Intelligent management system and method applied to clothing flexible customization

Abstract

The invention discloses an intelligent management system and method applied to clothing flexible customization, and relates to the technical field of clothing intelligent manufacturing, and the system comprises a multi-source color data acquisition module, a color space preprocessing module, a dynamic color space mapping module, an edge real-time calibration module, an equipment state monitoring module and a full-process data closed-loop module. Wherein the multi-source color data acquisition module constructs a cross-device color reference library, the color space preprocessing module standardizes color data and unifies a color gamut, the dynamic color space mapping module generates and optimizes a dynamic conversion matrix, and the edge real-time calibration module updates device parameters in combination with fabric characteristics. The equipment state monitoring module constructs a correlation model to analyze the equipment state and color deviation, and the full-process data closed-loop module stores data through a block chain and optimizes an algorithm. The system realizes a complete process from fabric color data acquisition, cross-device mapping calibration, device state monitoring to full-process data closed-loop management through cooperation of the modules.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent manufacturing of clothing, and specifically to an intelligent management system and method applied to flexible customization of clothing. Background Art

[0002] In the field of flexible customization of clothing, color management faces significant challenges in cross-device color space differences. In the traditional process, there are significant differences in the RGB color gamut coverage of cutting machines and the CMYK color gamut coverage of printing and dyeing machines, and there is a lack of a dynamic cross-device mapping mechanism, resulting in a high complaint rate of cross-device color differences for small-batch orders. When processing high-saturation colors, the color deviation problem is prominent, and multiple trial-and-error adjustments are required. At the same time, the existing static ICC profiles cannot adapt to device individual differences, aging, and changes in fabric characteristics. Changes in device status are likely to cause calibration parameters to fail, and manual recalibration takes a long time. Moreover, due to the significant differences in the light reflection characteristics of different fabrics, it is difficult for static profiles to make dynamic compensation, resulting in color differences generally exceeding the visually perceptible threshold, affecting product quality and production efficiency.

[0003] In addition, in the prior art, data in links such as cutting, printing and dyeing, and quality inspection are independent of each other, lacking means for real-time synchronization and reliable evidence storage. To trace quality problems, it is necessary to manually check multi-source data, which takes a long time to locate. Moreover, traditional databases can be tampered with, and it is difficult to define responsibilities. More critically, the existing solutions fail to build a closed-loop system of "data collection - dynamic calibration - quality traceability - algorithm evolution", lacking the ability of multi-modal data fusion, not introducing technologies such as edge computing and blockchain to solve industrial scenario requirements. Algorithm optimization relies on manual experience and cannot achieve adaptive evolution through historical data, making it difficult to cope with the frequent device adjustments and fabric replacements in flexible customization.

[0004] To solve the above problems, the present invention proposes an intelligent management system and method applied to flexible customization of clothing. Summary of the Invention

[0005] The purpose of the present invention is to provide an intelligent management system and method applied to flexible customization of clothing to solve the problems raised in the prior art.

[0006] To achieve the above purpose, the present invention provides the following technical solutions: An intelligent management system applied to the flexible customization of clothing, including a multi-source color data acquisition module, a color space preprocessing module, a dynamic color space mapping module, an edge real-time calibration module, a device status monitoring module, and a full-process data closed-loop module; the multi-source color data acquisition module uses a multispectral sensor and an industrial-grade RGB sensor to synchronously collect the spectral reflectance, RGB value, and ambient light parameters of the standard color card when the fabric is warehoused, and constructs a cross-device color reference library; the color space preprocessing module is used to standardize the original color data, and unify the gamut ranges of the RGB of the cutting machine and the CMYK of the printing and dyeing machine through the CIE XYZ space; the dynamic color space mapping module generates a dynamic conversion matrix of the device based on the iterative optimization algorithm, converts the RGB data of the cutting machine into the theoretical value of the CMYK of the printing and dyeing machine, and dynamically updates the matrix parameters by comparing the actual output error to achieve the mapping of the cross-device color space; the edge real-time calibration module is used to deploy edge computing nodes at the device end and update the device parameters in real time in combination with the physical characteristics of the fabric; the device status monitoring module is used to collect the device operation parameters in real time, combine the real-time detection data of the spectrophotometer, and construct a device status-color parameter correlation model; the full-process data closed-loop module is used to store the full-link color data through the blockchain, and optimize the dynamic conversion matrix and the compensation matrix generation algorithm based on the correlation model data of the device status monitoring module and the historical data stored in the blockchain through the adaptive learning engine optimization algorithm, forming a closed-loop management of quality traceability and continuous evolution.

[0007] The multi-source color data acquisition module includes a sensor array unit and a reference library construction unit; The sensor array unit is used to establish a multi-dimensional color data source in the fabric warehousing link, providing an original reference for the full-process color matching; specifically: first, use a multispectral sensor array to obtain the spectral reflectance curve S(λ) of the standard color card, and then use an industrial-grade RGB sensor to synchronously collect the tristimulus values R, G, B of the color card; the tristimulus values directly correspond to the input gamut space of the cutting machine, making the front-end data compatible with the device interface; finally, the ambient light perception sensor collects the ambient illuminance and the relevant color temperature in real time to form an ambient parameter vector L; The reference library construction unit is used to establish a cross-device color reference library with a time stamp and a device identifier; specifically: using the unique identifier ID of the fabric batch as an index, encapsulate S(λ), R, G, B, L, and the acquisition time stamp into a reference data structure D base ; then use the hash algorithm to generate a data fingerprint, store the hash value through the hardware security module, and deploy the reference library in the distributed file system, supporting the cutting and printing and dyeing links to retrieve the corresponding reference data through the ID to establish a cross-device color reference.

[0008] The color space preprocessing module includes a data standardization unit and a gamut unification unit; The data normalization unit is used to eliminate environmental noise and equipment deviation in the original data, and form standardized color data that can be processed across platforms. Specifically: perform illuminance normalization correction on the RGB data, and the formula is as follows: ; Wherein, are the tristimulus values after normalization respectively; E ref is the illuminance of the standard illuminant, and E is the ambient illuminance collected in real time. The color cast under different lighting conditions is eliminated through division operation; Then, perform extreme value normalization on the spectral data S(λ). By calculating the historical minimum response value and the maximum response value of the sensor at this wavelength, map the spectral reflectance S(λ) collected in real time to the [0,1] interval to obtain the normalized spectral data ; The gamut unification unit unifies and cuts the gamut descriptions of the RGB of the cutting machine and the CMYK of the printing and dyeing machine through the CIE XYZ standard space, and eliminates the incompatibility problems of the color spaces across devices. Specifically: First, map the RGB values output by the cutting machine through the conversion matrix M RGB of the device to the CIE XYZ space; this matrix is pre-calibrated based on the spectral tristimulus values of the CIE 1931 standard observer and contains 3×3 linear conversion coefficients. The matrix is as follows: ; Among them, the vector [X,Y,Z] represents the tristimulus value vector of the CIE XYZ color space, representing the coordinates of the color in the standardized color space; Then, based on the ink overprint principle, first calculate the lightness component Y through the ink overprint model, and the formula is as follows: ; Wherein, C, M, Y k , K are the concentrations of cyan, magenta, yellow, and black inks respectively, and the value range is [0,1]. Combine the chromaticity coordinates (x,y) to inversely deduce the X and Z values to obtain the CIE XYZ coordinates corresponding to CMYK.

[0009] The dynamic color space mapping module includes a matrix generation unit and an error feedback unit; The matrix generation unit is used to generate a dynamic conversion matrix exclusive to the device individual. Specifically: First, collect the RGB vector set {C RGB,i} output by the cutting machine and the theoretical CMYK output vector set of the printing and dyeing machine , where i is the sample index; then use the iterative least squares optimization algorithm to solve the optimal conversion matrix T ∗ , and the formula is as follows: ; Among them, T is a 4×3 conversion matrix to be optimized. By iteratively adjusting the matrix elements T mn , the Euclidean distance deviation between the RGB and CMYK vectors is minimized; T mn represents the conversion coefficient between the CMYK channel and the RGB channel in the dynamic conversion matrix T; Finally, a regularization term is introduced to suppress overfitting, where λ is a regularization parameter within a preset range, is the Frobenius norm of the matrix, enabling the model to have generalization ability on different devices; The error feedback unit dynamically adjusts the conversion matrix parameters based on the difference between the actual output data and the theoretical value; specifically: First, the actual color data of the printed and dyed fabric collected by the spectrophotometer is converted into the lightness component in the CIELAB space by using the method of non - linear transformation and the chromaticity component ; Then, through the k - th iteration matrix T k of the matrix generation unit, the RGB data of the cutting machine is converted into the theoretical CMYK value of the printing and dyeing machine, and then the theoretical components are obtained through CIELAB conversion; Then, the color difference between the measured value and the theoretical value is calculated according to the above conversion values, and the formula is as follows: ; When ΔE ab exceeds the preset threshold, the gradient descent algorithm is triggered to update the matrix to the (k + 1)-th version, and the formula is as follows: ; where, T k is the dynamic conversion matrix of the k - th iteration, η is the learning rate, ∇ is the gradient operator, and the matrix elements are adjusted through backpropagation until the color difference converges.

[0010] The edge real - time calibration module includes an edge computing node unit and a fabric characteristic compensation unit; The edge computing node unit is used to complete the real - time parsing of the dynamic conversion matrix and the update of device parameters on - site; specifically: Based on the edge server, a real - time operating system is deployed, and the dynamic conversion matrix T is subscribed through the industrial Internet of Things protocol and parsed into a set of device - executable parameters; and the matrix is parsed into a set of parameters executable by the device controller according to the implementation method described in the matrix generation unit, ensuring that the parameter format is compatible with the device interface; Then, a priority scheduling mechanism is adopted to assign the highest execution priority to matrix parsing and parameter update tasks. Specifically: The FreeRTOS real-time operating system is selected, and a hierarchical priority architecture is adopted. Matrix parameter update, matrix parsing and parameter generation, and communication protocol processing are respectively set as priority tasks of different levels. The priority inheritance mechanism is used to avoid priority inversion problems; The interrupt service program is optimized, the bottom half mechanism is used to process time-consuming tasks, dedicated memory resources are allocated for high-level tasks and bound to CPU cores to achieve resource isolation; Finally, the parameter writing is directly accessed through memory-mapped IO to the device register, and the double buffer mechanism is used to achieve lock-free parameter update; The fabric characteristic compensation unit generates dynamic compensation parameters for the physical characteristics of different fabrics to correct the change of light reflection characteristics caused by material differences. Specifically: First, the fabric physical characteristic data is collected in real time through sensors, and a feature vector F = [d, g, w] including fabric thickness d, glossiness g, and gram weight w is constructed; Among them, the thickness and gram weight can be collected by a weighing sensor and a laser rangefinder, and the glossiness is measured by a glossmeter; And the sensor type needs to be compatible with the sensor array of the multi-source color data acquisition module; Then, a long short-term memory network (LSTM) neural network is used to establish a fabric characteristic-compensation parameter mapping model; The input is the feature vector F, and the output is the compensation matrix K, which is used to adjust the device output curve; The model training data comes from the historical fabric characteristic data and the corresponding color deviation data stored in the full-process data closed-loop module; Specifically: First, a three-layer architecture including an input layer, two hidden layers with 64 neurons each and using the ReLU activation function, and an output layer is adopted. The continuously collected fabric thickness, glossiness, and gram weight feature vectors are organized as a time series input. The input layer receives a feature vector with a dimension of 3, and the output layer outputs a compensation matrix with a dimension of 4×4 corresponding to the CMYK color space conversion parameters through a fully connected layer; And the fabric characteristic data collected in real time is cleaned, standardized and constructed into a time series. The thickness is collected by a laser rangefinder and filtered, and the gram weight is collected by a weighing sensor and filtered by a Kalman filter; Then, the fabric characteristic sequence and the target compensation matrix corresponding to the color deviation conversion are extracted from the full-process data closed-loop module as training data, and a function including the mean square error and the CIEDE2000 color difference is used as the loss function, and the Adam optimizer and the learning rate scheduling strategy are used for training; During model inference, the predicted compensation matrix is de-normalized and reversible constraint processing is performed, and the output 4×4 compensation matrix is applied to adjust the device output curve; The sensor collects data at a preset frequency, updates the feature sequence every time a preset number of data points are collected and inputs the model to generate a compensation matrix, which is written into the device controller through an edge computing node, and the sliding window mechanism is used to update the historical window. When the change rate of the continuously predicted compensation matrix exceeds the threshold, the model is triggered for online fine-tuning, the normalization parameters are updated and the loss function weights are adjusted.

[0011] The device status monitoring module includes an operating parameter acquisition unit and a correlation model construction unit; The operating parameter acquisition unit is used to obtain the device operating status and color output data in real time, providing a multi-dimensional evidence chain for quality anomaly analysis; specifically: reading the printing and dyeing machine parameters in real time through the PLC data interface, including but not limited to temperature, pressure, speed, ink flow rate, and the acquisition frequency is synchronized with the industrial production rhythm; according to the on-line spectrophotometer installed at the discharge end of the printing and dyeing equipment, scanning the output fabric at a preset interval to obtain the real-time spectral reflectance S meas (λ), and retrieving the standard spectral reflectance S of the corresponding fabric batch from the cross-device color reference library of the full-process data closed-loop module base (λ), which is used to calculate the spectral difference vector ΔS(λ) to quantify the reflectance deviation at each wavelength. The formula is as follows: ; Then align the device parameters and spectral data according to the acquisition timestamp to generate a time-series marked dataset, where the reference library data is stored in the distributed file system indexed by the unique identifier ID of the fabric batch; The correlation model construction unit is used to identify the causal relationship between device status fluctuations and color deviations; specifically: first, use the CIEDE2000 color difference formula to calculate the comprehensive color deviation ΔE ab as the dependent variable ΔC of the model output; the calculation formula is as follows: ; Among them, is the component difference between the measured color and the reference color in the CIELAB space Then construct a multiple linear regression model, and the independent variables include device status parameters and spectral deviation characteristics ΔS(λ at equally spaced wavelength points within the visible spectral range j ); the device status parameters include temperature T m , pressure P m , speed V m and ink flow rate Q i , and the model structure is as follows: ; Among them, β0 is the intercept term, and β1-β 8+n are the regression coefficients of the respective independent variables, reflecting their contribution degrees to the color deviation; Finally, obtain from the full-process data closed-loop module the device status parameters, spectral deviation data, and the corresponding ΔE abFor the historical dataset, the least squares method is used to solve the regression coefficients, and the independent variables significantly affecting color deviation are screened through t-tests; when the predicted ΔC of the model exceeds the preset threshold, an alarm is triggered, and the priority of equipment parameter adjustment is determined according to the absolute value of the regression coefficients, where the regression coefficients reflect the contribution of the independent variables to color deviation.

[0012] The full-process data closed-loop module includes a blockchain evidence storage unit and an adaptive learning unit; The blockchain evidence storage unit is responsible for establishing an immutable full-link data ledger to achieve quality traceability and liability definition; specifically: First, a blockchain is constructed, and each block contains a data layer and a cryptographic layer. The data layer stores the conversion matrix, calibration timestamp, fabric characteristics, and full-process operation records of the equipment status. The cryptographic layer ensures data integrity through a hashing algorithm; then, the data is automatically uploaded to the blockchain through a smart contract, so that the operation records from fabric warehousing to finished product inspection are solidified in chronological order, supporting the retrieval of full-process color data through the fabric batch identifier to locate problem nodes; The adaptive learning unit is responsible for continuously optimizing using historical data-driven algorithms to form a closed-loop evolution ability; specifically: Establish a data storage architecture to integrate full-process historical data, and extract effective samples through feature engineering; effective samples refer to data samples that can truly reflect the correlation between equipment status, fabric characteristics, and color deviation after data cleaning and standardization preprocessing; optimize algorithm parameters based on an adaptive learning engine, and update the algorithm model by learning the correlation between equipment status, fabric characteristics, and color deviation in historical data, improving the accuracy of cross-device color mapping and fabric characteristic compensation, and forming a closed-loop management of collection - analysis - evolution.

[0013] An intelligent management method applied to flexible clothing customization includes the following steps: S1. Through a multispectral sensor and an industrial-grade RGB sensor, synchronously collect the spectral reflectance, RGB values, and ambient light parameters of the standard color card when the fabric is warehoused, and construct a cross-device color reference library with timestamps and device identifiers to provide an original reference for full-process color matching; S2. Standardize the original color data, eliminate environmental noise and device deviations through illuminance normalization correction and extreme value normalization, and then unify the gamut descriptions of RGB of the cutting machine and CMYK of the printing and dyeing machine through the CIE XYZ standard space to eliminate the incompatibility problem of cross-device color spaces; S3. Generate a dynamic conversion matrix exclusive to each device based on the iterative least squares optimization algorithm, convert the RGB data of the cutting machine into the theoretical CMYK value of the printing and dyeing machine, and dynamically update the matrix parameters using the gradient descent algorithm by comparing the actual output error collected by a spectrophotometer to achieve cross-device color space mapping; S4. Deploy edge computing nodes at the device end, parse the dynamic conversion matrix through a real-time operating system and update the device parameters. At the same time, combine the physical properties of fabric thickness, gloss, and gram weight collected in real time by sensors, and use the LSTM neural network to generate a compensation matrix to correct the change in light reflection characteristics caused by material differences, realizing the real-time update of device parameters; S5. Collect the operating parameters of the printing and dyeing machine in real time, such as temperature, pressure, speed, and ink flow rate. Combine the real-time spectral reflectance obtained by scanning the output fabric with a spectrophotometer, calculate the difference vector with the spectral library, and construct a multiple linear regression model to identify the causal relationship between device state fluctuations and color deviations, providing data support for quality anomaly analysis; S6. Store the full-link color data through the blockchain, including operation records such as the conversion matrix version, calibration timestamp, fabric characteristics, and device status, to form an immutable traceability ledger; at the same time, optimize the algorithm model using historical data, and analyze the correlation between device state, fabric characteristics, and color deviation through an adaptive learning engine to improve the accuracy of cross-device color mapping and fabric characteristic compensation, forming a closed-loop management of quality traceability and algorithm evolution.

[0014] Compared with the prior art, the beneficial effects of the present invention are: 1. The cross-device color consistency is significantly improved: By collecting multi-source color data to construct a cross-device color reference library and combining the iterative optimization algorithm of the dynamic color space mapping module, the dynamic mapping of the color spaces of the cutting machine and the printing and dyeing machine is realized. Compared with the traditional static calibration scheme, the present invention significantly reduces the cross-device color difference, meets the high-precision color matching requirements in clothing customization, and effectively reduces color difference complaints for small-batch orders.

[0015] 2. The dynamic calibration mechanism can cope with device and fabric changes and improve production flexibility: The edge real-time calibration module realizes the real-time parsing and parameter update of the dynamic conversion matrix through the edge computing node, and uses the neural network model to construct a fabric characteristic-compensation model to correct the light reflection deviation caused by material differences in real time. Compared with traditional static calibration, the device calibration efficiency is significantly improved, and it can adapt to the physical property changes of various fabrics, significantly improving the adaptability of flexible production to different device states and fabric types.

[0016] 3. A trusted closed-loop of full-process data is realized to quickly trace quality problems and define responsibilities: The blockchain evidence storage unit automatically solidifies the full-link data through a smart contract, uses encryption algorithms to ensure the immutability of the data, and supports the quick retrieval of full-process color data through a unique identifier. Combined with the device state-color correlation model, the key links and device parameters affecting color deviation can be accurately located, realizing the quick traceability of quality problems and the accurate definition of responsibilities, and improving the transparency and reliability of the production process. Description of the Drawings

[0017] Figure 1 It is the organizational structure diagram of an intelligent management system for flexible customization of clothing according to the present invention; Figure 2 It is the workflow diagram of an intelligent management system for flexible customization of clothing according to the present invention. Specific embodiments

[0018] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts shall fall within the protection scope of the present invention.

[0019] Embodiment: As Figure 1 - Figure 2 shown, the present invention provides a technical solution, An intelligent management system for flexible customization of clothing, including a multi-source color data acquisition module, a color space preprocessing module, a dynamic color space mapping module, an edge real-time calibration module, a device status monitoring module, and a full-process data closed-loop module; the multi-source color data acquisition module synchronously acquires the spectral reflectance, RGB values, and ambient light parameters of the standard color card through a multispectral sensor and an industrial-grade RGB sensor when the fabric is put into storage, and constructs a cross-device color reference library; the color space preprocessing module is used to perform standardization processing on the original color data, and unify the gamut ranges of the RGB of the cutting machine and the CMYK of the printing and dyeing machine through the CIE XYZ space; the dynamic color space mapping module generates a dynamic conversion matrix of the device based on an iterative optimization algorithm, converts the RGB data of the cutting machine into the theoretical value of the CMYK of the printing and dyeing machine, and dynamically updates the matrix parameters by comparing the actual output error to achieve the mapping of the cross-device color space; the edge real-time calibration module is used to deploy edge computing nodes at the device end and update the device parameters in real time in combination with the physical characteristics of the fabric; the device status monitoring module is used to collect the device operation parameters in real time, combine the real-time detection data of the spectrophotometer, and construct a device status-color parameter correlation model; the full-process data closed-loop module is used to store the full-link color data through the blockchain, and optimize the dynamic conversion matrix and the compensation matrix generation algorithm based on the correlation model data of the device status monitoring module and the historical data stored in the blockchain through an adaptive learning engine optimization algorithm, forming a closed-loop management of quality traceability and continuous evolution.

[0020] The multi-source color data acquisition module includes a sensor array unit and a reference library construction unit; The sensor array unit is used to establish a multi-dimensional color data source in the fabric warehousing process, providing an original reference for the full-process color matching; specifically: First, a multi-spectral sensor array is used to obtain the spectral reflectance curve S(λ) of the standard color card, and then an industrial-grade RGB sensor is used to synchronously collect the tristimulus values R, G, B of the color card; the tristimulus values directly correspond to the input color gamut space of the cutting machine, making the front-end data compatible with the device interface; finally, the ambient light perception sensor real-time collects the ambient illuminance and the relevant color temperature, forming an ambient parameter vector L; The reference library construction unit is used to establish a cross-device color reference library with time stamps and device identifiers; specifically: using the unique identifier ID of the fabric batch as an index, encapsulating S(λ), R, G, B, L and the collection time stamp into a reference data structure D base ; Then, a hash algorithm is used to generate a data fingerprint, the hash value is stored through a hardware security module, and the reference library is deployed in a distributed file system, supporting the cutting and printing processes to retrieve the corresponding reference data through the ID to establish a cross-device color reference.

[0021] The color space preprocessing module includes a data normalization unit and a color gamut unification unit; The data normalization unit is used to eliminate the ambient noise and device deviation in the original data, forming standardized color data that can be processed across platforms; specifically: performing illuminance normalization correction on the RGB data, and the formula is as follows: ; where, are the normalized tristimulus values respectively; E ref is the illuminance of the standard illuminant, E is the ambient illuminance collected in real time, and the color cast under different lighting conditions is eliminated through division operation; Then, extreme value normalization is performed on the spectral data S(λ). By calculating the historical minimum response value and the maximum response value of the sensor at this wavelength, the real-time collected spectral reflectance S(λ) is mapped to the [0,1] interval to obtain the normalized spectral data ; The color gamut unification unit unifies the color gamut descriptions of the cutting machine RGB and the printing machine CMYK through the CIE XYZ standard space, eliminating the incompatibility problem of the cross-device color space; specifically: First, the RGB value output by the cutting machine is mapped to the CIE XYZ space through the conversion matrix M of the device RGB ; This matrix is pre-calibrated based on the spectral tristimulus values of the CIE 1931 standard observer and contains 3×3 linear conversion coefficients. The matrix is as follows: ; where, the vector [X,Y,Z] represents the tristimulus value vector of the CIE XYZ color space, representing the coordinates of the color in the standardized color space; Then, based on the ink overprint principle, the lightness component Y is first calculated through the ink overprint model, and the formula is as follows: ; where C, M, Y k , and K are the concentrations of cyan, magenta, yellow, and black inks respectively, and the value range is [0, 1]. Combining the chromaticity coordinates (x, y), the X and Z values are inversely deduced to obtain the CIE XYZ coordinates corresponding to CMYK.

[0022] The dynamic color space mapping module includes a matrix generation unit and an error feedback unit; The matrix generation unit is used to generate a device-specific dynamic conversion matrix; specifically: first, collect the RGB vector set {C RGB,i} output by the cutting machine and the theoretical CMYK output vector set of the printing and dyeing machine , where i is the sample index; then use the iterative least squares optimization algorithm to solve the optimal conversion matrix T ∗ , and the formula is as follows: ; where T is the 4×3 conversion matrix to be optimized. By iteratively adjusting the matrix element T mn , the Euclidean distance deviation between the RGB and CMYK vectors is minimized; T mn represents the conversion coefficient between the CMYK channel and the RGB channel in the dynamic conversion matrix T; Finally, introduce the regularization term to suppress overfitting, where λ is the regularization parameter within a preset range, is the Frobenius norm of the matrix, enabling the model to have generalization ability on different devices; The error feedback unit dynamically adjusts the conversion matrix parameters based on the difference between the actual output data and the theoretical value; specifically: first, convert the actual color data of the printed and dyed fabric collected by the spectrophotometer into the lightness component and chromaticity component in the CIELAB space through a non-linear transformation method; then through the k-th iteration matrix T k of the matrix generation unit, convert the RGB data of the cutting machine into the theoretical CMYK value of the printing and dyeing machine, and then obtain the theoretical component through the CIELAB conversion; then calculate the color difference between the measured value and the theoretical value according to the above conversion values, and the formula is as follows: ; When ΔE ab exceeds the preset threshold, trigger the gradient descent algorithm to update the matrix to the (k + 1)-th version, and the formula is as follows: ; Among them, T k is the dynamic conversion matrix for the k-th iteration, η is the learning rate, ∇ is the gradient operator, and the matrix elements are adjusted through backpropagation until the color difference converges.

[0023] The edge real-time calibration module includes an edge computing node unit and a fabric characteristic compensation unit; The edge computing node unit is used to complete the real-time parsing of the dynamic conversion matrix and the update of device parameters on-site; specifically: deploy a real-time operating system based on the edge server, subscribe to the dynamic conversion matrix T through the industrial Internet of Things protocol, and parse it into a set of device-executable parameters; and parse the matrix into a set of parameters executable by the device controller according to the implementation method described by the matrix generation unit to ensure that the parameter format is compatible with the device interface; Then, a priority scheduling mechanism is adopted to assign the highest execution priority to the matrix parsing and parameter update tasks. Specifically: select the FreeRTOS real-time operating system, adopt a hierarchical priority architecture, set the matrix parameter update, matrix parsing and parameter generation, and communication protocol processing as priority tasks at different levels, and avoid the priority inversion problem through the priority inheritance mechanism; optimize the interrupt service program, adopt the bottom half mechanism to process time-consuming tasks, allocate dedicated memory resources for high-level tasks and bind the CPU core to achieve resource isolation; finally, directly access the device register through memory-mapped IO to implement parameter writing, and use the double buffer mechanism to achieve lock-free parameter update; The fabric characteristic compensation unit generates dynamic compensation parameters for the physical characteristics of different fabrics to correct the change in light reflection characteristics caused by material differences; specifically: first, the fabric physical characteristic data is collected in real time through sensors, and a feature vector F = [d, g, w] including fabric thickness d, glossiness g, and gram weight w is constructed; among them, the thickness and gram weight can be collected through a weighing sensor and a laser rangefinder sensor, and the glossiness is measured by a glossmeter; and the sensor type needs to be compatible with the sensor array of the multi-source color data acquisition module; Then, a fabric characteristic-compensation parameter mapping model is established using a long short-term memory network (LSTM) neural network; the input is the feature vector F, and the output is the compensation matrix K, which is used to adjust the device output curve; the model training data comes from the historical fabric characteristic data and the corresponding color deviation data stored in the full-process data closed-loop module; Specifically: First, a three-layer architecture including an input layer, two hidden layers with 64 neurons each using the ReLU activation function, and an output layer is adopted. The continuously collected fabric thickness, gloss, and gram weight feature vectors are organized into a time series input. The input layer receives a feature vector with a dimension of 3, and the output layer outputs a compensation matrix with a dimension of 4×4 corresponding to the CMYK color space conversion parameters through a fully connected layer; the fabric characteristic data collected in real time is cleaned, standardized, and a time series is constructed. The thickness is collected by a laser rangefinder and filtered, and the gram weight is collected by a load cell and filtered by a Kalman filter; then the fabric characteristic sequence and the target compensation matrix corresponding to the color deviation conversion are extracted from the full-process data closed-loop module as training data. A function including the mean square error and the CIEDE2000 color difference is used as the loss function, and the Adam optimizer and the learning rate scheduling strategy are used for training; during model inference, the predicted compensation matrix is de-normalized and reversible constraint processing is performed, and the output 4×4 compensation matrix is applied to adjust the device output curve; the sensor collects data at a preset frequency, updates the feature sequence every time a preset number of data points are collected and inputs it into the model to generate a compensation matrix, which is written into the device controller through an edge computing node. The sliding window mechanism is used to update the historical window. When the change rate of the continuously predicted compensation matrix exceeds the threshold, the model is triggered for online fine-tuning, the normalization parameters are updated, and the loss function weights are adjusted.

[0024] The device status monitoring module includes an operating parameter acquisition unit and a correlation model construction unit; The operating parameter acquisition unit is used to obtain the device operating status and color output data in real time, providing a multi-dimensional evidence chain for quality anomaly analysis; specifically: the printing and dyeing machine parameters are read in real time through the PLC data interface, including but not limited to temperature, pressure, speed, and ink flow, and the acquisition frequency is synchronized with the industrial production rhythm; according to the online spectrophotometer installed at the discharge end of the printing and dyeing equipment, the output fabric is scanned at a preset interval to obtain the real-time spectral reflectance S meas (λ), and the standard spectral reflectance S of the corresponding fabric batch is retrieved from the cross-device color reference library of the full-process data closed-loop module base (λ), which is used to calculate the spectral difference vector ΔS(λ) to quantify the reflectance deviation at each wavelength. The formula is as follows: ; Then, the device parameters and spectral data are aligned according to the acquisition timestamp to generate a dataset with time series tags, where the reference library data is stored in a distributed file system indexed by the unique identifier ID of the fabric batch. The correlation model construction unit is used to identify the causal relationship between device status fluctuations and color deviations; specifically: First, the CIEDE2000 color difference formula is used to calculate the comprehensive color deviation ΔE ab as the dependent variable ΔC of the model output; the calculation formula is as follows: ; Among them, is the component difference between the measured color and the reference color in the CIELAB space Then, a multiple linear regression model is constructed. The independent variables include the device state parameters and the spectral deviation characteristics ΔS(λ of equally spaced wavelength points within the visible spectral range j ); The device state parameters include temperature T m , pressure P m , speed V m and ink flow rate Q i . The model structure is as follows: ; Among them, β0 is the intercept term, and β1-β 8+n are the regression coefficients of their respective independent variables, reflecting their contribution degrees to the color deviation; Finally, a historical data set containing device state parameters, spectral deviation data, and the corresponding ΔE ab is obtained from the full-process data closed-loop module. The least squares method is used to solve the regression coefficients, and the independent variables that significantly affect the color deviation are screened through t-tests; When the predicted ΔC of the model exceeds the preset threshold, an early warning is triggered, and the adjustment priority of the device parameters is determined according to the absolute value ranking of the regression coefficients, where the regression coefficient reflects the contribution degree of the independent variable to the color deviation.

[0025] The full-process data closed-loop module includes a blockchain evidence storage unit and an adaptive learning unit; The blockchain evidence storage unit is responsible for establishing an immutable full-link data ledger to achieve quality traceability and responsibility definition; Specifically: First, a blockchain is constructed. Each block contains a data layer and a cryptographic layer. The data layer stores the conversion matrix, calibration timestamp, fabric characteristics, and the full-process operation records of the device state. The cryptographic layer ensures data integrity through a hash algorithm; Then, the data is automatically uploaded to the chain through a smart contract, so that the operation records from fabric warehousing to finished product inspection are solidified in chronological order, and it supports retrieving the full-process color data through the fabric batch identifier to locate the problem nodes; The adaptive learning unit is responsible for continuously optimizing using historical data-driven algorithms to form a closed-loop evolution ability; Specifically: Establish a data storage architecture to integrate the full-process historical data, and extract effective samples through feature engineering; Effective samples refer to data samples that can truly reflect the correlation between the device state, fabric characteristics, and color deviation after data cleaning and standardization preprocessing; Based on the adaptive learning engine, optimize the algorithm parameters, and update the algorithm model by learning the correlation between the device state, fabric characteristics, and color deviation in historical data, improving the accuracy of cross-device color mapping and fabric characteristic compensation, and forming a closed-loop management of collection - analysis - evolution.

[0026] An intelligent management method applied to flexible customization of clothing, comprising the following steps: S1. When the fabric is put into storage, synchronously collect the spectral reflectance, RGB values, and ambient light parameters of the standard color card through a multispectral sensor and an industrial-grade RGB sensor, and construct a cross-device color reference library with timestamps and device identifiers to provide an original reference for full-process color matching; S2. Perform standardization processing on the original color data, eliminate environmental noise and device deviations through illuminance normalization correction and extreme value normalization, and then unify the gamut descriptions of the RGB of the cutting machine and the CMYK of the printing and dyeing machine through the CIE XYZ standard space to eliminate the incompatibility problem of cross-device color spaces; S3. Generate a dynamic conversion matrix exclusive to each device based on the iterative least squares optimization algorithm, convert the RGB data of the cutting machine into the theoretical CMYK values of the printing and dyeing machine, and dynamically update the matrix parameters using the gradient descent algorithm by comparing the actual output error collected by the spectrophotometer to achieve the mapping of cross-device color spaces; S4. Deploy edge computing nodes at the device end, parse the dynamic conversion matrix through a real-time operating system and update the device parameters. At the same time, combine the physical properties of the fabric thickness, gloss, and gram weight collected in real time by the sensor, and use the LSTM neural network to generate a compensation matrix to correct the change in light reflection characteristics caused by material differences and achieve real-time update of device parameters; S5. Real-time collect the operating parameters such as the temperature, pressure, speed, and ink flow of the printing and dyeing machine, combine the real-time spectral reflectance obtained by scanning the output fabric with a spectrophotometer, calculate the difference vector with the reference library spectrum, construct a multiple linear regression model, identify the causal relationship between device state fluctuations and color deviations, and provide data support for quality anomaly analysis; S6. Store the full-link color data through the blockchain, including operation records such as the conversion matrix version, calibration timestamp, fabric characteristics, and device status, to form an immutable traceability ledger; at the same time, use historical data to optimize the algorithm model, analyze the correlation between device status, fabric characteristics, and color deviations through an adaptive learning engine, and improve the accuracy of cross-device color mapping and fabric characteristic compensation to form a closed-loop management of quality traceability and algorithm evolution.

[0027] In the flexible customized production of clothing, the system first constructs a cross-device color reference library through a multi-source color data acquisition module. Taking a certain silk fabric as an example, with a thickness of 0.15 mm, a glossiness of 92 GU, and a gram weight of 80 g / ㎡, a multi-spectral sensor array covering a wavelength range of 400 - 700 nm with an accuracy of ±0.5% is used to collect the spectral reflectance curve of the standard color card. For example, the reflectance at 550 nm is 68%. Synchronously, through an industrial-grade RGB sensor model IMX571 with a color gamut coverage of 95% sRGB, the tristimulus values R = 185, G = 210, B = 190 are obtained, and the illuminance of 5000 Lux and the color temperature of 6500 K are collected using an ambient light sensor. The data is processed by the reference library construction unit, and with the fabric batch ID "F20230516-001" as the index, a reference data structure containing a timestamp is generated. The data is ensured to be tamper-proof through the SHA-256 hash algorithm and stored in a distributed file system. The color space preprocessing module performs illuminance normalization and extreme value normalization on the original data, converts the RGB values to CIE XYZ space coordinates X = 45.2, Y = 52.8, Z = 39.5, and calculates the corresponding CMYK coordinates through an ink overprint model to achieve the unification of the RGB color gamut of the cutting machine and the CMYK color gamut of the printing and dyeing machine, with the coincidence degree increased from 62% to 89%.

[0028] The dynamic color space mapping module generates a device-specific conversion matrix based on the iterative least squares algorithm, converting the RGB data output by the cutting machine, such as R = 200, G = 150, B = 120, into the initial values of the CMYK theoretical values for the printing and dyeing machine, which are C = 30%, M = 45%, Y = 50%, K = 0%. Comparing with the measured data of the spectrophotometer, ΔE ab = 2.7. After 5 iterations through the gradient descent algorithm, the error between the theoretical value and the measured value converges to ΔE ab = 0.9. The edge real-time calibration module deploys the FreeRTOS system through an edge computing node, with a response time ≤ 100 ms. It analyzes the dynamic matrix and writes it into the printing and dyeing machine controller. At the same time, using a three-layer architecture of the LSTM neural network with 64 neurons in each hidden layer, it analyzes the fabric characteristic vectors d = 0.15 mm, g = 92 GU, w = 80 g / ㎡ collected in real time, generates a 4×4 compensation matrix, and adjusts the ink flow parameters, such as the cyan ink flow increasing from 22 mL / min to 25 mL / min, to make the actual output color difference ΔE* ab stable within 0.8. The device status monitoring module synchronously collects parameters such as the temperature of the printing and dyeing machine at 60℃ ± 1℃ and the pressure at 0.3 MPa ± 0.02 MPa, constructs a multiple linear regression model, identifies that the temperature contributes 42% to the color deviation, and automatically adjusts the temperature control system when the warning threshold is triggered.

[0029] The full-process data closed-loop module stores all-link data through blockchain. Each block contains the conversion matrix version such as V3.2, calibration timestamp 2025-05-16 14:30:05, fabric characteristics, and equipment status, and is automatically uploaded to the chain through a smart contract, improving the data storage efficiency by 70%. The adaptive learning engine extracts more than 100,000 historical data, including 2,000 fabric characteristics and 5,000 matrix update records. Through feature engineering, the key influencing factors are screened out as the glossiness weight of 35% and the temperature weight of 28%, optimizing the LSTM model parameters and improving the generation accuracy of the compensation matrix by 30%. When the change rate of the compensation matrix for 10 consecutive predictions exceeds 5%, online fine-tuning is triggered, updating the normalization parameters and adjusting the loss function weights to form a closed-loop of acquisition-calibration-learning-evolution. In practical applications, the system improves the pass rate of cross-device color consistency from 75% to 98%, shortens the equipment calibration time from 120 minutes per time to 15 minutes per time, reduces the fabric waste rate by 22%, and realizes precise color control for small-batch customized orders such as 50 pieces per order.

[0030] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic characteristics of the present invention. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be included in the present invention. Any reference signs in the claims should not be regarded as limiting the claims involved.

Claims

1. An intelligent management system applied to the flexible customization of clothing, characterized in that: It includes a multi-source color data acquisition module, a color space preprocessing module, a dynamic color space mapping module, an edge real-time calibration module, a device status monitoring module, and a full-process data closed-loop module; the multi-source color data acquisition module uses a multi-spectral sensor and an industrial-grade RGB sensor to synchronously collect the spectral reflectance, RGB values, and ambient light parameters of the standard color card when the fabric is stored in the warehouse, and constructs a cross-device color reference library; the color space preprocessing module is used to perform standardization processing on the original color data, and unify the gamut ranges of the RGB of the cutting machine and the CMYK of the printing and dyeing machine through the CIE XYZ space; the dynamic color space mapping module generates a dynamic conversion matrix for the device based on the iterative optimization algorithm, converts the RGB data of the cutting machine into the theoretical value of the CMYK of the printing and dyeing machine, and dynamically updates the matrix parameters by comparing the actual output error to achieve the mapping of the cross-device color space. The edge real-time calibration module is used to deploy edge computing nodes at the device end and update the device parameters in real time in combination with the physical characteristics of the fabric; the device status monitoring module is used to collect the device operation parameters in real time, combine the real-time detection data of the spectrophotometer, and construct a device status-color parameter association model; the full-process data closed-loop module is used to store the full-link color data through the blockchain, and through the adaptive learning engine optimization algorithm, based on the association model data of the device status monitoring module and the historical data stored in the blockchain, optimize the generation algorithms of the dynamic conversion matrix and the compensation matrix to form a closed-loop management of quality traceability and continuous evolution.

2. The intelligent management system for flexible customization of clothing according to claim 1, characterized in that: The multi-source color data acquisition module includes a sensor array unit and a reference library construction unit; The sensor array unit is used to establish a multi-dimensional color data source in the fabric storage link to provide an original reference for the full-process color matching; specifically: first, use a multi-spectral sensor array to obtain the spectral reflectance curve S(λ) of the standard color card, and then use an industrial-grade RGB sensor to synchronously collect the tristimulus values R, G, B of the color card; the tristimulus values directly correspond to the input gamut space of the cutting machine, making the front-end data compatible with the device interface; finally, the ambient light perception sensor collects the ambient illuminance and the relevant color temperature in real time to form an ambient parameter vector L. The reference library construction unit is used to establish a cross-device color reference library with timestamps and device identifiers; specifically: using the unique identifier ID of the fabric batch as an index, encapsulating S(λ), R, G, B, L, and the acquisition timestamp into a reference data structure D base ; then using a hashing algorithm to generate a data fingerprint, storing the hash value through a hardware security module, and deploying the reference library in a distributed file system to support retrieving corresponding reference data by ID in the cutting and printing processes to establish a cross-device color reference.

3. An intelligent management system applied to flexible customization of clothing according to claim 1, characterized in that: The color space preprocessing module includes a data standardization unit and a gamut unification unit; The data standardization unit is used to eliminate the environmental noise and device deviation in the original data to form standardized color data that can be processed across platforms; specifically: perform illuminance normalization correction on the RGB data, and the formula is as follows: ； Among them, are the tristimulus values after normalization; E ref is the illuminance of the standard illuminant, and E is the ambient illuminance collected in real time. The color cast under different lighting conditions is eliminated through division operations; Then, extreme value normalization is performed on the spectral data S(λ). By calculating the historical minimum response value and the maximum response value of the sensor at this wavelength, the spectral reflectance S(λ) collected in real time is mapped to the interval [0,1] to obtain the normalized spectral data ; The color gamut unification unit unifies the color gamut descriptions of RGB of the cutting machine and CMYK of the printing machine through the CIE XYZ standard space, eliminating the incompatibility problem of color spaces across devices; specifically: First, the RGB values output by the cutting machine are mapped to the CIE XYZ space through the conversion matrix M of the device. RGB This matrix is pre-calibrated based on the spectral tristimulus values of the CIE 1931 standard observer and contains 3×3 linear conversion coefficients. The matrix is as follows: ； Among them, the vector [X, Y, Z] represents the tristimulus value vector in the CIE XYZ color space, representing the coordinates of the color in the standardized color space; Then, based on the ink overprint principle, first calculate the lightness component Y through the ink overprint model, and the formula is as follows: ； Among them, C, M, Y k , and K are the concentrations of cyan, magenta, yellow, and black inks respectively, and the value range is [0, 1]. Combining with the chromaticity coordinates (x, y), the X and Z values are inversely deduced to obtain the CIE XYZ coordinates corresponding to CMYK.

4. An intelligent management system applied to flexible customization of clothing according to claim 1, characterized in that: The dynamic color space mapping module includes a matrix generation unit and an error feedback unit; The matrix generation unit is used to generate a dynamic conversion matrix exclusive to each device individual; specifically: First, collect the RGB vector set {C RGB,i} output by the cutting machine and the theoretical CMYK output vector set of the printing and dyeing machine , where i is the sample index; then use the iterative least squares optimization algorithm to solve the optimal conversion matrix T ∗ , and the formula is as follows: ； Among them, T is a 4×3 conversion matrix to be optimized, and the matrix elements T are iteratively adjusted mn , to minimize the Euclidean distance deviation between the RGB and CMYK vectors; T mn represents the conversion coefficients of the CMYK channels and the RGB channels in the dynamic conversion matrix T; Finally, a regularization term is introduced to suppress overfitting, where λ is a regularization parameter within a preset range, which is the Frobenius norm of the matrix, enabling the model to have generalization ability on different devices; The error feedback unit dynamically adjusts the conversion matrix parameters based on the difference between the actual output data and the theoretical value; specifically: First, the actual color data of the printed and dyed fabric collected by the spectrophotometer is converted into the lightness component in the CIELAB space by using a non-linear transformation method and the chromaticity component ; Then, through the k-th iteration matrix T k of the matrix generation unit, the RGB data of the cutting machine is converted into the theoretical CMYK value of the printing and dyeing machine, and then the theoretical components are obtained through CIELAB conversion ; Then, calculate the color difference between the measured value and the theoretical value according to the above conversion values. The formula is as follows: ； When ΔE ab exceeds the preset threshold, the gradient descent algorithm is triggered to update the matrix to the (k + 1)-th version, and the formula is as follows: ； Among them, T k is the dynamic transformation matrix for the k-th iteration, η is the learning rate, ∇ is the gradient operator, and the matrix elements are adjusted through backpropagation until the color difference converges.

5. The intelligent management system for flexible customization of clothing according to claim 4, characterized in that: The edge real-time calibration module includes an edge computing node unit and a fabric characteristic compensation unit; The edge computing node unit is used to complete the real-time parsing of the dynamic conversion matrix and the update of device parameters on-site; specifically: deploy a real-time operating system based on the edge server, subscribe to the dynamic conversion matrix T through the industrial Internet of Things protocol, and parse it into a set of device-executable parameters; and parse the matrix into a set of parameters executable by the device controller according to the implementation method described in the matrix generation unit, ensuring that the parameter format is compatible with the device interface; Then, adopt a priority scheduling mechanism to assign the highest execution priority to the matrix parsing and parameter update tasks. Specifically: select the FreeRTOS real-time operating system, adopt a hierarchical priority architecture, set matrix parameter update, matrix parsing and parameter generation, and communication protocol processing as priority tasks of different levels respectively, and avoid the priority inversion problem through the priority inheritance mechanism; optimize the interrupt service program, adopt the bottom half mechanism to process time-consuming tasks, allocate dedicated memory resources for high-level tasks and bind them to the CPU core to achieve resource isolation; finally, directly access the device register through memory-mapped IO to implement parameter writing, and use the double-buffer mechanism to achieve lock-free parameter update; The fabric characteristic compensation unit generates dynamic compensation parameters for the physical characteristics of different fabrics to correct the change in light reflection characteristics caused by material differences; specifically: first, collect fabric physical characteristic data in real time through sensors, and construct a feature vector F = [d, g, w] including fabric thickness d, glossiness g, and gram weight w; among them, the thickness and gram weight can be collected through a weighing sensor and a laser ranging sensor, and the glossiness is measured by a glossiness meter; and the sensor type needs to be compatible with the sensor array of the multi-source color data acquisition module; Then, use the long short-term memory network LSTM neural network to establish a fabric characteristic-compensation parameter mapping model; the input is the feature vector F, and the output is the compensation matrix K, which is used to adjust the device output curve; the model training data comes from the historical fabric characteristic data and the corresponding color deviation data stored in the full-process data closed-loop module; Specifically: First, a three-layer architecture including an input layer, two hidden layers with 64 neurons each using the ReLU activation function, and an output layer is adopted. The continuously collected fabric thickness, glossiness, and gram weight feature vectors are organized as time series inputs. The input layer receives feature vectors with a dimension of 3, and the output layer outputs a 4×4 compensation matrix corresponding to the CMYK color space conversion parameters through a fully connected layer; the fabric characteristic data collected in real time is cleaned, standardized, and a time series is constructed. The thickness is collected by a laser rangefinder and filtered, and the gram weight is collected by a load cell and filtered by a Kalman filter; then the fabric characteristic sequence and the target compensation matrix corresponding to the color deviation conversion are extracted from the full-process data closed-loop module as training data, and a function including the mean square error and the CIEDE2000 color difference is used as the loss function, and training is carried out using the Adam optimizer and the learning rate scheduling strategy; during model inference, the predicted compensation matrix is de-normalized and reversible constraint processing is performed, and the output 4×4 compensation matrix is applied to adjust the device output curve; the sensor collects data at a preset frequency, updates the feature sequence every time a preset number of data points are collected and inputs it into the model to generate a compensation matrix, which is written into the device controller through an edge computing node, and a sliding window mechanism is used to update the historical window. When the change rate of the compensation matrix predicted continuously for multiple times exceeds the threshold, the model is triggered for online fine-tuning, the normalization parameters are updated, and the loss function weights are adjusted.

6. The intelligent management system for flexible customization of clothing according to claim 1, wherein: The device status monitoring module includes an operating parameter collection unit and a correlation model construction unit; The operation parameter acquisition unit is used to obtain the device operation status and color output data in real time, providing a multi-dimensional evidence chain for quality anomaly analysis; specifically: reading the printing and dyeing machine parameters in real time through the PLC data interface, including but not limited to temperature, pressure, speed, ink flow rate, and the acquisition frequency is synchronized with the industrial production rhythm; according to the online spectrophotometer installed at the discharging end of the printing and dyeing equipment, scanning the output fabric at a preset interval to obtain the real-time spectral reflectance S meas (λ), and retrieving the standard spectral reflectance S base (λ) of the corresponding fabric batch from the cross-device color reference library of the full-process data closed-loop module, which is used to calculate the spectral difference vector ΔS(λ) to quantify the reflectance deviation at each wavelength. The formula is as follows: ； Then, the device parameters and spectral data are aligned according to the collection timestamp to generate a dataset with time series tags, where the reference library data is stored in a distributed file system indexed by the unique identifier ID of the fabric batch; The correlation model construction unit is used to identify the causal relationship between equipment state fluctuations and color deviations; specifically: First, the CIEDE2000 color difference formula is used to calculate the comprehensive color deviation ΔE ab As the dependent variable ΔC of the model output; the calculation formula is as follows: ； Among them, is the component difference between the measured color and the reference color in the CIELAB space Then, a multiple linear regression model is constructed. The independent variables include the device state parameters and the spectral deviation characteristics ΔS(λ j ) within the visible spectral range at equally spaced wavelength points; the device state parameters include temperature T m , pressure P m , speed V m and ink flow rate Q i . The model structure is as follows: ； Among them, β0 is the intercept term, and β1-β 8+n are the regression coefficients of their respective independent variables, reflecting their contribution degrees to the color deviation; Finally, obtain the historical data set containing device status parameters, spectral deviation data, and corresponding ΔE from the full-process data closed-loop module ab . Use the least squares method to solve the regression coefficients, and screen the independent variables that significantly affect color deviation through t-tests. When the predicted ΔC of the model exceeds the preset threshold, trigger an alarm, and determine the priority of device parameter adjustment according to the absolute value of the regression coefficients, where the regression coefficients reflect the contribution of the independent variables to color deviation.

7. An intelligent management system applied to the flexible customization of clothing according to claim 1, characterized in that: The full-process data closed-loop module includes a blockchain evidence storage unit and an adaptive learning unit; The blockchain evidence storage unit is responsible for establishing an immutable full-link data ledger to achieve quality traceability and responsibility definition; specifically: First, a blockchain is constructed. Each block includes a data layer and a cryptographic layer. The data layer stores the conversion matrix, calibration timestamp, fabric characteristics, and full-process operation records of the device status. The cryptographic layer ensures data integrity through a hashing algorithm; Then, the data is automatically uploaded to the blockchain through a smart contract, and the operation records from fabric warehousing to finished product inspection are solidified in chronological order, supporting the retrieval of full-process color data through the fabric batch identifier to locate problem nodes; The adaptive learning unit is responsible for continuously optimizing using historical data-driven algorithms to form a closed-loop evolution ability; specifically: A data storage architecture is established to integrate full-process historical data, and effective samples are extracted through feature engineering; effective samples refer to data samples that can truly reflect the correlation between device status, fabric characteristics, and color deviation after data cleaning and standardized preprocessing; Based on the adaptive learning engine, the algorithm parameters are optimized. By learning the correlation between device status, fabric characteristics, and color deviation in historical data, the algorithm model is updated to improve the accuracy of cross-device color mapping and fabric characteristic compensation, forming a closed-loop management of collection - analysis - evolution.

8. An intelligent management method applied to the flexible customization of clothing, which is applied to an intelligent management system for the flexible customization of clothing described in claims 1-7, characterized in that: It includes the following steps: S1. Use a multispectral sensor and an industrial-grade RGB sensor to synchronously collect the spectral reflectance, RGB values, and ambient light parameters of the standard color card when the fabric is put into storage, and construct a cross-device color reference library with timestamps and device identifiers to provide an original reference for the full-process color matching; S2. Standardize the original color data. Eliminate environmental noise and device deviations through illuminance normalization correction and extreme value normalization, and then unify the gamut descriptions of the RGB of the cutting machine and the CMYK of the printing and dyeing machine through the CIE XYZ standard space to eliminate the incompatibility problem of the cross-device color space; S3. Generate a dynamic conversion matrix exclusive to each device based on the iterative least squares optimization algorithm, convert the RGB data of the cutting machine into the theoretical CMYK values of the printing and dyeing machine, and use the gradient descent algorithm to dynamically update the matrix parameters by comparing the actual output error collected by the spectrophotometer to achieve the mapping of the cross-device color space; S4. Deploy edge computing nodes at the device end, parse the dynamic conversion matrix through a real-time operating system and update the device parameters. At the same time, combine the physical characteristics of the fabric thickness, gloss, and gram weight collected by the sensor in real time, and use the LSTM neural network to generate a compensation matrix to correct the change in the light reflection characteristics caused by the material difference, and achieve real-time update of the device parameters; S5. Collect the operating parameters such as the temperature, pressure, speed, and ink flow of the printing and dyeing machine in real time, combine the real-time spectral reflectance obtained by scanning the output fabric with a spectrophotometer, calculate the difference vector with the spectrum of the reference library, and construct a multiple linear regression model to identify the causal relationship between the device state fluctuation and the color deviation, providing data support for the quality anomaly analysis; S6. Store the full-link color data through the blockchain, including operation records such as the conversion matrix version, calibration timestamp, fabric characteristics, and device status, to form an immutable traceability ledger; at the same time, use historical data to optimize the algorithm model, analyze the correlation between the device state, fabric characteristics, and color deviation through an adaptive learning engine, and improve the accuracy of cross-device color mapping and fabric characteristic compensation, forming a closed-loop management of quality traceability and algorithm evolution.

Images