutting resting time is calculated, reducing the risk of fabric misalignment caused by electrostatic adsorption or elastic recoil. Furthermore, the fabric conveying direction is obtained through fabric edge images, enabling automatic correction of the fabric laying direction. Finally, the air cushion control parameters, fabric release roller speed, neutralization and ionization parameters, cutting resting time, and fabric conveying direction are integrated into fabric laying parameters, providing a high-precision fabric base for subsequent sewing, reducing quality problems caused by fabric laying errors, and improving the yield and efficiency of garment production. 3. By combining fabric air resistance and fabric layout to obtain negative pressure strength, the fabric layer can be stably fixed during the cutting process, reducing cutting deviation. The detection cutting path is obtained by compensating the initial cutting path through detection images and abnormal parameters, so as to avoid intrinsic physical damage to the fabric during the cutting process. The wear level is obtained by using acoustic emission signals and a correction strategy is generated to realize real-time monitoring and adaptive adjustment of the cutting blade status. Finally, the negative pressure strength, detection cutting path, bounding box parameters, and correction strategy are integrated into cutting parameters to improve cutting accuracy and blade life, reduce the defect rate of finished products caused by cutting defects, and improve the yield and efficiency of garment production. Attached Figure Description

[0027] Figure 1 This is a flowchart of an AI-driven rapid response garment manufacturing method according to an embodiment of the present invention.

[0028] Figure 2 This is a flowchart of a method for obtaining a detection image and transmission detection information according to an embodiment of the present invention; Figure 3 This is a flowchart of an AI-driven rapid response garment manufacturing method according to an embodiment of the present invention. Detailed Implementation

[0029] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0030] Reference Figure 1 and Figure 2 This application discloses an AI-driven rapid-response garment manufacturing method, including the following steps: S10: Collect order information.

[0031] Order information refers to parameters such as the size and material of the garments to be made. This information is obtained by connecting to e-commerce platforms in real time to retrieve trend forecast data and order details, or by scanning paper orders or reading electronic documents (PDF, Excel), and generating structured data as order information through image OCR recognition and field parsing. Order information can also be manually entered by operators through a human-computer interaction interface.

[0032] S11: Obtain the fabric for inspection based on order information.

[0033] Fabric testing refers to fabrics that meet the requirements for garment production in the order information. Based on the material composition (such as cotton, polyester, spandex, etc.) and weight range in the order information, candidate fabrics that meet the physical property similarity threshold (such as elastic modulus, thickness deviation ≤5%) are selected from the preset fabric library as test fabrics.

[0034] If the fabric exists in the fabric library, a fabric retrieval command is generated to transport the corresponding fabric roll to the workstation for subsequent testing operations.

[0035] If the inventory is insufficient or there is no matching fabric, a purchase requisition will be automatically generated, and the estimated delivery time of the order will be updated accordingly.

[0036] Before entering the production line, the fabrics retrieved from the fabric library must be accompanied by their historical traceability identifier (if any) or a temporary batch number to be inspected, so that subsequent detection data such as spectral fingerprints and defect coordinates can be linked to the batch.

[0037] S12: Collect the detection spectral parameters and transmission detection information of the fabric to be tested, and form a detection image based on the detection spectral parameters.

[0038] Reference Figure 2 The detection spectral parameters refer to the continuous spectral transmittance data of the fabric in the 400nm to 2500nm band, as well as the molecular-level material fingerprint of the fabric extracted by deep convolutional neural networks (PCA-AlexNet, MobileViT), including quantitative indicators such as fiber composition ratio, dye and chemical distribution consistency, moisture content, and weight.

[0039] The fabric was continuously sampled and tested using a hyperspectral imager in the 400-2500nm band. The molecular-level material fingerprint of the fabric was extracted from the band data by a deep convolutional neural network and used as the detection spectral parameter.

[0040] Transmission detection information refers to the time-domain waveform and frequency-domain spectrum of terahertz waves after penetrating the fabric being tested. The parameters obtained by performing a transmission scan on the fabric using a terahertz sensor are used as transmission detection information.

[0041] The detection image refers to the heat map of the defect distribution of the detected fabric. Each pixel in the detection image corresponds to the defect type (such as snags, color spots, weft skew) and severity at a certain geographic coordinate on the fabric. The system uses an autoencoder dimensionality reduction algorithm to eliminate the baseline drift of the hyperspectral imager caused by ambient light fluctuations, and then uses the detection spectral parameters to generate a defect distribution heat map with geographic coordinates in real time as the detection image.

[0042] In this embodiment, the detected image is automatically classified into tiny defects with a diameter of less than 0.1 mm through a multi-scale feature fusion network (FPN), and the defect position is geometrically aligned with the detected fabric in real time using a global coordinate mapping algorithm.

[0043] S13: Combine the detection image with the transmission detection information to obtain the abnormal parameters and material discharge traction parameters.

[0044] Anomaly parameters refer to the abnormal defect features found in the detected fabric, including the coordinates of surface defects (snagging, color spots, weft skew), the location of hidden broken needles or hard impurities, the area of ​​endogenous physical damage caused by uneven distribution of weaving stress, and the abrupt change points of local elastic modulus of the fabric. By automatically classifying the detected image using a semantic segmentation network (such as FPN), the coordinates of snags (diameter <0.1mm), color spots, weft skew angles (accuracy better than 0.1°), oil stains, etc., are extracted. Then, a one-dimensional convolutional neural network (1D-CNN) is used to identify the transmission detection information to obtain the endogenous physical damage caused by broken needles, hard impurities, and uneven distribution of weaving stress in the detected fabric. The coordinates and endogenous physical damage are combined to form anomaly parameters.

[0045] The feed traction parameters refer to the command parameters used to correct the direction of yarn arrangement on the rollers. Using a dynamic detection model based on a Convolutional Long Short-Term Memory (ConvLSTM) network, the system predicts the weft skew trend between consecutive detection frames, outputting a real-time weft skew angle (with an accuracy better than 0.1°). When the real-time weft skew angle exceeds a preset threshold (e.g., 0.5°), the system automatically calculates compensation vectors (Δx, Δy, Δrotation angle) as the feed traction parameters. Δx detects the change in lateral movement of the fabric on the rollers, and Δy detects the change in vertical movement of the fabric on the rollers.

[0046] S14: Generate a fabric layout plan based on order information, fabric inspection, and abnormal parameters, and then lay out the fabric using the fabric layout plan and the material laying traction parameters.

[0047] Fabric layout scheme refers to a two-dimensional layout diagram that combines and nests cut pieces from multiple orders across orders. The fabric layout scheme is obtained by analyzing order information, inspected fabrics, and abnormal parameters. The fabric is then laid out according to the fabric layout scheme, and the laying-out traction parameters are used to compensate for the contour of the subsequent cutting, so that the cut contour is aligned with the yarn direction after the weft skew.

[0048] S15: After completing the material layout, the detection processing parameters are obtained based on the detection image, the detection fabric and the fabric layout scheme, and the fabric is processed with the detection processing parameters to form the detection garment.

[0049] The processing parameters refer to the set of specific process parameters executed in each step of fabric spreading, cutting, handling, and sewing.

[0050] Inspection garments refer to finished garments that meet the order information requirements after processing. After the system completes the material layout, it analyzes the inspection images, inspection fabrics and fabric layout schemes to obtain inspection processing parameters, and then processes the fabrics according to the inspection processing parameters to form inspection garments.

[0051] S16: Combine the inspection of the garments with the order information to transport the garments.

[0052] After passing online inspection by the Transformer visual model, the garments are transported to the automated packaging station by conveyor belt or AGV (Automated Guided Vehicle). The delivery node and logistics address in the order information generate a delivery tag, which is associated with the RFID tag of the garment and used for outbound transportation. When the garment is scanned upon leaving the warehouse, the traceability data is uploaded to the system's blockchain. Consumers can obtain the manufacturing cycle information of the garment by scanning the RFID.

[0053] Methods for generating fabric layout schemes include: S140: Calculate the detection similarity based on the detected fabric.

[0054] Similarity detection refers to the degree of comprehensive matching between the current detected fabric and other fabrics (or leftovers) in the fabric library in terms of three-dimensional physical properties such as elastic modulus, weight, texture direction, and heat shrinkage rate. The value ranges from 0 to 1, and the closer it is to 1, the more suitable it is for joint layout.

[0055] The physical property vector is formed by retrieving the elastic modulus, weight, grain direction angle, and heat shrinkage rate from the tested fabric. Then, the elastic modulus, weight, grain direction angle, and heat shrinkage rate of any fabric are retrieved from the fabric library to form any vector. The similarity of the detected fabric is calculated by combining the weights determined by the entropy weight method (e.g., elastic modulus weight 0.4, weight 0.3, grain direction angle 0.2, heat shrinkage rate 0.1) with the allowable deviation scales set by the technicians for the corresponding properties (elastic modulus deviation 50MPa, weight deviation 5g / m², grain direction deviation 2°, heat shrinkage rate deviation 0.5%).

[0056] S141: The joint fabric is obtained by comparing the detection similarity with the preset benchmark similarity.

[0057] The baseline similarity is the minimum similarity threshold set by the technicians for joint material arrangement, usually set to 0.80.

[0058] Combined fabrics refer to other fabric rolls (or leftovers) whose detection similarity to the currently detected fabric exceeds the benchmark similarity. Combined fabrics are fabrics that have a detection similarity to the currently detected fabric that exceeds the benchmark similarity. Combined fabrics can be fabrics from the detected fabrics or other fabrics from the fabric library.

[0059] S142: Generate a fabric layout plan by combining the detected fabric, the combined fabric, and the order information.

[0060] By mapping the abnormal parameters (defect coordinates, intrinsic damage) of the detected fabric and the combined fabric to the fabric width coordinate system, tiny defects with a diameter <0.1mm, broken needles / impurities, and weft skew exceeding the tolerance are marked as cut piece forbidden areas. The cut piece geometric dimensions are retrieved from the order information to form the cutting path, and the fabric width template is formed based on the path.

[0061] The delivery priority is retrieved from the order information. The detected fabric and the combined fabric are combined according to the order delivery priority. With the optimization goals of maximizing fabric utilization, eliminating overlapping cut pieces, and ensuring compliance with texture direction, a heuristic genetic algorithm is used to generate an initial nested layout by combining the fabric size template and the combined fabric. The layout scheme corresponding to the optimal layout vector is formed by Monte Carlo tree search and iterative optimization as the fabric layout scheme.

[0062] S143: Obtain the remaining outline of the cut fabric based on the fabric layout plan and the inspected fabric, and generate a layout barcode based on the fabric layout plan and the remaining outline of the cut fabric and record it in the warehouse.

[0063] The residual outline of the cut fabric refers to the two-dimensional geometric outline of the fabric remaining after the fabric is cut according to the fabric layout scheme. Based on the coordinates of the cut pieces in the fabric layout scheme, the edge vectors of the residual area are extracted by the visual outline reconstruction algorithm to form the corresponding outline, and the area within the outline is calculated. Then, the outline with an area exceeding the preset area threshold is selected as the residual outline of the cut fabric.

[0064] The layout barcode refers to the digital encoding (QR code / barcode) of the associated detection fabric, joint fabric, cut residual outline, and fabric layout scheme. The feature vector of the cut residual outline, detection spectral parameters, joint fabric, and unique ID of the layout scheme are encrypted to generate a digital barcode as the layout barcode, and the layout barcode is bound to the database record.

[0065] Reference Figure 3 Methods for obtaining detection processing parameters include: S20: Retrieve the layout fabric and its weight from the fabric layout scheme, and extract the elastic modulus parameter of the layout fabric from the detected spectral parameters.

[0066] Fabric layout refers to the test fabric and combined fabric cut in the current batch in the fabric layout plan. Layout weight refers to the weight value of the layout fabric. The layout weight is calculated by retrieving the layout fabric from the fabric layout plan and then retrieving the fabric weight (accuracy 1g / m²) from the test spectral parameters and multiplying it by the total area of ​​the fabric width in the layout plan.

[0067] The elastic modulus parameter refers to the physical elastic property parameter of the fabric. It is obtained by extracting the elastic modulus parameter of the fabric from the measured spectral parameters.

[0068] S21: Obtain air cushion control parameters based on the weight of the arrangement.

[0069] Air cushion control parameters refer to the power of the airflow required to form an air cushion layer for the smooth laying of auxiliary fabric. Based on the weight of the fabric, the air pressure, airflow rate, and vector angle control parameters of each zone of the fabric are calculated using a proportional-integral (PI) control algorithm and used as air cushion control parameters.

[0070] In this embodiment, a micro-electromechanical system (MEMS) is used to control the micro-aperture nozzle to operate with air cushion control parameters to form an air cushion layer with a height stable in the range of 0.5 mm to 2.0 mm.

[0071] S22: Collect the fabric laying speed, electrostatic potential gradient, and fabric edge image, and calculate the difference between the fabric laying speed and the preset fabric feeding roller speed as the speed deviation value.

[0072] Fabric placement speed refers to the actual dynamic movement speed of the fabric on the air cushion layer. The dynamic movement speed of the fabric on the air cushion layer is collected in real time by a laser Doppler velocimeter as the fabric placement speed.

[0073] The electrostatic potential gradient refers to the gradient value of the electrostatic charge distribution on the surface of the fabric. The electrostatic potential gradient is a parameter detected in real time by the electrostatic sensor of the preset charge balance device.

[0074] Fabric edge image refers to a real-time image of the edge of the fabric layup, which is captured by a preset edge line array CCD camera.

[0075] The fabric feeding roller speed is the preset reference operating speed of the fabric feeding roller used for pulling the fabric, set by the technician.

[0076] The speed deviation value refers to the deviation between the fabric laying speed and the fabric feeding roller speed. It is calculated as the difference between the fabric laying speed and the fabric feeding roller speed.

[0077] S23: Update the fabric feeding roller speed based on the speed deviation value.

[0078] The correction coefficient is matched with the speed deviation value from the preset garment reference table. The product of the correction coefficient and the feed roller speed is calculated to obtain the new feed roller speed, ensuring that the longitudinal stretch rate of the fabric is less than 0.02%.

[0079] The garment reference table stores correction coefficients corresponding to different speed deviation values. The correction coefficient ranges from 0 to 1. The larger the speed deviation value, the smaller the correction coefficient, and the slower the fabric feeding roller speed. The parameters in the garment reference table are set in advance by those skilled in the art based on actual conditions through experiments.

[0080] S24: The neutralization and ionization parameters are obtained through the electrostatic potential gradient, and the cutting and resting time is calculated based on the elastic modulus parameter.

[0081] Neutralization and ionization parameters refer to the intensity and ionization time parameters of the ionization gas flow output by the charge balance device. The neutralization and ionization parameters are matched from the clothing reference table by the electrostatic potential gradient.

[0082] The clothing reference table stores the neutralization and ionization parameters corresponding to different electrostatic potential gradients. The larger the electrostatic potential gradient, the larger the neutralization and ionization parameter.

[0083] The cutting resting time refers to the optimal static relaxation time for the fabric stress to be released after the fabric is laid out. The time length value obtained by automatically calculating the stress release timer based on the elastic modulus parameter is used as the cutting resting time.

[0084] S25: Identify edge features from fabric edge images.

[0085] Edge features refer to the geometric features of the fabric edge, such as its contour, straightness, and alignment. An algorithm combining the Sobel operator and Hough transform is used to extract the geometric features of the fabric edge, such as its contour, straightness, and alignment, from the fabric edge image as edge features.

[0086] S26: Compare the edge features with the preset baseline features to obtain the contrast, and update the fabric edge image with the contrast.

[0087] The baseline feature is the standard geometric feature of the fabric layup edge set by the technician.

[0088] Contrast refers to the geometric difference between edge features and reference features. The contrast is calculated by using a machine vision feature matching algorithm and combining the geometric difference between edge features and reference features. The background backlight intensity is dynamically adjusted according to the contrast to obtain a clear fabric edge image after enhancing the edge features.

[0089] S27: Obtain the uniformity based on the fabric edge image and edge features, and compare the uniformity with the preset baseline uniformity to obtain the fabric conveying direction.

[0090] Neatness refers to the comprehensive index of the alignment accuracy and flatness of the edges of multi-layered fabrics. It is calculated by integrating the edge alignment error and flatness from the updated fabric edge image and edge features.

[0091] The baseline neatness is the acceptable threshold for the alignment of the fabric edges, set by the technicians.

[0092] The fabric conveying direction refers to the direction in which the fabric is conveyed to correct edge alignment errors. The difference between the neatness and the baseline neatness is calculated as the deviation value, and the deviation value is input into the garment reference table to match the fabric conveying direction.

[0093] The garment reference table stores the fabric conveying direction corresponding to different deviation values. The larger the deviation value, the greater the angle of the fabric conveying direction that needs to be corrected.

[0094] S28: Integrate the air cushion control parameters, fabric feeding roller speed, neutralization and ionization parameters, cutting and resting time, and fabric conveying direction into fabric laying parameters, and combine the fabric laying parameters with the detection image to obtain the detection and processing parameters.

[0095] Fabric laying parameters refer to the set of control parameters for the entire process of zero-tension fabric laying performed by the adaptive flexible fabric laying module. By integrating air cushion control parameters, fabric laying roller speed, neutralization and ionization parameters, cutting and resting time, and fabric conveying direction into fabric laying parameters, and analyzing the fabric laying parameters and detection images, the detection and processing parameters are obtained.

[0096] Methods for obtaining detection processing parameters also include: S30: Collects the air resistance of the fabric and the acoustic emission signals of the cutting blade.

[0097] The vacuum suction area is a zone set up by technicians to test the fabric's breathability and fit. The vacuum suction area is equipped with an array of vacuum suction devices, and these devices are equipped with pressure sensors to collect airflow resistance data.

[0098] Fabric air resistance refers to the airflow resistance value of the fabric in the vacuum suction area. The parameter detected by the pressure sensor set in the vacuum suction device is used as the fabric air resistance.

[0099] Acoustic emission signal refers to the frequency band vibration waveform signal (20kHz~200kHz) generated when the cutting knife cuts the fabric fibers. The frequency band vibration waveform signal during the cutting process is captured in real time by a preset ultrasonic sensor as the acoustic emission signal.

[0100] S31: Combining the air resistance of the fabric surface with the fabric arrangement to obtain negative pressure strength.

[0101] Negative pressure intensity refers to the negative pressure value of each zone in an array-type vacuum zone suction device. The negative pressure intensity of each zone is dynamically obtained by using reinforcement learning algorithms based on the air resistance of the fabric and the fabric arrangement.

[0102] S32: Obtain the initial cutting path based on the fabric layout scheme and the layout fabric.

[0103] The initial cutting path refers to the two-dimensional cut piece outline vector path corresponding to the fabric layout scheme. It is obtained by extracting the cut piece outline paths of the fabric layout scheme as the initial cutting path. In this embodiment, this initial cutting path does not undergo deformation or defect avoidance compensation.

[0104] S33: Compensate the initial cropping path based on the detected image and abnormal parameters to obtain the detection cropping path.

[0105] The inspection cutting path refers to the final cutting path after non-rigid deformation compensation and defect avoidance.

[0106] The TPS thin plate spline function, RGB-D three-dimensional point cloud mapping and A* pathfinding algorithm are used. The path obtained by combining the detected image and abnormal parameters (defect coordinates) for defect avoidance and deformation compensation is used as the detection clipping path.

[0107] The TPS thin plate spline function is: f(p) = Ap + ∑w i U(‖pp i ‖), A is the affine coefficient matrix, w i For the radial basis function weights, U(r) = r 2 lnr is the kernel function.

[0108] S34: Obtain the bounding box parameters of the cutting tool by detecting the cutting path and the preset cutting tool model.

[0109] The cutting blade model is a three-dimensional model of the cutting blade's geometric dimensions, range of motion, and blade spacing, as defined by technicians.

[0110] Bounding box parameters refer to the collision safety boundary coordinates, spacing, and movement range parameters during multi-head asynchronous cutting. The collision safety boundary coordinates, spacing, and movement range parameters obtained by calculating and detecting the cutting path and the cutting blade model through analytical geometry model are used as bounding box parameters.

[0111] S35: Identify high-frequency characteristic spectra from acoustic emission signals and obtain the wear level based on the high-frequency characteristic spectra and a preset reference spectrum signal.

[0112] The reference spectrum signal is a standard acoustic emission spectrum signal template set by technicians for cutting fabric with a normally worn cutter.

[0113] The high-frequency characteristic spectrum refers to the wear characteristic spectrum of the cutting blade when it actually cuts the fabric. The 20kHz to 200kHz frequency band of the cutting blade wear characteristic spectrum is extracted by Fourier transform and wavelet packet decomposition of the acoustic emission signal as the high-frequency characteristic spectrum.

[0114] Wear level refers to the degree of wear of the cutting blade. It is obtained by calculating the Euclidean distance between the high-frequency feature spectrum and the reference spectrum, and combining it with a deep learning classification model.

[0115] S36: Obtain a correction strategy based on the wear level.

[0116] Correction strategies refer to control strategies for cutter wear, including speed reduction, online sharpening, cutter change warning, and depth compensation. Correction strategies are matched from the clothing comparison table based on the wear level.

[0117] The equipment comparison table stores the correction strategies corresponding to different wear levels. For example: slight wear = speed reduction + depth compensation. severe wear = sharpening / replacing the tool.

[0118] S37: Integrate negative pressure strength, detection cutting path, bounding box parameters, and correction strategies into cutting parameters, and use the cutting parameters and fabric laying parameters as detection processing parameters.

[0119] Cutting parameters refer to the set of control parameters for the entire process of high-precision cutting performed by the non-rigid cutting module. The cutting parameters are obtained by measuring negative pressure strength, detecting the cutting path, bounding box parameters, and correction strategies. The cutting parameters and fabric laying parameters are then used as detection and processing parameters.

[0120] Other methods for obtaining clipping parameters include: S40: Obtain the cutting area based on the detected cutting path.

[0121] The cutting area refers to the two-dimensional plane area of ​​the cut piece enclosed by the detected cutting path. The cutting area is the parameter obtained by performing area integration on the vector contour of the detected cutting path.

[0122] S41: Obtain the cut piece weight and reference pressure value based on the cutting area and fabric layout.

[0123] Cut piece weight refers to the actual weight of a single cut piece, which is calculated by multiplying the cutting area by the fabric weight (1g / m²).

[0124] The baseline pressure value refers to the standard pressure threshold at which the CNT / PDMS flexible tactile layer grips a single-layer fabric piece. The baseline pressure value is determined by matching the fabric arrangement with the garment reference table.

[0125] The clothing reference table stores the baseline pressure values ​​corresponding to different fabric layouts.

[0126] S42: Obtain electrostatic power based on the weight of the cut piece.

[0127] Electrostatic power refers to the electrostatic excitation power used to attract cut pieces. The electrostatic power is matched from the garment reference table based on the weight of the cut pieces.

[0128] The garment comparison chart stores the electrostatic power corresponding to different fabric weights; the greater the fabric weight, the greater the electrostatic power.

[0129] S43: Use electrostatic power to adsorb the cut piece and collect the gripping pressure value.

[0130] The gripping pressure value refers to the actual pressure value of the CNT / PDMS flexible tactile layer gripping the cut piece. The cut piece is adsorbed by the preset alternating electrostatic adsorption operation controlled by electrostatic power, and the flexible tactile layer grips the cut piece. The parameters detected by the pressure sensor in the flexible tactile layer are then used as the gripping pressure value.

[0131] S44: Compare the captured pressure value with the reference pressure value to calculate the pressure deviation value and obtain the number of detection layers.

[0132] The pressure deviation value refers to the deviation between the gripping pressure value and the reference pressure value. By analyzing the consistency between the gripping pressure value and the reference pressure value, when the gripping pressure value is consistent with the reference pressure value, the gripped piece is only one layer, and no adjustment is made.

[0133] The number of detection layers refers to the number of layers of fabric pieces that the flexible tactile layer actually grasps. When the grasping pressure value is inconsistent with the reference pressure value, there are multiple layers of fabric pieces being grasped. The difference between the grasping pressure value and the reference pressure value is calculated as the pressure deviation value, and the number of detection layers is matched from the garment comparison table based on the pressure deviation value and the reference pressure value.

[0134] The clothing reference table stores the number of detection layers corresponding to different pressure deviation values ​​and reference pressure values. With the reference pressure value remaining unchanged, the larger the pressure deviation value, the larger the number of detection layers.

[0135] S45: Compare the number of detection layers with the preset reference number of layers and combine the pressure deviation value to obtain the pulse airflow parameters.

[0136] The baseline layer number is the number of layers used to grip the cut piece according to the standard for flexible tactile layer set by the technicians.

[0137] The pulse airflow parameter refers to the airflow parameter required for peeling off multi-layer fabric pieces. The difference between the number of detected layers and the reference number of layers is calculated as the layer deviation value. Then, the pulse airflow parameter is matched from the garment reference table using the layer deviation value and the pressure deviation value.

[0138] The clothing reference table stores pulse airflow parameters corresponding to different layer deviation values ​​and pressure deviation values. The larger the layer deviation value and pressure deviation value, the larger the pulse airflow parameter (Bernoulli effect).

[0139] S46: Run with pulsed airflow parameters to update the number of detection layers until the number of detection layers matches the reference number, and add electrostatic power and pulsed airflow parameters to the trimming parameters.

[0140] The micro-blowing peeling device operates with pulse airflow parameters and re-obtains the number of detected layers. When the updated number of detected layers matches the reference number, the micro-blowing peeling device stops operating, and the electrostatic power and pulse airflow parameters are added to the cutting parameters.

[0141] Methods for obtaining detection processing parameters also include: S50: After cutting is completed, the preset transport robot is controlled to collect the cut pieces and collect environmental detection information.

[0142] Transport robots are automated guided vehicles (AGVs) designed by technicians for transferring pieces between cutting processes.

[0143] Environmental monitoring information refers to environmental data such as the location of obstacles, operator motion vectors, channel width, and production line congestion status in the travel path of the transport robot. This data is collected in real time by the LiDAR, RGB-D depth camera, and V2X communication module carried by the transport robot.

[0144] S51: Define the currently collected transport robot as a marked robot, and treat the remaining transport robots as other robots.

[0145] A marking robot is a transport robot that is currently performing the task of collecting and transferring cut pieces. The transport robot currently performing the task of collecting cut pieces is defined as a marking robot.

[0146] Other robots refer to other transport robots that are in other task states, defined by transport robots that are in operation other than the marking robot.

[0147] S52: Retrieve other operating parameters from other robots.

[0148] Other operating parameters refer to the real-time position, remaining battery power, driving status, load status, and communication status parameters of other robots, which are retrieved from other robots.

[0149] S53: Obtain the detection operation vector based on environmental monitoring information and other operating parameters.

[0150] The detection running vector refers to the motion vector parameters of other robots and surrounding human operators. It is formed by retrieving the position, direction of movement, and speed of the human from the environmental detection information and combining them with other running parameters to form the motion vector parameters as the detection running vector.

[0151] S54: Obtain the marked driving strategy by detecting the running vector, and collect the workstation detection values ​​and order qualification rate.

[0152] The marking driving strategy refers to the dynamic obstacle avoidance, yielding and detour, speed adjustment and path optimization driving scheme of the marking robot. It adopts the dynamic driving trajectory with the shortest path and the best energy efficiency obtained by improving the dynamic potential field algorithm and social force model and integrating the detection running vector. When encountering human operators, it implements active deceleration or yielding and detour strategies as the marking driving strategy.

[0153] The workstation detection value refers to the real-time load rate, material waiting queue length, and equipment operating status parameters of the sewing workstation. It is used to express the fatigue level or real-time demand level of the workers at the workstation. The workstation detection value is obtained by retrieving the real-time load rate, material waiting queue length, and equipment operating status parameters from the system and performing weight calculation. The larger the workstation detection value (the smaller the real-time demand level), the more fatigued the workers at that workstation are.

[0154] Order qualification rate refers to the historical production qualification rate of a sewing station. It is calculated by retrieving the historical qualified garment quantity of that station from the system and dividing the historical qualified garment quantity by the total quantity.

[0155] S55: Combine the workstation inspection values ​​with the order qualification rate to obtain the cutting piece sorting order.

[0156] The cutting piece sorting sequence refers to the order in which cutting pieces that need to be sewn are transported. Based on the sorting strategy of deep reinforcement learning (DRL), the order in which cutting pieces are transported from each workstation is calculated by integrating the workstation detection values ​​and the order qualification rate as the cutting piece sorting sequence.

[0157] S56: Integrate the marking driving strategy and the piece sorting sequence into transportation parameters, and add the transportation parameters to the inspection and processing parameters.

[0158] Transportation parameters refer to the set of control parameters for the entire process of fabric transfer, sorting, and obstacle avoidance by the collaborative handling module. By integrating the marking driving strategy with the fabric sorting sequence into transportation parameters, and adding transportation parameters to the detection and processing parameters.

[0159] Other methods for obtaining transportation parameters include: S60: Collects the battery level and location of the marker robot.

[0160] The marked battery level refers to the real-time remaining battery percentage of the marked robot, which is obtained by using the remaining battery percentage collected in real time by the marked robot's battery management system (BMS).

[0161] The marked location refers to the real-time coordinate position of the robot, which is determined in real time by the self-built SLAM semantic map module of the transport robot.

[0162] S61: Compare the marked power level with the preset baseline power level to collect backup location data.

[0163] The baseline power level is the minimum power threshold (20%) set by technicians for the task to be performed normally.

[0164] The standby position refers to the position of the transport robot in standby mode. By analyzing the difference between the marked power level and the baseline power level, if the marked power level is greater than the baseline power level, it means that the marked robot can perform the task normally, and no adjustment is made.

[0165] When the marked battery level is not greater than the baseline battery level, it means that the marking robot cannot continue to perform the task. In this case, refer to S60 to retrieve the real-time coordinate position of the transport robot in standby mode as the backup position.

[0166] S62: Obtain the alternative path by marking the location and the alternative location.

[0167] Backup paths refer to the various travel paths that the backup robot takes to reach the marked robot's location. Based on the digital twin map, the various paths between the marked location and the backup location are calculated as backup paths.

[0168] S63: Select the backup robot corresponding to the backup path with the shortest distance as the target robot.

[0169] The target robot is the backup robot with the shortest alternative path distance. By retrieving the required travel distance for each path, the backup robot corresponding to the alternative path with the shortest distance is selected as the target robot.

[0170] S64: Control the target robot to move to the marked position to continue executing the marked driving strategy and the piece sorting sequence, and add the marked position and the target robot to the transportation parameters.

[0171] Control the target robot to move to the marked position to continue executing the remaining marking driving strategy and piece sorting sequence of the marking robot, and add the marked position and target robot to the transportation parameters.

[0172] Methods for obtaining detection processing parameters also include: S70: Based on transport parameters, acquire sewing images of the cut pieces after transport and needle bar resistance.

[0173] Sewing images refer to real-time images of needle tip movement, thread loop shape, and fabric alignment during the sewing process of fabric pieces. These images are captured in real-time by a 1200fps high-speed visual monitoring module.

[0174] Needle bar resistance refers to the real-time resistance value when a sewing needle penetrates a piece of fabric. The penetration resistance data is collected in real time by a preset needle bar resistance sensor and used as the needle bar resistance.

[0175] S71: Identify sewing abnormal parameters through sewing images.

[0176] Sewing anomaly parameters refer to parameters that indicate abnormal features during the sewing process, such as skipped stitches, thread loop misalignment, thread take-up delay, and skewed stitches. These parameters, which are extracted from sewing images through semantic recognition using the Feature Pyramid Network (FPN), are used as sewing anomaly parameters.

[0177] S72: Retrieve the fabric thickness corresponding to the sewing image from the transport parameters.

[0178] Cutting piece thickness refers to the thickness value of the current sewing cutting piece, which is obtained by retrieving the cutting piece thickness from the fabric detection spectral parameters bound to the transportation parameters.

[0179] S73: Predicts thread tension based on fabric thickness, sewing image, needle bar resistance, and preset sewing speed.

[0180] The sewing speed is the standard sewing speed set by the technicians.

[0181] Predicted thread tension refers to the predicted tension value of the sewing thread during the sewing process. It utilizes a Physical Information Neural Network (PINN) to fuse sewing images and needle bar resistance in real time. PINN explicitly introduces a sewing dynamics physical constraint term into its loss function, and its normalized constraint equation is: Lphysics = ||(T / T0)−φ(H / H0,V / V0)||², where T is the thread tension, H is the fabric thickness, V is the sewing speed, T0, H0, and V0 are the baseline normalized reference values ​​for each quantity, φ(·) is the polynomial mapping function fitted by the sensor data, and the ratios of the three terms are all dimensionless numbers.

[0182] Based on the above-predicted tension change pattern, the future tension value is obtained as the predicted line tension.

[0183] S74: Compare the predicted thread tension with the preset baseline thread tension and combine the preset baseline operating parameters with the sewing abnormal parameters to obtain the sewing operating parameters.

[0184] The baseline tension is the optimal baseline tension value of the sewing thread under standard working conditions set by the technician.

[0185] The reference operating parameters are the standard operating parameters of the sewing device, such as the reference spindle speed, the initial force of the electronic thread clamp, and the fabric feed ratio, set by the technicians.

[0186] Sewing operation parameters refer to the operating parameters of the sewing device after tension compensation and anomaly correction. The tension change value is calculated by comparing the predicted thread tension with the baseline thread tension. Based on the tension change value, the baseline operating parameters, and the sewing anomaly parameters, the sewing operation parameters are matched from the garment comparison table.

[0187] The garment reference table stores the sewing operating parameters corresponding to different tension change values, baseline operating parameters, and sewing abnormal parameters. When the baseline operating parameters and sewing abnormal parameters remain unchanged, the larger the tension change value, the larger the difference between the sewing operating parameters and the baseline operating parameters.

[0188] S75: Add the sewing operation parameters to the detection and processing parameters.

[0189] By incorporating sewing operation parameters into the detection processing parameters.

[0190] Methods for obtaining sewing operation parameters include: S80: Collects temperature detection information and presser foot pressure value.

[0191] Temperature detection information refers to the real-time temperature value at the contact point between the sewing needle and the fabric piece. This temperature detection information can be obtained by monitoring the temperature of the sewing needle or fabric piece in real time using a non-contact infrared temperature sensor.

[0192] The presser foot pressure value refers to the real-time pressure value of the sewing presser foot on the fabric piece. It is the pressure value detected by the pressure sensor preset on the sewing presser foot.

[0193] S81: Compare the temperature detection information with the preset reference temperature information to obtain the cooling spray parameters and re-collect the temperature detection information until the temperature detection information is lower than the reference temperature information.

[0194] The reference temperature information is the fabric melting point safety warning temperature set by the technicians.

[0195] Cooling spray parameters refer to the operating parameters of the cooling spray required to cool the temperature between the sewing needle and the fabric piece. By analyzing the deviation between the temperature detection information and the reference temperature information, when the temperature detection information exceeds the reference temperature information, the difference between the temperature detection information and the reference temperature information is calculated as the temperature deviation value. Based on the temperature deviation value, the cooling spray parameters are matched from the garment reference table, and the preset atomizing cooling device is controlled by the cooling spray parameters. The temperature detection information is re-acquired until the temperature detection information is lower than the reference temperature information, at which point the atomizing cooling device stops operating.

[0196] The clothing reference table stores the cooling spray parameters corresponding to different temperature deviation values; the larger the temperature deviation value, the larger the cooling spray parameter.

[0197] S82: The cut piece specifications corresponding to the sewing image are retrieved from the transportation parameters.

[0198] Cutting piece specifications refer to the size, number of layers, material, thickness and position of the cutting piece. These specifications are obtained by retrieving them from the fabric layout scheme that the cutting piece is bound to by the transportation parameters.

[0199] S83: Obtain the reference presser foot pressure based on the cut piece specifications.

[0200] The reference presser foot pressure refers to the standard presser foot pressure value that is adapted to the current pattern size. The reference presser foot pressure is matched from the garment comparison table based on the pattern size.

[0201] The garment reference chart stores the standard presser foot pressure corresponding to different pattern piece sizes.

[0202] S84: Compare the pressure foot pressure value with the reference pressure foot pressure to calculate the pressure deviation value.

[0203] The pressure deviation value refers to the deviation between the presser foot pressure value and the reference presser foot pressure. By analyzing the consistency between the presser foot pressure value and the reference presser foot pressure, when the presser foot pressure value is inconsistent with the reference presser foot pressure, the difference between the presser foot pressure value and the reference presser foot pressure is calculated as the pressure deviation value.

[0204] S85: Update the sewing operation parameters based on the pressure deviation value and add the cooling spray parameters to the sewing operation parameters.

[0205] By matching the pressure deviation value with the differential feed ratio in the sewing operation parameters from the garment reference table, a new differential feed ratio is obtained to replace the original differential feed ratio in the sewing operation parameters, thus obtaining new sewing operation parameters. The cooling spray parameter is then added to the sewing operation parameters.

[0206] In this embodiment, the system collects the operating vibration, energy efficiency parameters and work-in-process (WIP) location of each device in real time, constructs a 1:1 dynamic mirror model in the cloud virtual space, and uses a streaming big data analysis engine to perform real-time production simulation. It can issue early warnings before equipment failure or production bottlenecks occur, and supports parameter fine-tuning and process command issuance to remote physical devices through augmented reality (AR) interface, forming a transparent management and control closed loop that blends the virtual and real worlds.

[0207] An AI-driven rapid-response garment manufacturing method further includes the following steps: S100: Collects total operating parameters.

[0208] Total operating parameters refer to a comprehensive set of operational data for all equipment used in garment manufacturing, including vibration, motor current harmonics, energy efficiency parameters, belt tension, work-in-process (WIP) position, and equipment speed. This data is collected in real-time by vibration sensors, current sensors, energy efficiency monitoring modules, and positioning modules, and then aggregated via an industrial bus system.

[0209] S101: Form a detection operation model based on the total operating parameters.

[0210] The detection operation model refers to a real-time digital twin mirror model of all equipment used in garment manufacturing during operation. By inputting the total operating parameters into the cloud-based digital twin platform, a 1:1 real-time virtual equipment model is generated through streaming rendering as the detection operation model.

[0211] S102: The baseline operating model is obtained based on the fabric layout scheme and the total operating parameters.

[0212] The benchmark operating model refers to a fault-free digital twin benchmark model under standard operating conditions. It is a real-time virtual equipment model generated based on the fabric layout scheme, process standard presets, and historical best operating parameters, and serves as the benchmark operating model.

[0213] S103: Compare the consistency between the detection operation model and the benchmark operation model to obtain fault parameters.

[0214] Fault parameters refer to the difference features between the detection operation model and the benchmark operation model, including fault features such as belt slack, excessive vibration, and abnormal current. By comparing the real-time model and the benchmark model through a Bayesian network, the abnormal feature vectors are extracted to obtain the fault parameters.

[0215] S104: When the fault parameters are the preset abnormal belt characteristics, the abnormal tightness is obtained based on the fault parameters and the abnormal belt characteristics.

[0216] The abnormal belt feature is an abnormal characteristic of belt slack in the adaptive flexible fabric spreading module set by the technicians.

[0217] Abnormal tightness refers to the tightness of the belt in the adaptive flexible fabric spreading module. When the fault parameter is abnormal belt characteristics, it indicates that the adaptive flexible fabric spreading module has an abnormal situation of belt loosening. The abnormal tightness is calculated by fitting the belt vibration and current characteristics in the fault parameters.

[0218] S105: Calculate the difference between the abnormal tightness and the preset baseline tightness as the supplementary tightness.

[0219] The reference tension is the optimal tension threshold set by the technician for normal belt operation.

[0220] Supplemental tension refers to the extra tension required for the belt to return to its baseline tension. It is calculated by taking the difference between the abnormal tension and the baseline tension.

[0221] S106: Compressive force is obtained by adjusting the tightness and the preset belt specifications.

[0222] The belt specifications are the parameters such as model, width, length, and elasticity coefficient of the transmission belt set by the technicians.

[0223] The compressive force refers to the pressure value required to tighten the tensioner or belt. The compressive force is determined by matching the tightness and belt specifications from a preset belt reference table.

[0224] The belt reference table stores the compression force corresponding to different supplementary tension and belt specifications. With the belt specifications unchanged, the greater the supplementary tension, the greater the compression force. The parameters in the belt reference table are set in advance by those skilled in the art based on actual conditions through experiments.

[0225] S107: The inflation pressure is obtained based on the airflow control parameters and the preset airbag specifications.

[0226] The airbag specifications are defined by the technicians for the volume, micropore diameter, pressure bearing value, and tension stroke parameters of the porous airbag (porous structure, which can overflow gas to form an air cushion layer).

[0227] Inflation pressure refers to the air pressure at which the airbag outputs gas to form the air cushion layer. The inflation pressure is determined by matching the airflow control parameters with the preset airbag specifications from a preset airbag reference table.

[0228] The airbag reference table stores the gas volume and inflation pressure corresponding to different airflow control parameters and airbag specifications. When the airbag specifications remain unchanged, the larger the gas volume of the airflow control parameters, the greater the inflation pressure. The parameters in the airbag reference table are set in advance by those skilled in the art based on actual conditions through experiments.

[0229] S108: The position of the airbag is obtained by combining the squeezing force and the inflation pressure, and the operation of the airbag is controlled by the position of the airbag and the output air flow.

[0230] The airbag position refers to the point where the airbag needs to move to exert a squeezing force on the belt when the airbag is inflated. The normal distance that the belt needs to be squeezed under the reference tension is matched from the belt reference table by the squeezing force. Then, the radius of the reaction force on the airbag is calculated according to the squeezing force and the airbag specifications. The airbag position is obtained by combining the radius and the normal distance and the point where the center of the airbag needs to move. The operation of the airbag is controlled by the airbag position and the output air flow.

[0231] Based on the same inventive concept, embodiments of the present invention provide an AI-driven rapid-response garment manufacturing system, comprising: The fabric multimodal quality inspection module is used to collect detection spectral parameters and transmission detection information, and to generate detection images using the detection spectral parameters. An adaptive flexible fabric laying module is used to receive and execute fabric laying parameters to lay the fabric. The non-rigid cutting module is used to receive and execute cutting parameters to cut the fabric. The collaborative handling module is used to receive and execute transportation parameters to transport the cut pieces formed after the fabric is cut, and after transportation, to receive and execute sewing operation parameters to sew the cut pieces into garments. The production scheduling module is used to collect data uploaded by the fabric multimodal quality inspection module, the adaptive flexible fabric spreading module, the non-rigid cutting module, and the collaborative handling module, and generate fabric laying parameters, transportation parameters, and sewing operation parameters and output them to the corresponding modules.

[0232] Based on the same inventive concept, embodiments of the present invention provide a smart terminal, including a memory and a processor, wherein the memory stores a computer program that can be loaded and executed by the processor to provide an AI-driven fast-response garment manufacturing method.

[0233] Based on the same inventive concept, this invention provides a digital twin real-time synchronization and predictive maintenance system for a garment fast-response factory. The system is characterized by collecting total operating parameters and constructing a detection and operation model, and then using the detection and operation model to perform real-time production simulation to obtain fault parameters for early warning.

[0234] Based on the same inventive concept, this invention provides an adaptive material layout and surplus material utilization optimization system for fragmented production, characterized in that it executes steps S140 to S143 to generate a fabric layout scheme and material layout barcode, and records the material layout barcode in the warehouse.

[0235] Based on the same inventive concept, this invention provides a zero-tension automatic fabric laying system based on fluid dynamics feedback, characterized in that it controls a preset set of micro-pore array nozzles to execute the air cushion control parameters in the fabric laying parameters S28.

[0236] Based on the same inventive concept, this invention provides an automated piece gripping system based on the fusion of electrostatic field and flexible tactile sensation, characterized in that it is used to execute the electrostatic power and pulse airflow parameters generated in S40 to S46.

[0237] Based on the same inventive concept, embodiments of the present invention provide a physical sensing online control system for sewing stitch shape, characterized in that it is used to execute S70 to S75 to generate predicted thread tension and execute sewing operation parameters for sewing compensation.

[0238] Based on the same inventive concept, this embodiment of the invention provides a microelectromechanical system based on a proportional-integral control algorithm, characterized in that it is used to control the air cushion control parameters of each nozzle in the micro-orifice array.

[0239] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device, and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0240] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. An AI-driven rapid-response garment manufacturing method, characterized in that, include: S10: Collect order information; S11: Obtain the inspected fabric based on order information; S12: Collect the detection spectral parameters and transmission detection information of the fabric to be tested, and form a detection image based on the detection spectral parameters; S13: Combine the detection image with the transmission detection information to obtain the abnormal parameters and the material discharge traction parameters; S14: Generate a fabric layout plan based on order information, inspected fabric, and abnormal parameters; then lay out the fabric using the fabric layout plan and the material laying traction parameters. S15: After completing the material layout, the detection processing parameters are obtained based on the detection image, the detection fabric and the fabric layout scheme, and the fabric is processed with the detection processing parameters to form the detection garment. S16: Combine the inspection and order information to transport the garments; Methods for generating fabric layout schemes include: S140: Calculate the detection similarity based on the detected fabric; S141: The joint fabric is obtained by comparing the detection similarity with the preset benchmark similarity. S142: Generate a fabric layout plan by combining the tested fabric, combined fabric, and order information; S143: Obtain the remaining outline of the cut fabric based on the fabric layout plan and the inspected fabric, and generate a layout barcode based on the fabric layout plan and the remaining outline of the cut fabric and record it in the warehouse.

2. The AI-driven rapid-response garment manufacturing method according to claim 1, characterized in that, Methods for obtaining detection processing parameters include: S20: Retrieve the layout fabric and the layout weight of the layout fabric from the fabric layout scheme, and extract the elastic modulus parameter of the layout fabric from the detection spectral parameters. S21: Obtain air cushion control parameters based on the weight arrangement; S22: Collect the fabric laying speed, electrostatic potential gradient, and fabric edge image, and calculate the difference between the fabric laying speed and the preset fabric feeding roller speed as the speed deviation value. S23: Update the fabric feeding roller speed based on the speed deviation value; S24: The neutralization and ionization parameters are obtained through the electrostatic potential gradient, and the cutting and resting time is calculated based on the elastic modulus parameter; S25: Identify edge features from fabric edge images; S26: Compare the edge features with the preset baseline features to obtain the contrast, and update the fabric edge image with the contrast. S27: Obtain the neatness based on the fabric edge image and edge features, and compare the neatness with the preset benchmark neatness to obtain the fabric conveying direction; S28: Integrate the air cushion control parameters, fabric feeding roller speed, neutralization and ionization parameters, cutting and resting time, and fabric conveying direction into fabric laying parameters, and combine the fabric laying parameters with the detection image to obtain the detection and processing parameters.

3. The AI-driven rapid-response garment manufacturing method according to claim 2, characterized in that, Methods for obtaining detection processing parameters also include: S30: Collects the air resistance of the fabric and the acoustic emission signals of the cutting knife; S31: Combining fabric air resistance with fabric layout to achieve negative pressure strength; S32: Obtain the initial cutting path based on the fabric layout scheme and the laid-out fabric; S33: Compensate the initial cropping path based on the detected image and abnormal parameters to obtain the detection cropping path; S34: Obtain the bounding box parameters of the cutting blade by detecting the cutting path and the preset cutting blade model; S35: Identify high-frequency characteristic spectra from acoustic emission signals and obtain the wear level based on the high-frequency characteristic spectra and a preset reference spectrum signal; S36: Obtain a correction strategy based on the wear level; S37: Integrate negative pressure strength, detection cutting path, bounding box parameters, and correction strategies into cutting parameters, and use the cutting parameters and fabric laying parameters as detection processing parameters.

4. The AI-driven rapid-response garment manufacturing method according to claim 3, characterized in that, Other methods for obtaining clipping parameters include: S40: Obtain the cutting area based on the detected cutting path; S41: Obtain the cut piece weight and reference pressure value based on the cutting area and fabric layout; S42: Calculate the electrostatic power based on the weight of the cut piece; S43: Use electrostatic power to adsorb the cut piece and collect the gripping pressure value; S44: Compare the gripping pressure value with the reference pressure value to calculate the pressure deviation value and obtain the number of detection layers; S45: Compare the number of detection layers with the preset reference number of layers and combine the pressure deviation value to obtain the pulse airflow parameters; S46: Run with pulsed airflow parameters to update the number of detection layers until the number of detection layers matches the reference number, and add electrostatic power and pulsed airflow parameters to the trimming parameters.

5. The AI-driven rapid-response garment manufacturing method according to claim 4, characterized in that, Methods for obtaining detection processing parameters also include: S50: After cutting is completed, control the preset transport robot to collect the cut pieces and collect environmental detection information; S51: Define the currently collected transport robot as a marked robot, and treat the remaining transport robots as other robots; S52: Retrieve other operating parameters from other robots; S53: Obtain the detection operation vector based on environmental monitoring information and other operating parameters; S54: Obtain the marked driving strategy by detecting the running vector, and collect the workstation detection values ​​and order pass rate; S55: Combine the workstation inspection values ​​with the order qualification rate to obtain the cutting piece sorting order; S56: Integrate the marking driving strategy and the piece sorting sequence into transportation parameters, and add the transportation parameters to the inspection and processing parameters.

6. The AI-driven rapid-response garment manufacturing method according to claim 5, characterized in that, Other methods for obtaining transportation parameters include: S60: Collects the battery level and location of the marker robot; S61: Compare the marked power level with the preset baseline power level to collect backup location data; S62: Obtain the alternative path by marking the location and the alternative location; S63: Select the backup robot corresponding to the backup path with the shortest distance as the target robot; S64: Control the target robot to move to the marked position to continue executing the marked driving strategy and the piece sorting sequence, and add the marked position and the target robot to the transportation parameters.

7. The AI-driven rapid-response garment manufacturing method according to claim 6, characterized in that, Methods for obtaining detection processing parameters also include: S70: Based on transport parameters, acquire sewing images of the cut pieces after transport and needle bar resistance; S71: Identify abnormal sewing parameters using sewing images; S72: Retrieve the fabric thickness corresponding to the sewing image from the transportation parameters; S73: Predicts thread tension based on fabric thickness, sewing image, needle bar resistance, and preset sewing speed; S74: Compare the predicted thread tension with the preset baseline thread tension and combine the preset baseline operating parameters with the sewing abnormal parameters to obtain the sewing operating parameters; S75: Add the sewing operation parameters to the detection and processing parameters.

8. The AI-driven rapid-response garment manufacturing method according to claim 7, characterized in that, Methods for obtaining sewing operation parameters include: S80: Collects temperature detection information and presser foot pressure value; S81: Compare the temperature detection information with the preset reference temperature information to obtain the cooling spray parameters and re-collect the temperature detection information until the temperature detection information is less than the reference temperature information. S82: Retrieve the cut piece specifications corresponding to the sewing image from the transportation parameters; S83: Obtain the reference presser foot pressure based on the cut piece specifications; S84: Compare the pressure foot pressure value with the reference pressure foot pressure to calculate the pressure deviation value; S85: Update the sewing operation parameters based on the pressure deviation value and add the cooling spray parameters to the sewing operation parameters.

9. An AI-driven rapid-response garment manufacturing system, characterized in that, include: The fabric multimodal quality inspection module is used to collect detection spectral parameters and transmission detection information, and to generate detection images using the detection spectral parameters. An adaptive flexible fabric laying module is used to receive and execute fabric laying parameters to lay the fabric. The non-rigid cutting module is used to receive and execute cutting parameters to cut the fabric. The collaborative handling module is used to receive and execute transportation parameters to transport the cut pieces formed after the fabric is cut, and after transportation, to receive and execute sewing operation parameters to sew the cut pieces into garments. The production scheduling module is used to collect data uploaded by the fabric multimodal quality inspection module, the adaptive flexible fabric spreading module, the non-rigid cutting module, and the collaborative handling module, and generate fabric laying parameters, transportation parameters, and sewing operation parameters and output them to the corresponding modules.

10. A smart terminal, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed as described in any one of claims 1 to 8, which is an AI-driven rapid-response garment manufacturing method.

11. A digital twin real-time synchronization and predictive maintenance system for a garment fast-response factory, characterized in that, By collecting total operating parameters and constructing a detection and operation model, and then using the detection and operation model to conduct real-time production simulation to obtain fault parameters for early warning.

12. An adaptive material feeding and waste material utilization optimization system for fragmented production, characterized in that, This is used to execute S140 to S143 to generate fabric layout schemes and layout barcodes, and to record the layout barcodes in the warehouse.

13. A zero-tension automatic fabric spreading system based on fluid dynamics feedback, characterized in that, This is used to control a preset set of micro-pore array nozzles to execute the air cushion control parameters in the S28 fabric laying parameters.

14. An automated piece-grabbing system based on the fusion of electrostatic field and flexible tactile sensation, characterized in that, Used to execute the electrostatic power and pulsed airflow parameters generated in S40 to S46.

15. A physical sensing online control system for sewing stitch patterns, characterized in that, Used to execute S70 to S75 to generate predicted thread tension and execute sewing operation parameters for sewing compensation.

16. A microelectromechanical system based on a proportional-integral control algorithm, characterized in that, Air cushion control parameters used to control the individual nozzles in the micropore array.

Images