Garment processing quality detection platform and method

Abstract

The invention relates to the field of garment quality detection, in particular to a garment processing quality detection platform and method. The method comprises the following steps: acquiring a double-sided garment multi-source data set, performing cross-process collaborative analysis on an interlayer fitting balance effect, an internal wrapping form and a three-dimensional ironing stress field based on the double-sided garment multi-source data set, and generating a garment process quality ground state information set; based on the clothing process quality ground state information set, establishing a defect transfer network to quantitatively evaluate the evolution path and contribution degree of potential defects, and generating a clothing comprehensive process offset index; and based on the garment comprehensive process offset index, performing adaptive feedforward control strategy mapping, and generating a double-sided garment processing detection report. In the garment quality detection process, the first-pass yield and the production consistency of the double-sided garment are integrally improved, the quality cost is reduced, and meanwhile continuous optimization of the production process is promoted through a digital report.

Description

Technical Field

[0001] This application relates to the field of garment quality testing, and in particular to a garment processing quality testing platform and method. Background Technology

[0002] In the field of intelligent manufacturing of apparel, the quality of garment manufacturing processes is a core element that determines product value, brand reputation and market competitiveness. Its precise control and assurance are of vital strategic significance for achieving high-quality development and intelligent transformation and upgrading of the textile and apparel industry.

[0003] However, existing garment quality inspection methods lack a mechanism for systematically and quantitatively assessing the evolution path of potential defects in each process and the contribution of cross-process transmission when faced with a process system consisting of multiple processes, various fabrics, and complex physical interactions. This makes it impossible to effectively locate the root process that causes the final double-sided appearance defects, leading to global quality risks for the entire batch of products and causing serious economic losses and waste of resources. Summary of the Invention

[0004] This application provides a garment processing quality inspection platform and method to solve the above-mentioned technical problems.

[0005] In a first aspect, this application provides a method for inspecting the quality of garment processing. The method includes: acquiring a multi-source dataset of reversible garments; based on the multi-source dataset of reversible garments, performing cross-process collaborative analysis on the interlayer bonding balance, internal wrapping morphology, and three-dimensional ironing stress field to generate a garment process quality baseline information set; based on the garment process quality baseline information set, constructing a defect transmission network to quantitatively assess the evolution path and contribution of potential defects to generate a garment comprehensive process deviation index; and based on the garment comprehensive process deviation index, performing adaptive feedforward control strategy mapping to generate a reversible garment processing inspection report.

[0006] The above technical solutions enable a shift in quality control from passive detection to proactive prevention. Through cross-process collaborative analysis, complex issues such as interlayer bonding abnormalities and shape deformation can be identified at an early stage. The defect transmission network accurately locates the root cause of defects, significantly reducing rework rates. Feedforward control based on the offset index enables adaptive adjustment of process parameters, effectively preventing defects from occurring. Overall, the first-pass yield and production consistency of double-sided garments are improved, reducing quality costs. At the same time, digital reporting promotes continuous optimization of production processes.

[0007] Optionally, the step of performing cross-process collaborative analysis on interlayer bonding balance, internal wrapping morphology, and three-dimensional ironing stress field based on the multi-source dataset of double-sided garments to generate a garment process quality base state information set includes: the multi-source dataset of double-sided garments includes interlayer dynamic tension difference, interlayer relative displacement, wrapping three-dimensional morphology, interlayer temperature difference, and interlayer pressure field distribution; based on the interlayer dynamic tension difference, analyzing the interlayer dynamic bonding effect during the sewing process to generate interlayer bonding balance effect information to characterize the double-sided bonding balance state at the seam; based on the interlayer relative displacement and the wrapping three-dimensional morphology, analyzing the internal wrapping morphology during the folding and wrapping process to generate internal wrapping morphology information characterizing the stable state of the inner wrapping structure; based on the interlayer temperature difference and the interlayer pressure field distribution, analyzing the double-sided stress field coupling relationship during the three-dimensional ironing process to generate three-dimensional ironing stress field information characterizing the stable state of the finished garment; and generating the garment process quality base state information set based on the interlayer dynamic interaction information, the internal wrapping morphology information, and the three-dimensional ironing stress field information.

[0008] Optionally, the step of analyzing the interlayer dynamic bonding effect during the sewing process based on the interlayer dynamic tension difference to generate interlayer bonding balance information for characterizing the double-sided bonding balance state at the seam includes: decoupling the interlayer dynamic tension difference to separate the static deviation component characterizing the uneven force on the two fabrics and the dynamic fluctuation component characterizing the asynchronous instantaneous fabric feeding during sewing; based on the static deviation component, quantifying the residual stress at the seam caused by the difference in shrinkage and stiffness during the sewing of heterogeneous fabrics, analyzing its symmetrical influence on the flatness of the front and back surfaces, and generating bidirectional... The flatness attenuation gradient is used to predict the stable range where the seam does not produce visible distortion on either side. Based on the dynamic fluctuation component, its frequency and amplitude characteristics are analyzed to detect the periodic pause and sudden advance phenomenon caused by the mismatch of the friction coefficients between the two fabrics and the presser foot / feed dog. The irreversible tensile deformation accumulated on one side of the fabric by this phenomenon is evaluated, and an interlayer slippage risk index is generated to associate the wavy seam defect that may occur on either side of the two sides. Based on the bidirectional flatness attenuation gradient and the interlayer slippage risk index, the interlayer bonding balance information is constructed.

[0009] Optionally, the step of analyzing the internal wrapping morphology during the folding process based on the interlayer relative displacement and the three-dimensional shape of the wrapping, and generating internal wrapping morphology information representing the stable state of the inner wrapping structure, includes: tracking and calculating the cumulative amount of the interlayer relative displacement; analyzing the asynchronicity and path deviation of the inner and outer fabrics in the three-dimensional folding motion; focusing on evaluating the mismatch of rebound amount after folding due to the difference in stiffness of the two fabrics; generating double-sided contour symmetry; performing point cloud reconstruction and surface curvature analysis on the three-dimensional shape of the wrapping; detecting local morphological bulges / collapses caused by the stacking of internal sutures and functional components in a limited space; quantifying the visibility and distribution range of these bulges / collapses on both outer surfaces; and generating internal component migration risk; and constructing internal wrapping morphology information to represent that the wrapping structure does not produce non-designed contours on either side, based on the double-sided contour symmetry and the internal component migration risk.

[0010] Optionally, the step of analyzing the coupling relationship of the two-sided stress field during the three-dimensional ironing process based on the interlayer temperature difference and the interlayer pressure field distribution, and generating three-dimensional ironing stress field information characterizing the steady state of the garment, includes: performing time-series analysis and gradient calculation on the interlayer temperature difference, solving the asymmetric conduction process when heat penetrates the garment interlayer composed of heterogeneous fabrics, quantifying the time difference and temperature difference between the front and back sides reaching the fiber glass transition temperature, and assessing the resulting risk of mismatch in two-sided heat setting memory, generating a two-sided heat memory uniformity; performing spatial vector decomposition on the interlayer pressure field distribution, identifying the uniform pressure distribution applied on the three-dimensional curved surface to achieve two-sided flatness, and then analyzing the shear stress generated internally due to the difference in stiffness of the two fabrics, causing deformation, and quantifying the tendency of this stress to cause shrinkage / bulging on either side when the garment is statically suspended, generating a two-sided stress manifestation difference; and constructing the three-dimensional ironing stress field information based on the two-sided heat memory uniformity and the two-sided stress manifestation difference.

[0011] Optionally, the step of constructing a defect transmission network based on the garment process quality baseline information set to quantitatively assess the evolution path and contribution of potential defects and generate a garment comprehensive process deviation index includes: mapping the interlayer bonding balance information, the internal wrapping morphology information, and the three-dimensional ironing stress field information into garment processing network nodes, and constructing directed connection edges between nodes according to the physical process sequence and causal relationship of garment processing to form a defect transmission network; calculating the initial defect potential value of each node based on the quantitative indicators contained in the information of each garment processing network node; the initial defect potential value is the risk potential characterization value of the defect type and degree detected in the initial inspection; based on the defect transmission network, simulating the process of the defect potential of any node being transmitted and evolved to subsequent nodes along the directed edges, and dynamically calculating the additional defect potential of downstream nodes stimulated by the upstream influence; aggregating the initial and additional defect potential of all nodes to generate the garment comprehensive process deviation index, which is used to characterize the global risk level caused by the local deviation of a single or multiple processes to the double-sided appearance of the finished garment.

[0012] Optionally, the process of simulating the transmission and evolution of the defect potential of any node along the directed edge to subsequent nodes, and dynamically calculating the additional defect potential of downstream nodes stimulated by the upstream influence, includes: defining a transmission relationship matrix of directed edges in the defect transmission network based on historical defect data and material physical properties; the transmission relationship matrix quantifies the influence intensity of the unit defect potential of the upstream node on the downstream node, and defines the possible changes in the manifestation of the defect when it is transmitted across processes; taking the defect potential of the preceding node as input, performing nonlinear mapping calculation through the transmission relationship matrix, iterating until the end of the network; introducing a critical enhancement effect during the iterative calculation process, that is, when the cumulative defect potential transmitted from the upstream to this node exceeds its process tolerance threshold, triggering a nonlinear amplification of the current defect potential of this node; synchronously recording and updating the real-time defect potential state of each network node after being affected by all upstream influences, thereby generating a defect potential field that can reflect the overall picture of the dynamic propagation and evolution of defects in the processing link.

[0013] Optionally, defining the directed edge transmission relationship matrix of the defect transmission network based on historical defect data and material physical properties includes: based on historical defect data, calculating the conditional probability between the quantitative indicators of upstream process nodes and the quantitative indicators of downstream process nodes, and initializing the basic influence weights of the corresponding directed edges in the transmission relationship matrix with these conditional probability values; based on material physical property test data, extracting the stiffness ratio, thermal shrinkage rate difference, and interlayer friction coefficient difference of paired fabrics, and mapping them to a dimensionless material coupling coefficient through a nonlinear fusion function; using the material coupling coefficient as a modulation factor, multiplying it with the basic influence weights, physically correcting the weights, and generating the final transmission intensity of each directed edge in the transmission relationship matrix; normalizing the final transmission intensity into a probabilistic form to complete the construction of the transmission relationship matrix, where the value of any element in the transmission relationship matrix quantifies the probability and magnitude of a unit defect potential of an upstream node triggering a defect in a downstream node.

[0014] Optionally, the step of generating a double-sided garment processing inspection report by performing adaptive feedforward control strategy mapping based on the garment comprehensive process deviation index includes: inputting the garment comprehensive process deviation index into a pre-built feedforward control strategy library, which stores the mapping relationship between different deviation index intervals and dynamic process compensation schemes; matching the optimal compensation scheme from the feedforward control strategy library based on the numerical characteristics and growth trend of the current deviation index, wherein the optimal compensation scheme includes the coordinated adjustment of seam tension, folding trajectory, and ironing parameters; correcting the optimal compensation scheme in real time according to the material coupling coefficient to generate personalized process control instructions for the current batch of fabrics; and integrating the personalized process control instructions, key defect evolution paths, and double-sided quality compliance rate to generate the double-sided garment processing inspection report.

[0015] Secondly, this application provides a garment processing quality inspection platform, the platform comprising: a factor analysis module, used to acquire a multi-source dataset of reversible garments, and based on the multi-source dataset of reversible garments, to perform cross-process collaborative analysis on interlayer dynamic interaction, internal wrapping morphology and three-dimensional ironing stress field, generating a garment process quality base state information set; a defect transmission module, used to construct a defect transmission network based on the garment process quality base state information set to quantitatively evaluate the evolution path and contribution of potential defects, generating a garment comprehensive process deviation index; and a report generation module, used to perform adaptive feedforward control strategy mapping based on the garment comprehensive process deviation index, generating a reversible garment processing inspection report. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram illustrating an application scenario provided in one embodiment of this application; Figure 2 A flowchart illustrating a garment processing quality inspection method provided in one embodiment of this application; Figure 3 This is a schematic diagram of the structure of a garment processing quality inspection platform provided in one embodiment of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0019] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article, unless otherwise specified, generally indicates that the preceding and following related objects have an "or" relationship.

[0020] The embodiments of this application will now be described in further detail with reference to the accompanying drawings.

[0021] Existing garment quality inspection methods lack a mechanism for systematically and quantitatively assessing the evolution path of potential defects in each process and the contribution of cross-process transmission when dealing with a process system consisting of multiple processes, various fabrics, and complex physical interactions. This makes it impossible to effectively locate the root process that causes the final double-sided appearance defects, leading to global quality risks for the entire batch of products and causing serious economic losses and waste of resources.

[0022] Based on this, this application provides a garment processing quality inspection platform and method. First, multi-source data from the production process of reversible garments is collected via IoT sensors, forming a dataset covering fabric parameters, sewing data, and ironing parameters. Based on this dataset, data fusion technology is used to conduct cross-process collaborative analysis of interlayer bonding balance, internal wrapping morphology, and three-dimensional ironing stress field: the influence of interlayer bonding state on internal morphology is analyzed through association rule mining, and the shaping effect of the ironing process is evaluated in conjunction with stress distribution data, generating a process quality baseline information set containing standardized parameters and deviation indicators. Subsequently, a defect propagation network model is constructed, establishing inter-process defect propagation paths based on historical defect data. A graph traversal algorithm is used to quantify the contribution of each process to defect evolution, ultimately calculating a comprehensive process deviation index. Based on this index value, a feedforward control strategy is dynamically mapped and adapted through a threshold rule base to adjust process parameters such as sewing tension and ironing temperature in real time. During execution, an inspection report containing quality trends, control logs, and improvement suggestions is generated and output to staff. This method realizes the transformation of quality control from passive detection to proactive prevention. Through cross-process collaborative analysis, it can identify complex problems such as interlayer bonding abnormalities and shape deformation at an early stage; the defect transmission network accurately locates the root cause process of defects, significantly reducing the rework rate; feedforward control based on offset index realizes adaptive adjustment of process parameters, effectively preventing the generation of defects; it improves the first pass rate and production consistency of double-sided garments, reduces quality costs, and promotes continuous optimization of production processes through digital reporting.

[0023] Figure 1 This is a schematic diagram illustrating an application scenario provided by this application. In the process of garment quality inspection, the method provided in this application improves the overall first-pass yield and production consistency of reversible garments, reduces quality costs, and simultaneously promotes continuous optimization of production processes through digital reporting.

[0024] Specifically, the method of this application is applied to any server that communicates with a multimodal IoT sensor to obtain a multi-source dataset of reversible garments provided by the multimodal IoT sensor. First, multi-source data from the reversible garment production process is collected through the IoT sensor, forming a dataset covering fabric parameters, sewing data, and ironing parameters. Based on this dataset, data fusion technology is used to conduct cross-process collaborative analysis of interlayer bonding balance, internal wrapping morphology, and three-dimensional ironing stress field. Association rule mining is used to analyze the impact of interlayer bonding state on internal morphology, and stress distribution data is combined to evaluate the shaping effect of the ironing process, generating a process quality baseline information set containing standardized parameters and deviation indicators. Subsequently, a defect propagation network model is constructed, establishing inter-process defect propagation paths based on historical defect data. A graph traversal algorithm is used to quantify the contribution of each process to defect evolution, ultimately calculating a comprehensive process deviation index. Based on this index value, a feedforward control strategy is dynamically mapped and adapted through a threshold rule base to adjust process parameters such as sewing tension and ironing temperature in real time. During execution, a detection report containing quality trends, control logs, and improvement suggestions is generated and output to the staff.

[0025] For specific implementation details, please refer to the following examples.

[0026] Figure 2 This is a flowchart illustrating a garment processing quality inspection method according to an embodiment of this application. The method of this embodiment can be applied to the server in the above scenario. For example... Figure 2 As shown, the method includes: S201. Obtain multi-source dataset of double-sided garments. Based on the multi-source dataset of double-sided garments, conduct cross-process collaborative analysis on the interlayer bonding balance, internal wrapping morphology and three-dimensional ironing stress field to generate a base state information set of garment process quality.

[0027] The multi-source dataset for reversible garments can be a collection of multi-source data collected during the processing of reversible garments, including interlayer dynamic tension difference, interlayer relative displacement, three-dimensional wrapping morphology, interlayer temperature difference, and interlayer pressure field distribution. The data comes from multimodal IoT sensors on the garment processing line. Interlayer bonding balance refers to the balance of flatness, tightness, and alignment of the bonding interfaces between the front and back fabrics of a reversible garment during sewing, bonding, and other processes. If this balance is disrupted, defects such as interlayer misalignment and wrinkling will occur. Internal wrapping morphology refers to the integrity and consistency of the wrapping of the internal filling materials (such as cotton, down, or fiber cotton) or interlayer structure of a reversible garment, directly affecting the garment's warmth and three-dimensional appearance. Three-dimensional ironing stress field refers to the stress distribution state formed by the combined effects of temperature and pressure on the garment fabric and internal structure during the three-dimensional ironing process. The uniformity of the stress field determines the morphological stability of the garment after ironing. The basic state information set of garment process quality can be a set of information reflecting the basic state of garment process quality obtained through analysis, including information on interlayer dynamic interaction, internal wrapping morphology, and three-dimensional ironing stress field.

[0028] Specifically, reversible garments present unique challenges for quality inspection because they can be worn on both sides. Traditional quality inspection methods are often limited to isolated checks of single processes, failing to comprehensively assess the complex coupling relationships between interlayer bonding, internal shape, and ironing stress. This leads to potential quality issues (such as interlayer wrinkling, internal deformation, or uneven stress) being overlooked, thus affecting the overall performance and user experience of the garment. Furthermore, existing technologies lack the ability to collaboratively analyze data from multiple processes, making it difficult to identify process deviations early, resulting in increased rework rates and decreased production efficiency. This step collects multi-source data in real time from the production line, captures the internal wrapping shape using a 3D scanner, and records the ironing stress distribution using stress sensors. Data fusion algorithms (such as principal component analysis or association rule mining) are used to integrate this data, analyzing how the interlayer bonding balance affects the internal wrapping shape and the feedback effect of the three-dimensional ironing stress field on the former two aspects. This generates a structured set of garment process quality baseline information, enabling early defect detection, improving quality early warning capabilities, reducing quality fluctuations caused by process disconnections, and laying the foundation for high-quality reversible garment production.

[0029] S202. Based on the basic information set of garment process quality, a defect transmission network is constructed to quantitatively assess the evolution path and contribution of potential defects, and a comprehensive garment process deviation index is generated.

[0030] A defect propagation network can be a graph-based network structure representing the propagation, evolution, and accumulation of defects between different processes in garment manufacturing. Nodes represent specific processes (such as cutting, sewing, and ironing), and edges represent defect propagation paths and probabilities. The garment comprehensive process deviation index can be a comprehensive quantitative indicator obtained by quantifying the deviation of process parameters from the baseline standard and the evolution risk of potential defects based on the garment process quality baseline information set.

[0031] Specifically, in garment manufacturing, defects (such as skewed seams, fabric deformation, or uneven ironing) often propagate and amplify across multiple processes. Traditional quality assessment methods typically rely on post-process inspection or simple statistics, failing to dynamically track defect sources and assess their cross-process impact. This leads to lagging quality control and wasted resources. For example, a minor deviation in the cutting stage may be amplified in the sewing stage and ultimately manifest as a significant defect in the ironing stage. However, current technologies struggle to quantify the contribution of each process, making it impossible to specifically optimize key processes. This step defines defect types and process nodes based on the garment process quality baseline information set. It then trains a network model using historical data to determine defect propagation rules and conditional probabilities. A network traversal algorithm is then used to analyze the evolution path of potential defects, calculating the probability of occurrence and process contribution of each path. Finally, a comprehensive garment process deviation index is generated. Based on this, the root cause of quality problems can be accurately identified, process parameter settings optimized, defect incidence reduced, and production consistency and product reliability improved.

[0032] S203. Based on the comprehensive garment process offset index, perform adaptive feedforward control strategy mapping to generate a double-sided garment processing inspection report.

[0033] Adaptive feedforward control strategies can dynamically map the overall garment process deviation index to corresponding control strategies, thereby adjusting production parameters in advance and preventing defects. Double-sided garment processing inspection reports can be comprehensive reports containing personalized process control instructions, critical defect evolution paths, and double-sided quality compliance rates, used to guide production decisions and continuous optimization.

[0034] Specifically, traditional garment quality inspection often employs feedback control, adjusting only after defects occur. This can lead to production interruptions, resource waste, and inconsistent product quality. This is especially true for reversible garments, whose high standards make post-correction costly. Existing technologies lack adaptability and cannot dynamically adjust processes based on real-time quality indicators, hindering preventative control. This step, based on the garment's comprehensive process deviation index, determines suitable control strategies through mapping rules. These strategies are then applied to production line actuators to achieve real-time parameter control, ultimately outputting a reversible garment processing inspection report. This enables preventative quality management and dynamic adjustment, ensuring production process stability and high product consistency, ultimately promoting the garment processing industry towards intelligent and efficient development.

[0035] The method provided in this embodiment first collects multi-source data from the production process of reversible garments using IoT sensors, forming a dataset covering fabric parameters, sewing data, and ironing parameters. Based on this dataset, data fusion technology is used to conduct cross-process collaborative analysis of the interlayer bonding balance, internal wrapping morphology, and three-dimensional ironing stress field: the influence of interlayer bonding state on internal morphology is analyzed through association rule mining, and the shaping effect of the ironing process is evaluated in combination with stress distribution data, generating a process quality base state information set containing standardized parameters and deviation indicators. Subsequently, a defect propagation network model is constructed, and the defect propagation path between processes is established based on historical defect data. A graph traversal algorithm is used to quantify the contribution of each process to the evolution of defects, and finally, a comprehensive process deviation index is calculated. According to the index value, a feedforward control strategy is dynamically mapped and adapted through a threshold rule base to adjust process parameters such as sewing tension and ironing temperature in real time. During the execution process, an inspection report containing quality trends, control logs, and improvement suggestions is generated and output to the staff. This method realizes the transformation of quality control from passive detection to proactive prevention. Through cross-process collaborative analysis, it can identify complex problems such as interlayer bonding abnormalities and shape deformation at an early stage; the defect transmission network accurately locates the root cause process of defects, significantly reducing the rework rate; feedforward control based on offset index realizes adaptive adjustment of process parameters, effectively preventing the generation of defects; it improves the first pass rate and production consistency of double-sided garments, reduces quality costs, and promotes continuous optimization of production processes through digital reporting.

[0036] In some embodiments, the multi-source dataset for reversible garments includes interlayer dynamic tension difference, interlayer relative displacement, wrapping three-dimensional morphology, interlayer temperature difference, and interlayer pressure field distribution. Based on the interlayer dynamic tension difference, the interlayer dynamic bonding effect during the sewing process is analyzed to generate interlayer bonding balance information to characterize the double-sided bonding balance state at the seam. Based on the interlayer relative displacement and wrapping three-dimensional morphology, the internal wrapping morphology during the folding and wrapping process is analyzed to generate internal wrapping morphology information characterizing the stable state of the inner wrapping structure. Based on the interlayer temperature difference and interlayer pressure field distribution, the coupling relationship of the double-sided stress field during the three-dimensional ironing process is analyzed to generate three-dimensional ironing stress field information characterizing the stable state of the garment. Based on the interlayer dynamic interaction information, internal wrapping morphology information, and three-dimensional ironing stress field information, a garment process quality base state information set is generated.

[0037] Interlayer dynamic tension difference refers to the dynamic tension difference generated at the interface between the two layers of fabric during sewing, folding, and other processes in reversible garments. It characterizes the stress balance state when the layers are bonded. Interlayer relative displacement refers to the offset (including lateral and longitudinal offset) of the relative positions of the two layers of fabric during sewing, folding, and wrapping processes in reversible garments. It is a core indicator for evaluating the alignment accuracy between layers and directly affects the symmetry of the garment's appearance. The three-dimensional shape of the wrapping refers to the three-dimensional structural morphology data formed after the internal filling material (such as cotton or down) of the reversible garment is filled. Interlayer temperature difference refers to the temperature difference between the two layers of fabric and the interlayer during the three-dimensional ironing process in reversible garments. Excessive temperature difference can easily lead to uneven fabric shrinkage and deformation, while insufficient temperature difference cannot achieve the desired ironing effect. Interlayer pressure field distribution refers to the distribution data of the pressure applied by the ironing equipment between the garment layers during the three-dimensional ironing process. Information on interlayer bonding balance can be generated by analyzing the stress balance state of the two layers of fabric during the sewing process, based on interlayer dynamic tension difference data, thus representing the stability of the double-sided bonding at the seam. Information on internal wrapping morphology can be generated by analyzing the folding and wrapping process, based on interlayer relative displacement and three-dimensional wrapping morphology data, thus representing the stability of the garment's internal structure. Information on three-dimensional ironing stress field can be generated by analyzing the stress coupling relationship between the fabric and its internal structure during three-dimensional ironing, based on interlayer temperature difference and interlayer pressure field distribution data, thus representing the garment's steady state.

[0038] Specifically, in the processing of reversible garments, because both sides of the fabric can be worn, the complexity of the process and the quality requirements are significantly higher than those of traditional single-sided garments. In key processes such as sewing, folding and wrapping, and three-dimensional ironing, the interaction between the two layers of fabric directly affects the comfort, aesthetics, and durability of the finished garment. Traditional garment quality inspection methods are usually limited to isolated analysis of a single process, lacking comprehensive consideration of the synergistic effects across processes, leading to the following problems: First, in the sewing process, if the dynamic tension difference between the layers is not monitored and balanced in real time, it will cause one side of the seam to be too tight and the other too loose, resulting in seam distortion or wrinkling, affecting the wearing experience and appearance quality; Second, in the folding and wrapping process, if the relative displacement between the layers and the three-dimensional shape of the wrapping are not precisely controlled, it may lead to a loose or deformed internal wrapping structure, resulting in unevenness inside the garment and reducing the product grade; Finally, in the three-dimensional ironing process, if the temperature difference and pressure field distribution between the layers are not coordinated, residual stress will be introduced, causing the garment to deform during subsequent use or washing, shortening its service life. To address the above issues, this step activates the multi-source data acquisition system for the double-sided garment processing production line, collecting five types of core data: In the sewing process, tension sensors acquire the dynamic tension difference between layers (e.g., ±5N); visual alignment detection equipment acquires the relative displacement between layers (e.g., ≤2mm); in the filling process, a 3D shape scanner and industrial CT equipment acquire the three-dimensional shape data of the wrapping; in the ironing process, an infrared temperature sensor array acquires the temperature difference between layers (e.g., ≤3℃); and a pressure sensor array acquires the pressure field distribution between layers (e.g., 50-80kPa). Simultaneously, data on fabric material... Auxiliary data such as sewing speed (e.g., 100 stitches / minute) are calibrated and integrated into a multi-source dataset. Then, dimensional analysis is performed: information on interlayer bonding balance is generated based on tension difference (e.g., exceeding 8N can easily cause seam misalignment); information on internal wrapping shape is generated by combining displacement (e.g., exceeding 3mm can easily cause filling voids) and three-dimensional morphology; and information on three-dimensional ironing stress field is generated by using temperature difference (e.g., exceeding 5℃ can easily leave stress) and pressure field. Finally, the three types of information are integrated, and the parameter range is determined by combining historical qualified data (e.g., data from 1000 qualified products) to generate a basic state information set of garment process quality.

[0039] The method provided in this embodiment clarifies the core constituent parameters of the multi-source dataset, avoiding the blindness of traditional data collection and ensuring that the collected data can accurately represent the unique quality dimensions of reversible garments. The dimensional analysis logic enables the precise decomposition of the three key quality factors: interlayer bonding, internal wrapping, and three-dimensional ironing. This makes the generated base state information set no longer a general empirical value, but a quantitative standard based on actual processing data, providing a reliable benchmark for subsequent quality assessment.

[0040] In some embodiments, the interlayer dynamic tension difference is decoupled to separate the static deviation component characterizing the uneven stress on the two fabrics and the dynamic fluctuation component of the instantaneous asynchronous fabric feeding during sewing. Based on the static deviation component, the residual stress of the seam caused by the difference in shrinkage and stiffness during the sewing of heterogeneous fabrics is quantified, and its symmetrical influence on the appearance flatness of the two sides is analyzed to generate a bidirectional flatness attenuation gradient, which is used to predict the stable range of the seam where no visible distortion occurs on either side. Based on the dynamic fluctuation component, its frequency and amplitude characteristics are analyzed to detect the periodic pause and sudden advance phenomenon caused by the mismatch of the friction coefficients between the two fabrics and the presser foot / feed teeth, and the irreversible tensile deformation accumulated on one side of the fabric by this phenomenon is evaluated to generate an interlayer slippage risk index, which is used to associate the wavy sewing defect that may occur on either side of the two sides. Based on the bidirectional flatness attenuation gradient and the interlayer slippage risk index, information on the interlayer bonding balance is constructed.

[0041] Uneven fabric stress can occur during double-sided garment sewing when the two layers are different (e.g., different materials, thicknesses, shrinkage rates, stiffness), leading to a lack of consistent stress between the layers. Instantaneous feed asynchrony can occur during double-sided garment sewing when the feed dogs and presser feet of the sewing machine experience momentary inconsistencies in their feed speeds due to deviations in operating parameters (e.g., uneven feed dog speeds, unstable presser foot pressure). Dynamic fluctuation components can be decoupled from the interlayer dynamic tension difference, characterizing the tension changes caused by instantaneous interference factors during sewing (e.g., instantaneous feed asynchrony, mismatch in friction coefficients between the fabric and presser foot / feed dogs). Residual seam stress can be the potential stress remaining at the seam during double-sided garment sewing due to the inability of different fabrics to fully release internal stress caused by differences in shrinkage rate and stiffness; this stress will slowly release over time or during wear. The bidirectional flatness attenuation gradient can be an index generated based on the quantification of residual seam stress, used to predict the stable range where no visible distortion occurs on either side of the double-sided garment seam. The coefficient of friction between the presser foot and feed dog can be defined as the coefficient that hinders relative movement between the contact surfaces of the presser foot / feed dog and the fabric during double-sided garment sewing. Periodic pauses and sudden advances can be an abnormal phenomenon during double-sided garment sewing where a mismatch in the coefficient of friction between the two layers of fabric and the presser foot / feed dog leads to periodic brief pauses followed by sudden advances in the fabric feeding process. The interlayer slippage risk index is a quantitative indicator that assesses the risk of wavy seam defects formed after irreversible tensile deformation accumulates on the fabric due to the periodic pauses and sudden advances, based on the dynamic fluctuation components (frequency and amplitude characteristics) of the interlayer dynamic tension difference. Wavy seam defects can be wavy unevenness defects that appear on both sides or one side of the seam due to irreversible tensile deformation caused by periodic pauses and sudden advances during double-sided garment sewing.

[0042] Specifically, in the double-sided garment sewing process, accurate analysis of the dynamic tension difference between layers is crucial to ensuring the flatness of both sides of the garment. Traditional testing methods only focus on the overall tension average and cannot analyze the dynamic characteristics of the tension difference, which has obvious limitations: First, they do not separate static deviation from dynamic fluctuation components—static deviation stems from residual stress caused by the difference in shrinkage rates of heterogeneous fabrics, while dynamic fluctuation is caused by periodic tension peaks due to frictional mismatch in the fabric feeding mechanism; Second, they lack quantitative indicators to predict the stable range of double-sided seam balance, leading to undetectable single-sided wrinkling or wavy defects in the early stages. These defects will be amplified in subsequent processes, and relying solely on empirical tension monitoring cannot meet the high standards required for double-sided garments. To address the above issues, this step utilizes a high-precision tension sensor (sampling frequency e.g., 500Hz, synchronized with the sewing machine speed) located beneath the sewing machine needle plate to collect real-time raw data on the dynamic tension difference between the two layers of fabric during sewing. Simultaneously, it retrieves current fabric physical property parameters (e.g., shrinkage 3%, stiffness 20N / m) from the production management system. The raw tension data is decoupled, separating the static deviation component characterizing fabric properties from the dynamic fluctuation component characterizing instantaneous factors. Based on the static deviation component, the residual stress at the seam formed by sewing dissimilar fabrics is quantified (e.g., 5MPa), generating a bidirectional flatness attenuation gradient (e.g., 0.5mm / day, corresponding to a stable tension range of 5-8N). Based on the dynamic fluctuation component, its frequency (e.g., 2Hz) and amplitude (e.g., ±2N) are identified, correlated with periodic pauses and sudden advances, and an interlayer slippage risk index (e.g., 0.8, corresponding to high risk) is generated after tensile deformation. Finally, the two are integrated, supplementing process optimization suggestions (e.g., adjusting the presser foot pressure to 0.3MPa) to form information on the interlayer bonding balance.

[0043] The method provided in this embodiment solves the problem that traditional detection cannot locate the root cause of defects by decoupling the dynamic tension difference between layers, enabling process adjustments to be precisely targeted at the problem type and avoiding blind adjustments; the bidirectional flatness attenuation gradient can predict the long-term stable range of the joint in advance, and the interlayer slippage manifestation risk index can warn of potential wavy stitching defects, effectively reducing the exposure of defects in semi-finished products in subsequent processes, and reducing rework costs and production losses.

[0044] In some embodiments, the relative displacement between layers is tracked and accumulated to analyze the asynchronicity and path deviation of the inner and outer fabrics during folding motion in three-dimensional space. The focus is on evaluating the mismatch in rebound after folding caused by the difference in stiffness between the two fabrics, generating double-sided contour symmetry. The three-dimensional shape of the package is reconstructed by point cloud and the curvature of the surface is analyzed to detect local morphological bulges / collapses caused by the stacking of internal sutures and functional components in a limited space. The visibility and distribution range of these bulges / collapses on the two outer surfaces are quantified to generate internal component migration risk. Based on the double-sided contour symmetry and internal component migration risk, internal package morphology information is constructed to characterize the package structure so that it does not produce non-designed contours on either side.

[0045] Trajectory tracking is an analytical method that continuously records and tracks the movement paths of fabric feature points (such as edge markers and seam feature points) corresponding to the relative displacement between layers throughout the folding process. Rebound mismatch refers to the phenomenon where, after folding, the inner and outer fabrics recover to their original shape to different degrees due to differences in stiffness, resulting in asymmetrical deformation of the double-sided contour. Double-sided contour symmetry is an index obtained by quantitatively evaluating the degree of overlap, dimensional consistency, and morphological symmetry of the outer contours of the front and back of the garment after folding and wrapping, used to characterize whether the double-sided contour meets the design's symmetry requirements. Point cloud reconstruction is an analytical method that reconstructs a continuous and complete three-dimensional model from discrete data points of the three-dimensional shape of the garment collected by a 3D shape scanner and industrial CT equipment through data processing technology, used to visually present the internal and external spatial morphology. Surface curvature analysis is an analytical method that calculates and analyzes the curvature of the reconstructed three-dimensional model surface and the surface of internal components, identifying morphological anomalies such as local bulges and collapses through curvature changes. Functional components can be non-decorative parts integrated into the garment, such as thermal linings, hidden pockets, and support panels. Their spatial position directly affects the stability of the internal wrapping structure. The risk of internal component migration can be assessed by quantitatively evaluating the likelihood of internal seam seams and functional components shifting or stacking during folding, wrapping, and subsequent processes. This indicator can be used to warn of the risk of internal defects becoming apparent.

[0046] Specifically, in the folding and wrapping process of reversible garments, the analysis of the internal wrapping morphology is crucial to ensuring the structural integrity and visual appeal of the finished garment. Traditional inspection methods rely on manual vision or simple size measurement, which has significant limitations: First, they cannot analyze the dynamic characteristics of relative displacement between layers—this asynchronous folding caused by differences in fabric stiffness leads to asymmetry in the double-sided contours, but traditional methods lack trajectory tracking and cumulative calculation, making it difficult to detect early risks; Second, the evaluation of the three-dimensional shape of the wrapping only focuses on the external contour, ignoring the local bulges / collapses caused by the stacking of internal components. These morphological anomalies are manifested differently on both sides, and traditional methods are difficult to identify in the early stages, causing problems to appear only in later processes; In addition, as a key intermediate link, the quality defects of folding and wrapping are transferable and can be amplified to subsequent processes such as ironing. Therefore, traditional methods cannot meet the high standards required for reversible garments. To address the above issues, this step uses visual alignment detection equipment (e.g., sampling frequency 10 frames / second) during the folding process to collect relative displacement data between the inner and outer fabric layers. Simultaneously, a 3D morphology scanner with a precision of 0.1mm is used to acquire the external contour, and industrial CT equipment is used to acquire the internal structure, forming a three-dimensional morphology dataset of the wrapping. Fabric stiffness parameters (e.g., 50N for the inner layer and 80N for the outer layer) are retrieved. The relative displacement trajectory between layers is tracked (e.g., identifying a 0.5s time difference between the inner and outer layers) and the cumulative amount is calculated (e.g., a total displacement deviation of 3mm). Combined with stiffness analysis to determine the springback mismatch, the double-sided contour symmetry is generated (e.g., overlap ≥95% is acceptable). After reconstructing the point cloud of the wrapping morphology, surface curvature analysis is performed to locate defects caused by the stacking of internal suture bone positions (e.g., local bulges of 2mm), quantify the degree of manifestation (e.g., slight manifestation visible only at specific angles) and distribution range (e.g., a 5cm diameter area), and assess the risk of internal component migration (e.g., a risk value ≥0.7 is considered high risk). Finally, the two are integrated, and process suggestions are added (e.g., adjusting the folding sequence) to form internal wrapping morphology information.

[0047] The method provided in this embodiment captures the asynchrony and path deviation of the inner and outer fabrics during the folding process through trajectory tracking and cumulative calculation, solving the problem that traditional detection cannot accurately measure the symmetry state and providing a standardized basis for quality control. With the help of point cloud reconstruction and surface curvature analysis, it can penetrate and identify the bulge / collapse defects caused by internal suture bone positions and functional components, quantitatively assess the migration risk of internal components, avoid the risk of internal defects appearing in subsequent processes in advance, and reduce rework costs.

[0048] In some embodiments, time-series analysis and gradient calculation are performed on the interlayer temperature difference to solve the asymmetric conduction process when heat penetrates the garment interlayer composed of heterogeneous fabrics. The time difference and temperature difference between the front and back sides reaching the fiber glass transition temperature are quantified, and the resulting risk of mismatch in double-sided heat setting memory is assessed to generate double-sided heat memory uniformity. The interlayer pressure field distribution is decomposed into spatial vectors to identify the uniform pressure distribution applied on the three-dimensional curved surface to achieve double-sided flatness. The shear stress generated inside due to the difference in stiffness between the two fabrics is analyzed, and the tendency of this stress to cause shrinkage / bulging on either side when the garment is statically suspended is quantified to generate double-sided stress manifestation difference. Based on the double-sided heat memory uniformity and double-sided stress manifestation difference, three-dimensional ironing stress field information is constructed.

[0049] Asymmetric heat transfer occurs when heat is transferred between layers of garments made of heterogeneous fabrics. The difference in thermal conductivity and thickness between the two sides and the interlayer fabrics leads to an asymmetric distribution of heat transfer speed and penetration depth, which is the core mechanism causing differences in heat setting on both sides. Fiber vitrification is the critical temperature at which fabric fibers transition from a glassy state to a highly elastic state, representing a key temperature threshold for heat setting. The risk of mismatched heat setting memory on both sides can arise from the time and temperature difference between the two sides reaching the fiber vitrification transition temperature. This difference can lead to variations in the shape memory stability of the two fabrics after heat setting. A higher risk increases the likelihood of inconsistent shape on both sides during subsequent use (e.g., one side is flat while the other is wrinkled). The evenness of heat memory on both sides is a quantitative indicator of the matching degree of heat setting memory stability between the two sides. It characterizes whether both sides maintain consistent shape stability after ironing; a more even value indicates a more uniform heat setting effect. Spatial vector decomposition can decompose interlayer pressure field distribution data into pressure components in different directions according to a spatial coordinate system (such as the normal component perpendicular to the fabric surface and the tangential component parallel to the fabric surface). Equilibrium pressure distribution refers to the theoretically uniform pressure distribution applied to the surface of a garment by an ironing device to achieve a smooth, double-sided surface. Shear stress is the stress parallel to the fabric layer generated within the garment due to the difference in stiffness between the front and back fabrics, resulting from the equilibrium pressure. It is the core internal stress that causes defects such as shrinkage and bulging when the garment is statically hung. The difference in the apparent intensity of double-sided stress can be used as an indicator to quantitatively assess the difference in the degree of visible surface defects (shrinkage, bulging) caused after the release of internal shear stress on the front and back sides.

[0050] Specifically, in the three-dimensional ironing process of reversible garments, accurate analysis of stress field coupling relationships is crucial to ensuring the stability of the garment's shape and its long-term wear performance. Traditional ironing quality monitoring relies on manual experience and basic temperature and pressure monitoring, which has significant limitations: First, it cannot analyze the dynamic conduction characteristics of interlayer temperature differences—this asymmetric heat conduction caused by the thermal conductivity differences of heterogeneous fabrics can lead to a mismatch in the heat setting memory of both sides, but traditional methods lack time tracking and gradient analysis, making it difficult to detect potential risks early. Second, the assessment of the interlayer pressure field distribution only focuses on the overall pressure value, ignoring the internal shear stress generated on the three-dimensional curved surface due to differences in fabric stiffness. This stress will gradually release and cause deformation when the garment is suspended, but traditional methods cannot identify its spatial distribution trend. In addition, as the final critical process, ironing defects are irreversible and will directly affect the quality of the finished product. Therefore, traditional methods cannot meet the high standards required for reversible garments. To address the above issues, this step uses an infrared temperature sensor array (sampling frequency e.g., 10Hz, synchronized with the ironing equipment) to collect interlayer temperature difference time-series data throughout the three-dimensional ironing process. A flexible pressure sensor array acquires pressure data from the garment contact area and processes it into interlayer pressure field distribution data. Simultaneously, auxiliary parameters such as the glass transition temperature (e.g., 120℃), thermal conductivity (e.g., 0.1W / (m•K)), and stiffness (e.g., 20N / m) of the fibers on both sides of the fabric are retrieved from a fabric material database. The characteristics of each stage are tracked and analyzed using a gradient meter. The asymmetric heat conduction process is calculated, and the time difference (e.g., 3 seconds) and temperature difference (e.g., 5℃) between the front and back sides reaching the glass transition temperature are quantified. After assessing the risk of heat setting memory mismatch, the heat memory balance of both sides is generated. The pressure field data is spatially vector decomposed (e.g., separating the normal pressure of 50kPa and the tangential pressure of 10kPa), and the internal shear stress is analyzed in combination with the fabric stiffness. The shrinkage / bulging trend during static hanging is assessed, and the stress manifestation difference of both sides is generated. Finally, the two are integrated, and process suggestions are added (e.g., adjusting the ironing temperature gradient by 2℃ / min) to construct three-dimensional ironing stress field information.

[0051] The method provided in this embodiment captures the asymmetric heat conduction process in heterogeneous fabrics through time series analysis and gradient calculation, which solves the limitations of traditional detection methods that cannot penetrate the heat conduction mechanism and can only be observed on the surface, thus improving the foresight of the assessment. By identifying internal shear stress through spatial vector decomposition and associating the intrinsic relationship between stress release and surface defects, potential risks such as shrinkage and bulging can be accurately assessed, enabling pressure field analysis to truly support the steady-state assessment of garments.

[0052] In some embodiments, information on interlayer bonding balance, internal wrapping morphology, and three-dimensional ironing stress field is mapped to garment processing network nodes. Directed connections between nodes are constructed based on the physical sequence and causal relationships of garment processing, thus forming a defect transmission network. Based on the quantitative indicators contained in the information of each garment processing network node, the initial defect potential value of each node is calculated. This initial defect potential value represents the risk potential of the initially detected defect type and its severity. Based on the defect transmission network, the process of the defect potential of any node being transmitted and evolving along directed edges to subsequent nodes is simulated, and the additional defect potential of downstream nodes affected by upstream influences is dynamically calculated. The initial and additional defect potentials of all nodes are aggregated to generate a comprehensive garment process deviation index, which characterizes the global risk level ultimately caused by local deviations in one or more processes on the double-sided appearance of the finished garment.

[0053] The garment processing network nodes can be network nodes mapped to information on interlayer bonding balance, internal wrapping morphology, and three-dimensional ironing stress field, respectively. Each node corresponds to the quality status of a core process. Directed connecting edges can be directed line segments connecting adjacent network nodes, constructed based on the physical process sequence of garment processing (e.g., sewing → folding and wrapping → three-dimensional ironing) and causal relationships (the quality of preceding processes affects the quality of subsequent processes). These edges characterize the direction and path of defect transmission from upstream to downstream nodes. The initial defect potential value is the initial defect risk quantification value of each garment processing network node, calculated based on the quantitative indicators it contains. It represents the risk potential of the initially detected defect type and its severity, reflecting the severity of local defects in a single process. Additional defect potential is the additional defect risk quantification value generated when a downstream node is affected by the defect potential transmission from an upstream node. It reflects the cumulative impact of cross-process defect transmission on subsequent processes. The global risk level is the overall impact of local process defects, after cross-process transmission and superposition, on core quality indicators such as the symmetry, flatness, and stability of the garment's double-sided appearance.

[0054] Specifically, in the quality inspection of double-sided garments, the construction of a defect transmission network and the generation of a comprehensive garment process deviation index are crucial for achieving full-process quality control and risk warning. Traditional quality inspection methods rely on independent process inspection and finished product inspection, which have significant shortcomings: First, they lack analysis of cross-process defect transmission mechanisms and cannot track the evolution path of defects. For example, residual stress in the sewing process may evolve into surface distortion in subsequent processes, but traditional methods are difficult to detect in time. Second, defect risk assessment is mostly limited to qualitative description and ignores the risk accumulation effect under the coupling of multiple processes. In addition, traditional methods have not established a directed correlation model between processes and cannot simulate the transmission process of defects along the process chain. To address the above issues, this step first maps the information on interlayer bonding balance (including bidirectional flatness attenuation gradient, e.g., acceptable standard ≤0.3mm / day), internal wrapping morphology (including double-sided contour symmetry, e.g., acceptable standard ≥95%), and three-dimensional ironing stress field (including double-sided thermal memory uniformity, e.g., acceptable standard ≥0.8) to three garment processing network nodes: sewing, folding and wrapping, and three-dimensional ironing. Directed connecting edges are constructed according to the physical process sequence of "sewing → folding and wrapping → three-dimensional ironing" and the causal relationship (e.g., sewing quality affects folding and wrapping quality), forming a defect transmission network. Then, quantitative indicators for each node are extracted and combined with preset risk level standards. (For example, if the index exceeds the ground state range by 10%, it corresponds to low risk). Calculate the initial defect potential value (e.g., if the index of the sewing node exceeds the ground state by 8%, the initial value is 0.35). Then, simulate the transmission of defect potential along the directed edge (e.g., the 0.35 value from the sewing node is transmitted to the folding and wrapping node). Combined with the quality status of the downstream node itself (e.g., the initial index of the folding and wrapping node exceeds the ground state by 5%), dynamically calculate the additional defect potential (e.g., the additional value of the folding and wrapping node is 0.12). Finally, assign weights according to the importance of the process (sewing 0.4, folding and wrapping 0.3, three-dimensional ironing 0.3), aggregate the initial and additional potential values ​​of all nodes, and generate the garment comprehensive process deviation index (e.g., 0.32).

[0055] The method provided in this embodiment, by constructing a defect propagation network, clearly presents the path and pattern of defects from preceding processes to subsequent processes, solving the problem that traditional detection cannot link defects across processes, and providing a clear direction for tracing the source of quality problems and optimizing processes; the calculation of initial defect potential value and additional defect potential value makes defect risk assessment uniform and objective, and provides a scientific basis for comparing the quality of different batches and different processes.

[0056] In some embodiments, a transmission relationship matrix of directed edges in the defect transmission network is defined based on historical defect data and material physical properties. The transmission relationship matrix quantifies the influence intensity of the unit defect potential of the upstream node on the downstream node, and defines the possible changes in the manifestation of the defect when it is transmitted across processes. The defect potential of the preceding node is used as input, and nonlinear mapping calculation is performed through the transmission relationship matrix. The calculation is iterated until the end of the network. During the iterative calculation, a critical enhancement effect is introduced, that is, when the cumulative defect potential transmitted from the upstream to this node exceeds its process tolerance threshold, a nonlinear amplification of the current defect potential of this node is triggered. The real-time defect potential state of each network node after being affected by all upstream nodes is recorded and updated simultaneously, thereby generating a defect potential field that can reflect the dynamic propagation and evolution of defects in the processing link.

[0057] Historical defect data can be a complete dataset recorded in the past production process of reversible garments, including defect type, occurrence process, transmission path, and degree of impact. It covers defect cases and handling results for each process, such as sewing, folding, and three-dimensional ironing. Material physical properties can be the core physical parameters of the fabrics and accessories used in reversible garments, including fabric stiffness, shrinkage rate, coefficient of friction, and thermal stability, which directly affect the generation and transmission efficiency of defects. The transmission relationship matrix can be a matrix model constructed based on historical defect data and material physical properties to quantify the transmission law of defect potential. It includes two core dimensions: first, the influence intensity coefficient of the unit defect potential of the upstream node on the downstream node; and second, the morphological transformation rules when defects are transmitted across processes. Nonlinear mapping calculation can be a nonlinear calculation of the defect potential transmission process based on the transmission relationship matrix, used to simulate the nonlinear changes in defect potential caused by changes in transmission path and process characteristics. The critical enhancement effect can be the phenomenon of nonlinear amplification of the current defect potential of a node when the cumulative defect potential transmitted from upstream to a certain node exceeds the process tolerance threshold of that node, used to simulate the abrupt change law of defect risk. A defect potential field can be a structured data set that integrates the real-time defect potential status of all network nodes, forming a comprehensive and dynamic reflection of the propagation and evolution of defects in the processing link.

[0058] Specifically, in the quality inspection of double-sided garments, simulating the evolution of defect potential transmission is crucial for accurate risk warning. Traditional methods have significant shortcomings: First, the lack of systematic modeling makes it impossible to quantify the cumulative impact of defects across processes. For example, residual stress in the seams of the sewing process may be transformed into contour distortion in the folding process and fixed as permanent deformation in the ironing process, but traditional independent inspections cannot capture this evolution path. Second, the linear assumption ignores the nonlinear characteristics in actual processing. When defects exceed the process tolerance, they may trigger a sharp amplification, but traditional assessments cannot warn of such abrupt risks. In addition, the morphology of defects changes during transmission, but traditional methods have not established corresponding mapping models. To address the above issues, this step constructs a transmission relationship matrix based on historical defect data (such as records of the impact of interlayer slippage defects in the sewing process on folding and ironing processes over the past two years) and material physical properties (such as fabric stiffness of 80 N / m and shrinkage rate of 3%). This matrix clarifies the influence intensity of the upstream node's unit defect potential on the downstream (e.g., the influence coefficient of the sewing node on the folding and wrapping node is 0.7) and the defect morphology transformation rules (e.g., "interlayer tension deviation" is transformed into "double-sided contour asymmetry" after transmission). Using the initial defect potential of the sewing node (e.g., 0.4) as input, a nonlinear mapping calculation is performed through the matrix (e.g., multiplying the initial potential by the influence coefficient 0.7, and then combining the surface...). The material friction coefficient is corrected by 1.2 to obtain an initial influence value of 0.336 transmitted to the folding node. This value is combined with the initial potential of the folding node itself (e.g., 0.2) as a new input, and iterative calculation is performed until the ironing node. During the process, it is judged in real time whether the accumulated defect potential of the node (e.g., 0.336 + 0.2 = 0.536 at the folding node) exceeds the process tolerance threshold (e.g., 0.5). If it exceeds, a critical enhancement effect is triggered (e.g., the potential is amplified by 1.5 times to 0.804). Finally, the real-time potential of each node (e.g., 0.4 for sewing, 0.804 for folding, and 0.6 for final ironing) is recorded in real time and integrated to form a defect potential field that reflects the evolution of the entire defect propagation chain.

[0059] By constructing a transmission relationship matrix, the influence intensity and morphological transformation rules of defect transmission are quantified through the method provided in this embodiment. This solves the problem of "fuzziness" in the transmission simulation of existing technologies, enabling the calculation of additional defect potential based on objective quantitative evidence and significantly improving the reliability of the calculation results. The introduction of the critical enhancement effect breaks through the limitations of traditional linear calculation and can accurately simulate the real scenario of "accumulated defects exceeding the threshold triggering risk amplification", making the calculation of additional defect potential conform to the defect evolution law in actual production.

[0060] In some embodiments, based on historical defect data, the conditional probabilities between the quantitative indicators of upstream process nodes and the quantitative indicators of downstream process nodes are statistically analyzed. These conditional probability values ​​are used to initialize the basic influence weights of the corresponding directed edges in the transmission relationship matrix. Based on material physical property test data, the stiffness ratio, thermal shrinkage rate difference, and interlayer friction coefficient difference of the paired fabrics are extracted and mapped to a dimensionless material coupling coefficient through a nonlinear fusion function. The material coupling coefficient is used as a modulation factor and multiplied with the basic influence weights to physically correct the weights, generating the final transmission strength of each directed edge in the transmission relationship matrix. The final transmission strength is normalized to a probabilistic form, completing the construction of the transmission relationship matrix. The value of any element in the transmission relationship matrix quantifies the probability and magnitude of a unit defect potential of an upstream node triggering a defect in a downstream node.

[0061] Conditional probability can be the probability that when a certain quantitative indicator of an upstream process node becomes abnormal, the corresponding quantitative indicator of a downstream process node will also become abnormal, based on historical defect data. Stiffness ratio can be the ratio of the stiffness parameters of the two fabrics in a pair, used to characterize the difference in their resistance to deformation; the greater the difference, the more significant the modulation effect on defect transmission. Heat shrinkage rate difference can be the difference in the heat shrinkage rate parameters of the two fabrics in a pair, used to characterize the difference in the degree of shrinkage of the two fabrics during heat processing; the greater the difference, the easier it is for defects to undergo morphological changes during transmission. Interlayer friction coefficient difference can be the difference in the friction coefficients of the two fabrics and the interlayer layer in a pair, used to characterize the difference in frictional resistance of the two fabrics during relative motion; the greater the difference, the easier it is to accelerate or slow down defect transmission. Material coupling coefficient can be a dimensionless coefficient obtained through a nonlinear fusion function, used to modulate the weight of the basic influence. Final transmission intensity can be a value obtained by correcting the basic influence weights with the material coupling coefficient, accurately characterizing the intensity of defect transmission, integrating historical statistical patterns and material physical mechanisms.

[0062] Specifically, in the quality inspection of double-sided garment processing, traditional methods have significant shortcomings in defect propagation modeling: First, they lack systematic analysis of historical defect data, making it impossible to quantify the statistical regularity of defect impact between processes. For example, there is a correlation between the interlayer dynamic tension difference in the sewing process and the internal wrapping morphology defects in the folding process, but traditional methods rely solely on experience and cannot accurately calculate conditional probabilities, leading to strong subjectivity in weight initialization. Second, they completely ignore the modulating effect of material physical properties. When the stiffness ratio of the two fabrics differs significantly, the residual stress from sewing is more likely to cause contour distortion during folding, but traditional models do not integrate these parameters, causing the prediction to be disconnected from the actual physical mechanism. In addition, traditional methods use linear assumptions and cannot handle nonlinear effects in defect propagation, such as abrupt changes in propagation intensity caused by the interaction of material parameters. To address the above issues, this step extracts correlation cases between "upstream process defects and downstream process defects" from historical defect data, statistically analyzes conditional probabilities (e.g., when the interlayer slippage risk index exceeds the standard by 15% in the sewing process, the conditional probability of the double-sided contour symmetry exceeding the standard in the folding process is 60%), and maps these probabilities to the basic influence weights (e.g., 0.6) of the directed edges corresponding to the transmission relationship matrix; and obtains paired fabric parameters (e.g., fabric A stiffness 80 N / m, fabric B stiffness 50 N / m, heat shrinkage rates A 2%, B 3%, interlayer friction coefficient A 0.3) from material physical property test data. B0.2), calculate the stiffness ratio (1.6), thermal shrinkage difference (1%), and interlaminar friction coefficient difference (0.1), input the nonlinear fusion function to obtain the material coupling coefficient (e.g., 0.75); use this coefficient as a modulation factor, multiply it with the basic influence weight (0.6×0.75=0.45) to obtain the final transmission strength; finally, normalize all the final transmission strengths to the 0-1 interval (e.g., 0.45 to maintain the probabilistic form), complete the construction of the transmission relationship matrix, and the matrix elements (e.g., 0.45) quantify the probability and amplitude of the upstream unit defect potential triggering the downstream defect.

[0063] The method provided in this embodiment initializes the basic weights by statistically analyzing historical defect data, and generates coupling coefficients to correct the weights by combining them with the physical properties of the materials. This ensures that the final transmission intensity conforms to both historical transmission patterns and the physical mechanism of the material combination, solving the problem that existing technology matrices are "detached from the actual materials". The introduction of material coupling coefficients enables the transmission relationship matrix to adapt to the defect transmission patterns of different fabric combinations, breaking through the limitation of existing technology matrices "relying on specific historical data" and improving the matrix's general reusability.

[0064] In some embodiments, the comprehensive garment process deviation index is input into a pre-built feedforward control strategy library, which stores the mapping relationship between different deviation index ranges and dynamic process compensation schemes. Based on the numerical characteristics and growth trend of the current deviation index, the optimal compensation scheme is matched from the feedforward control strategy library. The optimal compensation scheme includes the coordinated adjustment of seam tension, folding trajectory, and ironing parameters. The optimal compensation scheme is corrected in real time according to the material coupling coefficient to generate personalized process control instructions for the current batch of fabrics. The personalized process control instructions, key defect evolution paths, and double-sided quality compliance rates are integrated to generate a double-sided garment processing inspection report.

[0065] The feedforward control strategy library is a structured database pre-built based on historical quality data, process optimization cases, and fabric characteristic parameters. It includes a mapping relationship between "offset index range and dynamic process compensation scheme," enabling rapid matching of appropriate process adjustment strategies based on index characteristics. The offset index range can be a numerical range categorized by risk level (e.g., low-risk, medium-risk, high-risk range). The coordinated adjustment amount can be the synchronous adjustment value for seam tension, folding trajectory, and ironing parameters in the optimal compensation scheme, ensuring that multi-process parameter adjustments are mutually compatible and collaboratively improve quality. Personalized process control instructions can be customized for the current batch of fabric after material coupling coefficient correction, and can be directly issued to production equipment. The double-sided quality compliance rate can be the proportion of products in the current batch of garments whose core quality indicators such as double-sided appearance and structural stability meet design requirements, used to quantify the quality improvement effect.

[0066] Specifically, the processing chain of reversible garments is highly complex and coupled. Minor deviations in preceding processes can be transmitted and amplified through physical and technological connections between processes, ultimately causing significant appearance defects on both sides of the garment. However, traditional quality inspection methods are fundamentally insufficient in defect control and process adjustment. Traditional methods lack a systematic use of the global risk index and cannot map the quantified process deviation index into effective compensation measures. For example, when the risk index of interlayer slippage generated by the sewing process is high, traditional methods may only adjust the sewing parameters in isolation, ignoring the synergistic effects of contour symmetry in the folding process and stress distribution in the ironing process. This results in a one-sided compensation scheme that cannot fundamentally block the defect transmission chain. To address the above issues, this step inputs the garment comprehensive process deviation index (e.g., 6.8, in the medium-risk range) into the pre-built feedforward control strategy library. Combining the index's numerical characteristics and growth trend (e.g., rapid increase in the last three process stages), it matches the optimal dynamic process compensation scheme (e.g., adjusting seam tension by +5N and improving folding trajectory accuracy by 0.1mm). Then, it retrieves the material coupling coefficient (e.g., 0.85) and corrects the collaborative adjustment amount in the scheme (e.g., changing the seam tension adjustment amount to +4.25N), generating personalized process control instructions for the current batch of fabric. Finally, it integrates these instructions, the evolution path of key defects (e.g., "seam slippage between layers → asymmetrical folding contour"), and the double-sided quality compliance rate (e.g., 92%) to form a double-sided garment processing inspection report.

[0067] The test report provided in this embodiment integrates personalized process control instructions, critical defect evolution paths, and dual-sided quality compliance rates. It not only presents the quality results but also clarifies "how to adjust the process" and "where the defects come from." This provides production operators with direct equipment adjustment basis and management personnel with decision support for defect tracing and process optimization, thus solving the pain point of traditional reports that "emphasize results but neglect guidance."

[0068] Figure 3 This is a schematic diagram of the structure of a garment processing quality inspection platform provided in one embodiment of this application, as shown below. Figure 3 As shown, the garment processing quality inspection platform 300 of this embodiment includes: a factor analysis module 301, a defect transmission module 302, and a report generation module 303; The factor analysis module 301 is used to acquire a multi-source dataset of reversible garments. Based on the multi-source dataset, it performs cross-process collaborative analysis on interlayer dynamic interaction, internal wrapping morphology, and three-dimensional ironing stress field to generate a garment process quality base state information set. The defect transmission module 302 is used to construct a defect transmission network based on the garment process quality base state information set to quantitatively evaluate the evolution path and contribution of potential defects and generate a garment comprehensive process deviation index. The report generation module 303 is used to perform adaptive feedforward control strategy mapping based on the garment comprehensive process deviation index to generate a reversible garment processing inspection report.

[0069] Optionally, when the factor analysis module 301 performs cross-process collaborative analysis on the interlayer bonding balance, internal wrapping morphology, and three-dimensional ironing stress field based on the multi-source dataset of double-sided garments to generate a base state information set of garment process quality, it is specifically used for: the multi-source dataset of double-sided garments includes interlayer dynamic tension difference, interlayer relative displacement, wrapping three-dimensional morphology, interlayer temperature difference, and interlayer pressure field distribution; based on the interlayer dynamic tension difference, it analyzes the interlayer dynamic bonding effect during the sewing process to generate interlayer bonding balance effect information to characterize the double-sided bonding balance state at the seam; based on the interlayer relative displacement and the wrapping three-dimensional morphology, it analyzes the internal wrapping morphology during the folding and wrapping process to generate internal wrapping morphology information characterizing the stable state of the inner wrapping structure; based on the interlayer temperature difference and the interlayer pressure field distribution, it analyzes the coupling relationship of the double-sided stress field during the three-dimensional ironing process to generate three-dimensional ironing stress field information characterizing the stable state of the finished garment; and based on the interlayer dynamic interaction information, the internal wrapping morphology information, and the three-dimensional ironing stress field information, it generates the base state information set of garment process quality.

[0070] Optionally, when the factor analysis module 301 analyzes the interlayer dynamic bonding effect during the sewing process based on the interlayer dynamic tension difference and generates interlayer bonding balance information to characterize the double-sided bonding balance state at the seam, it specifically performs the following: decoupling the interlayer dynamic tension difference, separating the static deviation component characterizing the uneven force on the two fabrics, and the dynamic fluctuation component characterizing the asynchronous instantaneous fabric feeding during sewing; based on the static deviation component, quantifying the residual stress at the seam caused by the difference in shrinkage and stiffness during the sewing of heterogeneous fabrics, and analyzing its symmetry with the flatness of the front and back surfaces. The system generates a bidirectional flatness attenuation gradient to predict the stable range where the seam will not produce visible distortion on either side. Based on the dynamic fluctuation component, its frequency and amplitude characteristics are analyzed to detect periodic pauses and sudden advances caused by the mismatch in friction coefficients between the two fabrics and the presser foot / feed dog. The irreversible tensile deformation accumulated on one side of the fabric due to this phenomenon is assessed, and an interlayer slippage risk index is generated to correlate wavy seam defects that may occur on either side. Based on the bidirectional flatness attenuation gradient and the interlayer slippage risk index, the interlayer bonding balance information is constructed.

[0071] Optionally, when the factor analysis module 301 analyzes the internal wrapping morphology during the folding process based on the interlayer relative displacement and the three-dimensional shape of the wrapping, and generates internal wrapping morphology information representing the stable state of the inner wrapping structure, it is specifically used for: tracking and calculating the cumulative amount of the interlayer relative displacement; analyzing the asynchronicity and path deviation of the inner and outer fabrics in the three-dimensional folding motion; focusing on evaluating the mismatch of the rebound amount after folding due to the difference in stiffness of the two fabrics; generating double-sided contour symmetry; performing point cloud reconstruction and surface curvature analysis on the three-dimensional shape of the wrapping; detecting local morphological bulges / collapses caused by the stacking of internal sutures and functional components in a limited space; quantifying the visibility and distribution range of these bulges / collapses on the two outer surfaces; generating internal component migration risk; and constructing internal wrapping morphology information to represent that the wrapping structure does not produce non-designed contours on either side, based on the double-sided contour symmetry and the internal component migration risk.

[0072] Optionally, when the factor analysis module 301 analyzes the coupling relationship of the two-sided stress field during the three-dimensional ironing process based on the interlayer temperature difference and the interlayer pressure field distribution, and generates three-dimensional ironing stress field information characterizing the steady state of the garment, it is specifically used for: performing time-series analysis and gradient calculation on the interlayer temperature difference, solving the asymmetric conduction process when heat penetrates the garment interlayer composed of heterogeneous fabrics, quantifying the time difference and temperature difference between the front and back sides reaching the fiber glass transition temperature, and assessing the resulting risk of mismatch in two-sided heat setting memory, generating a two-sided heat memory balance; performing spatial vector decomposition on the interlayer pressure field distribution, identifying the balanced pressure distribution applied on the three-dimensional curved surface to achieve two-sided flatness, and then analyzing the shear stress that causes deformation due to the difference in stiffness between the two fabrics caused by this balanced pressure, and quantifying the tendency of this stress to cause shrinkage / bulging on either side when the garment is statically suspended, generating a two-sided stress manifestation difference; and constructing the three-dimensional ironing stress field information based on the two-sided heat memory balance and the two-sided stress manifestation difference.

[0073] Optionally, when the defect transmission module 302 generates a comprehensive garment process offset index by constructing a defect transmission network based on the garment process quality base state information set to quantify the evolution path and contribution of potential defects, it specifically performs the following: mapping the interlayer bonding balance information, the internal wrapping morphology information, and the three-dimensional ironing stress field information into garment processing network nodes, and constructing directed connection edges between nodes according to the physical process sequence and causal relationship of garment processing, thereby forming a defect transmission network; calculating the initial defect potential value of each node based on the quantitative indicators contained in the information of each garment processing network node; the initial defect potential value is the risk potential characterization value of the defect type and degree detected in the initial inspection; based on the defect transmission network, simulating the process of the defect potential of any node being transmitted and evolved to subsequent nodes along the directed edges, dynamically calculating the additional defect potential of downstream nodes stimulated by the upstream influence; aggregating the initial and additional defect potential of all nodes to generate the comprehensive garment process offset index, which is used to characterize the global risk level caused by the local deviation of a single or multiple processes to the double-sided appearance of the finished garment.

[0074] Optionally, when the defect propagation module 302 dynamically calculates the additional defect potential of downstream nodes affected by upstream influences during the process of the defect potential of any simulated node being transmitted and evolved along directed edges to subsequent nodes, it is specifically used for: defining a transmission relationship matrix of directed edges in the defect propagation network based on historical defect data and material physical properties; the transmission relationship matrix quantifies the influence intensity of the upstream node's unit defect potential on the downstream node, and defines the possible changes in the manifestation of defects during cross-process transmission; taking the defect potential of the preceding node as input, performing nonlinear mapping calculation through the transmission relationship matrix, iterating until the end of the network; introducing a critical enhancement effect during the iterative calculation process, that is, when the cumulative defect potential transmitted from upstream to this node exceeds its process tolerance threshold, triggering a nonlinear amplification of the current defect potential of this node; synchronously recording and updating the real-time defect potential state of each network node after being affected by all upstream influences, thereby generating a defect potential field that can reflect the overall picture of the dynamic propagation and evolution of defects in the processing link.

[0075] Optionally, when defining the directed edge transmission relationship matrix of the defect transmission network based on historical defect data and material physical properties, the defect transmission module 302 is specifically used for: calculating the conditional probability between the quantitative indicators of upstream process nodes and the quantitative indicators of downstream process nodes based on historical defect data, and initializing the basic influence weights of the corresponding directed edges in the transmission relationship matrix with these conditional probability values; extracting the stiffness ratio, thermal shrinkage rate difference, and interlayer friction coefficient difference of the paired fabrics based on material physical property test data, and mapping them to a dimensionless material coupling coefficient through a nonlinear fusion function; multiplying the material coupling coefficient as a modulation factor with the basic influence weights to physically correct the weights and generate the final transmission intensity of each directed edge in the transmission relationship matrix; normalizing the final transmission intensity into a probabilistic form to complete the construction of the transmission relationship matrix, wherein the value of any element of the transmission relationship matrix quantifies the probability and magnitude of the upstream node's unit defect potential triggering the downstream node to generate a defect.

[0076] Optionally, the report generation module 303 is specifically used for: inputting the garment comprehensive process offset index into a pre-built feedforward control strategy library, which stores the mapping relationship between different offset index intervals and dynamic process compensation schemes; matching the optimal compensation scheme from the feedforward control strategy library based on the numerical characteristics and growth trend of the current offset index, wherein the optimal compensation scheme includes the coordinated adjustment of seam tension, folding trajectory and ironing parameters; correcting the optimal compensation scheme in real time according to the material coupling coefficient to generate personalized process control instructions for the current batch of fabrics; and integrating the personalized process control instructions, key defect evolution paths and double-sided quality compliance rate to generate the double-sided garment processing inspection report.

[0077] The platform in this embodiment can be used to execute the methods of any of the above embodiments, and its implementation principle and technical effect are similar, so they will not be described again here.

Claims

1. A method for inspecting the quality of garment processing, characterized in that, include: Obtain a multi-source dataset of reversible garments. Based on the multi-source dataset of reversible garments, conduct cross-process collaborative analysis on the interlayer bonding balance, internal wrapping morphology and three-dimensional ironing stress field to generate a base state information set of garment process quality. Based on the aforementioned garment process quality ground state information set, a defect transmission network is constructed to quantitatively assess the evolution path and contribution of potential defects, thereby generating a garment comprehensive process deviation index. Based on the comprehensive garment process offset index, an adaptive feedforward control strategy mapping is performed to generate a reversible garment processing inspection report.

2. The method according to claim 1, characterized in that, Based on the multi-source dataset of reversible garments, a cross-process collaborative analysis is performed on the interlayer bonding balance, internal wrapping morphology, and three-dimensional ironing stress field to generate a basic state information set of garment process quality, including: The multi-source dataset of the double-sided clothing includes interlayer dynamic tension difference, interlayer relative displacement, wrapping three-dimensional shape, interlayer temperature difference and interlayer pressure field distribution; Based on the interlayer dynamic tension difference, the interlayer dynamic bonding effect during the suturing process is analyzed, and interlayer bonding balance effect information is generated to characterize the double-sided bonding balance state at the joint. Based on the interlayer relative displacement and the three-dimensional shape of the package, the internal package shape during the folding process is analyzed to generate internal package shape information that characterizes the stable state of the inner package structure. Based on the interlayer temperature difference and the interlayer pressure field distribution, the coupling relationship of the two-sided stress field in the three-dimensional ironing process is analyzed to generate three-dimensional ironing stress field information that characterizes the steady state of the garment. Based on the interlayer dynamic interaction information, the internal wrapping morphology information, and the three-dimensional ironing stress field information, the base state information set of garment process quality is generated.

3. The method according to claim 2, characterized in that, The process of analyzing interlayer dynamic bonding during suturing based on the interlayer dynamic tension difference, and generating interlayer bonding balance information to characterize the double-sided bonding balance state at the joint, includes: The interlayer dynamic tension difference is decoupled to separate the static deviation component that characterizes the uneven force on the two fabrics and the dynamic fluctuation component that causes the instantaneous asynchronous fabric feeding during sewing. Based on the static deviation component, the residual stress of the seam caused by the difference in shrinkage and stiffness during the sewing of heterogeneous fabrics is quantified, and its symmetric influence on the appearance flatness of the front and back sides is analyzed. A bidirectional flatness attenuation gradient is generated to predict the stable range of the seam where no visible distortion occurs on either side. Based on the dynamic fluctuation components, analyze their frequency and amplitude characteristics, detect the periodic pause and advance phenomenon caused by the mismatch of friction coefficients between the two fabrics and the presser foot / feed dog, assess the irreversible tensile deformation accumulated on one side of the fabric by this phenomenon, generate the interlayer slippage risk index, and use it to associate the wavy stitching defects that may occur on either side of the two fabrics. The interlayer bonding balance information is constructed based on the bidirectional smoothness attenuation gradient and the interlayer slip manifestation risk index.

4. The method according to claim 3, characterized in that, The process of analyzing the internal packaging morphology during the folding process based on the interlayer relative displacement and the three-dimensional shape of the package, and generating internal packaging morphology information representing the stable state of the inner packaging structure, includes: The relative displacement between the layers is tracked and the cumulative amount is calculated. The asynchronicity and path deviation of the inner and outer fabrics in the three-dimensional folding motion are analyzed. The focus is on evaluating the mismatch of the rebound amount after folding due to the difference in stiffness of the two fabrics, and generating the double-sided contour symmetry. The three-dimensional shape of the package is reconstructed by point cloud and the curvature of the surface is analyzed to detect the local morphological bulges / collapses caused by the internal suture bone positions and the stacking of functional components in a limited space. The visibility and distribution range of these bulges / collapses on the front and back outer surfaces are quantified to generate the risk of internal component migration. Based on the bi-face contour symmetry and the internal component migration risk, internal wrapping morphology information is constructed to characterize the wrapping structure so that it does not produce unintended contours on either side.

5. The method according to claim 4, characterized in that, The process of analyzing the coupling relationship of the two-sided stress fields during the three-dimensional ironing process based on the interlayer temperature difference and the interlayer pressure field distribution, and generating three-dimensional ironing stress field information characterizing the steady state of the garment, includes: The interlayer temperature difference is analyzed by time series and gradient calculation to solve the asymmetric conduction process when heat penetrates the garment interlayer made of heterogeneous fabrics, quantify the time difference and temperature difference between the front and back sides reaching the fiber glass transition temperature, and assess the risk of mismatch between the two sides' heat setting memory, and generate the two sides' heat setting memory balance. The interlayer pressure field distribution is decomposed into spatial vectors to identify the balanced pressure distribution applied to achieve double-sided flatness on the three-dimensional curved surface. Then, the shear stress that causes deformation is generated inside due to the difference in stiffness between the two fabrics caused by this balanced pressure is analyzed, and the tendency of this stress to cause shrinkage / bulging on either side when the garment is statically suspended is quantified, generating double-sided stress manifestation difference. Based on the uniformity of double-sided thermal memory and the difference in the manifestation of double-sided stress, the three-dimensional ironing stress field information is constructed.

6. The method according to claim 5, characterized in that, Based on the garment process quality ground state information set, a defect propagation network is constructed to quantitatively assess the evolution path and contribution of potential defects, generating a comprehensive garment process deviation index, including: The interlayer bonding balance information, the internal wrapping morphology information, and the three-dimensional ironing stress field information are mapped as garment processing network nodes, and directed connection edges between nodes are constructed according to the physical process sequence and causal relationship of garment processing, thereby forming a defect transmission network. Based on the quantitative indicators contained in the information of each garment processing network node, the initial defect potential value of each node is calculated. The initial defect potential value is the risk potential characterization value of the defect type and its degree detected in the initial inspection; Based on the aforementioned defect propagation network, the process of defect potential of any node being propagated and evolving along directed edges to subsequent nodes is simulated, and the additional defect potential of downstream nodes being stimulated by the influence of upstream nodes is dynamically calculated. By aggregating the initial and additional defect potential of all nodes, the comprehensive garment process offset index is generated to characterize the global risk level caused by local deviations in one or more processes to the double-sided appearance of the garment.

7. The method according to claim 6, characterized in that, The simulation describes the process of the defect potential of any node being transmitted and evolved along directed edges to subsequent nodes, dynamically calculating the additional defect potential of downstream nodes stimulated by the influence of upstream nodes, including: Based on historical defect data and material physical properties, a transmission relationship matrix of directed edges in the defect transmission network is defined. The transmission relationship matrix quantifies the influence intensity of the unit defect potential of the upstream node on the downstream node, and defines the possible changes in the manifestation of the defect when it is transmitted across processes. Using the defect potential of the preceding node as input, nonlinear mapping calculation is performed through the transfer relation matrix, and the calculation is iterated until the end of the network; In the iterative calculation process, a critical enhancement effect is introduced, that is, when the cumulative defect potential transmitted from upstream to this node exceeds its process tolerance threshold, a nonlinear amplification of the current defect potential of this node is triggered. The system synchronously records and updates the real-time defect potential state of each network node after being affected by all upstream factors, thereby generating a defect potential field that reflects the dynamic propagation and evolution of defects in the processing link.

8. The method according to claim 7, characterized in that, Based on historical defect data and material physical properties, the transmission relationship matrix of directed edges in the defect transmission network is defined, including: Based on historical defect data, the conditional probabilities between the quantitative indicators of upstream process nodes and the quantitative indicators of downstream process nodes are statistically analyzed, and the basic influence weights of the corresponding directed edges in the transmission relationship matrix are initialized using these conditional probability values. Based on material physical property test data, the stiffness ratio, thermal shrinkage rate difference, and interlayer friction coefficient difference of the paired fabrics are extracted and mapped to a dimensionless material coupling coefficient through a nonlinear fusion function. The material coupling coefficient is used as a modulation factor and multiplied with the basic influence weight. The weight is then physically corrected to generate the final transmission strength of each directed edge in the transmission relationship matrix. The final transmission intensity is normalized into a probabilistic form to complete the construction of the transmission relationship matrix. The value of any element in the transmission relationship matrix quantifies the probability and magnitude of a downstream node generating a defect due to the unit defect potential of the upstream node.

9. The method according to claim 8, characterized in that, The process of mapping an adaptive feedforward control strategy based on the garment's comprehensive process offset index to generate a double-sided garment processing inspection report includes: The garment comprehensive process offset index is input into a pre-built feedforward control strategy library, which stores the mapping relationship between different offset index ranges and dynamic process compensation schemes. Based on the numerical characteristics and growth trend of the current offset index, the optimal compensation scheme is matched from the feedforward control strategy library. The optimal compensation scheme includes the coordinated adjustment of the seam tension, folding trajectory and ironing parameters. The optimal compensation scheme is corrected in real time based on the material coupling coefficient to generate personalized process control instructions for the current batch of fabrics. The personalized process control instructions, key defect evolution paths, and double-sided quality compliance rates are integrated to generate the double-sided garment processing inspection report.

10. A garment processing quality inspection platform, characterized in that, The method applied to any one of claims 1-9 includes: The factor analysis module is used to acquire multi-source datasets of double-sided garments. Based on the multi-source datasets of double-sided garments, cross-process collaborative analysis is performed on the dynamic interaction between layers, the internal wrapping morphology and the three-dimensional ironing stress field to generate a set of basic information on garment process quality. The defect propagation module is used to generate a comprehensive garment process deviation index by constructing a defect propagation network based on the garment process quality ground state information set to quantitatively evaluate the evolution path and contribution of potential defects. The report generation module is used to perform adaptive feedforward control strategy mapping based on the comprehensive garment process offset index to generate a double-sided garment processing inspection report.

Images