Dye sample collaborative proofing system based on network interconnection

Abstract

The invention relates to the technical field of dyeing and finishing automation and color science, and discloses a dye sample collaborative proofing system based on network interconnection, which comprises a collaborative process management platform, a networked automatic proofing execution station and a multi-mode appearance remote detection terminal, the collaborative process management platform issues a digital chemical single-drive execution station to automatically prepare a physical sample and instructs a remote detection terminal to collect a hyperspectral data stream of the sample so as to realize reliable remote collaborative evaluation. When evaluation is rejected, an intelligent correction algorithm module in the platform can automatically calculate the formula correction amount through a formula reverse solution model according to hyperspectral data in a problem state, data-driven closed-loop correction is achieved, and the problems that an existing proofing process is low in efficiency, remote evaluation is distorted, and formula correction depends on manpower are solved. Manual trial and error are replaced, and the proofing efficiency and the success rate are improved.

Description

Technical Field

[0001] This invention relates to the fields of dyeing and finishing automation and color science technology, specifically to a network-connected collaborative sampling system for small dye samples. Background Technology

[0002] In industries such as textiles, coatings, and plastics, the prototyping of dye samples is an indispensable and crucial step before product development and mass production. Its purpose is to verify whether the formulation accurately matches the target color and to obtain approval from customers or decision-makers by preparing small batches of physical samples.

[0003] Existing sampling processes typically rely heavily on manual operation and physical handling. Technicians mix dyes based on experience or preliminary calculations, manually preparing small physical samples. These samples must then be delivered to clients or designers in other locations via courier or other physical means for evaluation. This process is not only time-consuming, severely slowing down the overall product launch schedule, but also carries the risk of sample damage or loss during transportation.

[0004] In the evaluation process, the results are highly susceptible to subjective and environmental factors. Different observers may perceive the color of the same sample differently under varying lighting conditions, leading to inconsistent evaluation standards and inefficient communication. To overcome the lag in physical transfer, the industry often uses photographs or videos for remote communication, but this method has serious limitations. Due to significant differences between camera sensors, image compression algorithms, and terminal display devices, there is a substantial and uncontrollable color difference between the colors presented in digital images and the true colors of the physical samples. This makes remote evaluation based on ordinary digital images extremely unreliable, often resulting in misjudgments and unnecessary rework. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a network-connected collaborative sampling system for dye samples, which solves the problems of low efficiency caused by physical sample transfer, color information distortion during remote collaborative evaluation, and the high dependence on human experience and lack of accurate data closure in the formulation correction process.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a network-connected collaborative sampling system for small-scale dye samples, comprising:

[0007] The collaborative process management platform is used to generate digital chemical orders containing formulas and send remote inspection commands to multimodal appearance remote inspection terminals.

[0008] The networked automated sampling execution station is connected to the collaborative process management platform to receive the digital work order and automatically complete the preparation of physical samples according to the digital work order.

[0009] The multimodal appearance remote detection terminal is communicatively connected to the collaborative process management platform. It is used to receive the remote exploration command, collect the hyperspectral data stream of the physical sample according to the remote exploration command, and transmit the hyperspectral data stream to the collaborative process management platform for users to conduct collaborative evaluation.

[0010] Preferably, the collaborative process management platform includes a virtual sample and physical rendering module, which is used to generate an interactive three-dimensional virtual sample based on the initial formula before generating the digital work order, so that the user can perform interactive virtual sample testing of the initial formula, and after the user confirms, the initial formula is fixed in the digital work order.

[0011] Preferably, the collaborative process management platform includes an intelligent correction algorithm module, which, in response to the user's rejection decision, captures the hyperspectral data specified by the user under the problem state, calculates the correction amount of the dye concentration based on the hyperspectral data under the problem state through a formula reverse solving model, generates a corrected formula, and generates a new digital work order based on the corrected formula.

[0012] Preferably, the formulation reverse solution model calculates the correction amount of the dye concentration by constructing a Jacobian matrix representing the sensitivity of spectral reflectance to dye concentration, and solving the matrix equation based on the difference between the hyperspectral data and the target spectrum under the problem state to obtain the correction amount of the dye concentration.

[0013] Preferably, the multimodal appearance remote inspection terminal includes:

[0014] A programmable multi-angle light source array, used to change the type of light source and the incident angle illuminating the physical sample according to the remote exploration command;

[0015] A hyperspectral imaging and robotic arm system is used to adjust the observation angle of the physical sample according to the remote exploration command and to acquire the hyperspectral data stream.

[0016] Preferably, the collaborative process management platform is configured to convert the user's interface operations into remote exploration commands in real time, so as to drive the programmable multi-angle light source array and the hyperspectral imaging and robotic arm system to respond synchronously.

[0017] Preferably, the networked automated proofing station includes:

[0018] A high-precision automatic solution preparation unit is used to prepare dye solution according to the formula in the digital work order;

[0019] An automated dyeing and post-processing unit is used to dye fabric samples according to the process parameters in the digital work order;

[0020] An automatic sample transfer unit is used to transfer fabric samples between a high-precision automatic solution preparation unit and an automated dyeing and post-treatment unit.

[0021] Preferably, the collaborative process management platform also includes a data and model management module, which is used to store the full-process data of each sampling cycle in a structured manner to form a digital twin archive, and to use the accumulated digital twin archive to perform background iterative optimization of the formula reverse engineering model.

[0022] Preferably, the collaborative process management platform is configured to synchronously process the received hyperspectral data stream during the collaborative evaluation to simultaneously present the subjective visual appearance of the physical sample and its corresponding objective colorimetric value on the user interface.

[0023] A network-connected collaborative sampling method for dye samples includes the following steps:

[0024] An interactive 3D virtual sample is generated based on the initial formula, and a digital chemical invoice containing the initial formula is generated after user confirmation.

[0025] The digital work order is sent to the networked automated sampling execution station to automatically complete the preparation of physical samples;

[0026] The multimodal appearance remote inspection terminal adjusts the light source and observation angle according to the remote exploration command issued by the user, and collects the hyperspectral data stream of the physical sample for the user to conduct collaborative evaluation;

[0027] In response to the user's rejection decision, the system captures hyperspectral data in the problematic state, calculates the correction amount of the dye concentration using a formulation inverse solving model to generate a corrected formulation, and returns to execute the preparation steps of the physical sample based on the corrected formulation.

[0028] This invention provides a network-connected collaborative sampling system for small-scale dye samples. It offers the following advantages:

[0029] 1. This invention, by setting up an interactive virtual sample preparation step and integrating virtual sample preparation and physical rendering modules on a collaborative workflow management platform, enables users to perform high-fidelity visual effect estimation and screening of initial formulations before physical sample preparation. This transfers a significant amount of physical trial-and-error in traditional prototyping to the digital domain, effectively avoiding physical input for obviously unqualified formulations, thereby reducing the number of initial prototyping rounds, saving on material costs such as dyes and substrates, and shortening the initial product development cycle.

[0030] 2. This invention solves the problem of distorted appearance information caused by differences in equipment and environment in traditional remote assessments by employing a multimodal remote appearance inspection terminal and combining it with the synchronous processing capabilities of a collaborative process management platform. This terminal provides an objective and comprehensive physical data foundation for remote assessments by collecting hyperspectral data streams under controlled conditions, enabling efficient and reliable remote collaborative assessments.

[0031] 3. This invention constructs a closed-loop process for intelligent formula self-correction, transforming the formula correction process after sampling failure from relying on manual experience to automated, data-driven, and precise calculation. This greatly improves the efficiency and success rate of sampling iteration. When the user makes a rejection decision, the intelligent correction algorithm module in the system can automatically capture hyperspectral data under the problem state and directly calculate the correction amount of dye concentration using the formula reverse solution model. This automated closed-loop correction mechanism shortens the reaction time from evaluation to re-production, thereby accelerating the entire sampling process. Attached Figure Description

[0032] Figure 1 This is a schematic diagram of the overall system architecture of the present invention;

[0033] Figure 2 This is a schematic diagram of the multimodal appearance remote inspection terminal structure of the present invention;

[0034] Figure 3 This is an overall flowchart of the method of the present invention;

[0035] Figure 4 This is a schematic diagram of the remote dynamic evaluation interface of the present invention;

[0036] Figure 5 This is a flowchart of the intelligent formula self-correction algorithm of the present invention. Detailed Implementation

[0037] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0038] Please see the appendix Figure 1 - Appendix Figure 5 The present invention provides a network-connected collaborative sampling system for dye samples, which may include: a collaborative process management platform, a networked automatic sampling execution station, and a multimodal appearance remote inspection terminal.

[0039] The collaborative process management platform, the networked automated prototyping station, and the multimodal appearance remote inspection terminal establish a communication connection via a network. As the control and computational core of the system, the collaborative process management platform is responsible for sending digital work orders containing formulas and process parameters to the networked automated prototyping station and receiving its task status feedback. The collaborative process management platform is also responsible for receiving high-dimensional data streams of sample appearance from the multimodal appearance remote inspection terminal and sending it real-time control commands for light sources and viewing angles.

[0040] The collaborative process management platform can be deployed in the cloud or on a local server, and it can specifically include: a user interaction and process engine module, a virtual sample and physical rendering module, an intelligent correction algorithm module, and a data and model management module.

[0041] The user interaction and workflow engine module provides a graphical user interface for users to create, query, and manage prototyping tasks. This module has a built-in workflow engine that automatically flows and updates the status of prototyping tasks across different stages, such as formulation design, virtual prototyping, physical prototyping, remote evaluation, and formulation modification, based on preset business rules.

[0042] A virtual sample and physically rendered module is used to generate interactive virtual samples before physical prototyping. This module receives an initial formula input by the user and invokes a built-in color science model to calculate predicted spectral reflectance data. In one embodiment, this model can be based on the Kubelka-Munk theory. For opaque samples of sufficient thickness, their spectral reflectance... With absorption coefficient and scattering coefficient The relationship is:

[0043] ;

[0044] For a mixture containing n dyes, its K / S value It can be calculated using the following formula:

[0045] ;

[0046] in:

[0047] The wavelength of light;

[0048] For the first The concentration of the dye;

[0049] and The first The absorption and scattering coefficients of a dye at a unit concentration;

[0050] and These are the absorption and scattering coefficients of the fabric substrate, respectively.

[0051] After calculating the predicted spectral reflectance data, the module hands it over to the physical rendering engine, which, combined with the virtual light source and viewing angle selected by the user, renders the appearance of the virtual sample on the interface.

[0052] The intelligent correction algorithm module automatically calculates a compensation formulation when the remote evaluation result is unacceptable. This module is triggered by the built-in formulation inverse solving model. Based on the captured spectral data under the problem state and the user's correction intention, it calculates the correction amount of the dye concentration to generate a new round of iterative formulation.

[0053] The data and model management module, employing a structured database, is responsible for storing and managing all data generated during system operation. This data includes formulations, process parameters, hyperspectral data, user interaction commands, and evaluation decisions, collectively forming a digital profile for each sample. This module also utilizes the accumulated digital profile data to perform background training and parameter optimization of the color science model in the virtual sample and physical rendering module, and the algorithm model in the intelligent correction algorithm module.

[0054] A networked automated sample preparation station is a hardware entity that enables automated physical sample preparation. Specifically, it may include: a central control and network interface unit, a high-precision automated solution preparation unit, an automated staining and post-processing unit, and an automated sample transfer unit.

[0055] The central control and network interface unit, with a built-in programmable logic controller (PLC) or embedded computer, receives digital work orders from the collaborative process management platform via a network interface (such as an Ethernet interface). This unit is responsible for parsing the formula and process data in the work orders and converting them into precise control instructions for other units.

[0056] The high-precision automatic dispensing unit includes multiple micro-pumps, solenoid valves, and pipelines connected to different dye and auxiliary agent reservoirs. It can employ a high-precision peristaltic pump or syringe pump array. A central control and network interface unit, based on the formulation data, controls the corresponding pumps and valves to inject precise doses of dye and auxiliary agents into the mixing container, completing the dispensing process.

[0057] The automated staining and post-processing unit includes one or more laboratory sample preparation machines, whose control system is connected to a central control and network interface unit. Based on the analyzed process parameters, this unit automatically sets and executes the temperature profile, running time, and stirring speed for the staining process.

[0058] The automated sample transfer unit can be a multi-axis robotic arm or a linear conveyor system. This unit is responsible for transferring and loading / unloading the fabric samples to be dyed sequentially between the solution preparation unit, dyeing unit, drying unit, and finally the testing position without human intervention.

[0059] The multimodal appearance remote inspection terminal is a hardware entity that enables remote dynamic sensing. Specifically, it may include: a programmable multi-angle light source array, a hyperspectral imaging and robotic arm system, and a data acquisition and transmission unit.

[0060] The programmable multi-angle light source array consists of various types of LEDs (such as analog D65, CWF, A light sources, etc.) distributed on a hemispherical or arc-shaped support around the sample stage. This array receives instructions from a collaborative process management platform, allowing it to instantly switch light source types and illuminate LEDs at different locations to change the incident angle of the light.

[0061] The hyperspectral imaging and robotic arm system includes a hyperspectral camera and a multi-degree-of-freedom robotic arm or goniometer. The hyperspectral camera is used to acquire spectral reflectance information of the sample surface in a continuous narrow wavelength band. The robotic arm or goniometer carries the hyperspectral camera and, according to remote commands, precisely adjusts the camera's observation zenith and azimuth angles, thereby enabling multi-angle observation of the sample.

[0062] The data acquisition and transmission unit is responsible for integrating the data cubes acquired by the hyperspectral camera, adding current light source status parameters and robotic arm angle parameters, and forming a timestamped data frame. This unit transmits the data frame in real time to the collaborative process management platform 100 via a network interface in the form of streaming media for remote user evaluation.

[0063] This invention provides a network-based collaborative sampling method for dye samples, which may include the following steps:

[0064] S100 receives sampling tasks submitted by users through the collaborative process management platform and predicts the initial formula based on the color science model; according to the initial formula, it generates an interactive 3D virtual sample through the physical rendering engine for users to conduct interactive virtual sample testing; after user confirmation, it generates a digital chemical bill of materials containing the initial formula.

[0065] S101, the user initiates a sampling task through the client interface of the collaborative workflow management platform. The target color can be initialized in several ways during task creation. In one implementation, the user can directly input the instrument measurement value of the target color, which can be a CIELAB colorimetric value or a spectral reflectance curve under a preset standard light source. In another implementation, the user can place a physical standard sample on an external spectrophotometer for measurement and upload the obtained spectral reflectance data to the platform. After receiving the target color data, the platform creates a new task record with a unique identifier in its internal database.

[0066] S102, after receiving the target color data, the collaborative process management platform invokes its built-in virtual sample and physical rendering module to predict one or more initial formulations for the user. This prediction process is based on a preset color science model. In a specific implementation, the color science model can be a color prediction model based on Kubelka-Munk theory. This model describes the relationship between dye concentration and spectral reflectance using the following formula:

[0067] ;

[0068] in:

[0069] The wavelength of light;

[0070] For dye and fabric substrate mixture at wavelength The corresponding K / S value;

[0071] For the first in the formula The concentration of the dye;

[0072] and The first Absorption coefficient and scattering coefficient of the dye at unit concentration;

[0073] and These are the absorption coefficient and scattering coefficient of the fabric substrate itself, respectively.

[0074] By solving the above formula, we can obtain the initial dye concentration combination that theoretically minimizes the color difference with the target color data, i.e., the initial formulation. The calculated theoretical spectral reflectance across the entire wavelength range corresponds to this initial formulation. This will be used as input for the next step.

[0075] S103, Virtual Sample and Physical Rendering Module, uses the theoretical spectral reflectance calculated in the previous step. The data is processed by its built-in physically based rendering engine, which can be GPU-based and uses ray tracing or physically based rasterization techniques to combine spectral data with a pre-defined microstructure model of the fabric surface, rendering a three-dimensional virtual sample that can be interacted with in real time on the user interface. Users can modify the digital environment of the virtual sample in real time using a virtual controller on the interface, including switching the type of virtual light source, for example, from a standard D65 light source to a CWF or A light source, and adjusting the incident angle of the virtual light source and the viewing angle of the virtual camera. With each interaction, the physically based rendering engine recalculates and renders the appearance of the virtual sample. This interactive process aims to allow users to proactively and dynamically evaluate the appearance of the sample under different conditions before investing physical resources, especially for predicting potential problems such as metamerism or angle-dependent color variations.

[0076] S104. After the user is satisfied with the appearance of the virtual sample, they perform a confirmation operation through the interface, such as clicking the "Submit Physical Verification" button. At this time, the collaborative process management platform solidifies all parameters under the current virtual sample state and automatically generates a standardized digital work order. This digital work order is a structured data object, which includes at least: the initial formula confirmed by the user. Relevant process parameters For example, the staining temperature profile and processing time set in the virtual sample stage or the system default; the unique identifier of the task; and metadata such as the user and generation timestamp. After the digital work order is generated, the user interaction and workflow engine module updates the system status of the task to "pending production" and prepares to send the work order to the next step.

[0077] The S200 sends digital work orders to the networked automated sampling execution station; the networked automated sampling execution station automatically completes the preparation, staining, post-processing and transmission of physical samples according to the formula and process parameters in the digital work order.

[0078] After generating a digital work order, the S201 collaborative workflow management platform, through its user interaction and workflow engine module, sends the digital work order via the network to a designated or idle networked automated prototyping execution station. The transmission of digital work orders can be based on standard network protocols such as TCP / IP to ensure data integrity and reliability.

[0079] S202, the central control and network interface unit of the networked automated sample preparation station, parses the received digital work order to extract all the instruction information guiding the physical sample preparation. Specifically, this parsing process involves separating the initial formulation from the structured work order data. The defined dye components and their precise dosages, as well as the process parameters. Defined staining temperature curve A series of process instructions, such as processing time at each stage and heating / cooling rates.

[0080] S203, based on the analyzed initial formula The central control and network interface unit sends precise dispensing control commands to the high-precision automatic dispensing unit. In this implementation, the commands can be target volumes or weights for each dye or auxiliary agent. Based on these commands, the high-precision automatic dispensing unit activates specific micro-pumps or solenoid valves within it, drawing a specified dose of fluid from its respective storage tank and delivering it to a shared mixing container to prepare a dye solution perfectly consistent with the formulation.

[0081] S204, simultaneously with or after dye liquor preparation, the central control and network interface unit instructs the automatic sample transfer unit to operate. The unit's robotic arm or conveyor belt picks up standardized fabric samples from the preparation area and accurately loads them into the dyeing vat of the automated dyeing and post-treatment unit. Subsequently, based on the analyzed process parameters... The central control and network interface unit performs programmed control over the entire staining process of the automated staining and post-processing unit. This control specifically includes: based on a preset temperature profile... It precisely controls the heating, holding, and cooling processes of the dye liquor by controlling the heating or cooling system; it precisely maintains the processing time of each stage by controlling the timer; and it performs post-processing procedures such as cleaning and drainage by controlling the relevant valves and water pumps.

[0082] S205, after the staining and post-processing steps are completed, the automatic sample transfer unit activates again, removing the prepared physical sample from the staining tank. The sample undergoes rapid drying via the integrated drying unit and is then precisely placed on the sample stage of the multimodal appearance remote inspection terminal for further remote evaluation. After the physical sample preparation and placement are complete, the networked automatic sampling execution station sends a task completion status feedback signal to the collaborative workflow management platform via its network interface. Upon receiving this signal, the collaborative workflow management platform updates the status of the corresponding task in its database to "Pending Evaluation" and can automatically send an evaluation notification to the user who initiated the task, thereby triggering the subsequent step S300.

[0083] After the physical sample is transmitted to the multimodal appearance remote inspection terminal, the S300 receives the remote inspection command sent by the user through the collaborative process management platform; drives the multimodal appearance remote inspection terminal to adjust the light source and observation angle according to the inspection command, and collects the hyperspectral data stream of the sample; transmits the hyperspectral data stream to the user terminal for real-time rendering, so that the user can conduct collaborative evaluation and make decisions.

[0084] S301: After receiving a notification that the task status has been updated to "Pending Evaluation," the user logs into the collaborative workflow management platform via a client and enters the remote evaluation interface for the task. At this time, the platform establishes a real-time, bidirectional communication link between the user's client and the multimodal appearance remote inspection terminal. This link is used to transmit remote inspection commands issued by the user and to send back high-dimensional appearance data of the sample collected by the terminal.

[0085] S302, the user actively performs remote physical exploration of the physical sample placed on the sample stage of the multimodal appearance remote inspection terminal through the remote evaluation interface. In this implementation, the user's interactive operations, such as selecting the light source type or dragging the virtual camera icon on the interface, are converted into standardized remote exploration commands in real time by the client program. These commands can be data packets containing multiple parameters, and their format can be:

[0086] ;

[0087] in:

[0088] Select the type of light source;

[0089] and These are the incident zenith angle and azimuth angle of the light source, respectively;

[0090] and These are the zenith angle and azimuth angle of the observation (hyperspectral camera), respectively.

[0091] The collaborative process management platform sends the instruction to the multimodal appearance remote inspection terminal via the network.

[0092] The S303, a multimodal appearance remote inspection terminal's control system, immediately drives its hardware system to make a physical response upon receiving a remote inspection command. Specifically, the programmable multi-angle light source array responds according to the command... Switch the parameters to the corresponding light source, and according to... and The parameters illuminate an LED located at a specific spatial position; simultaneously, the hyperspectral imaging and robotic arm system operates according to the instructions... and The parameters drive the robotic arm or goniometer to adjust the hyperspectral camera to the specified spatial observation angle.

[0093] S304. After the positions and states of the light source and camera have stabilized, the hyperspectral camera acquires data from the sample surface and outputs a data cube, which is the spectral reflectance data stream of the sample surface under specific geometric and illumination conditions. The data acquisition and transmission unit packages the spectral data stream and the corresponding exploration command parameter Cmd from the acquisition time into a data frame and transmits it back to the user client of the collaborative workflow management platform in real time. This process continues as the user interacts continuously, forming a dynamic video data stream, but each frame contains complete physical measurement information.

[0094] S305, the client of the collaborative process management platform receives the spectral data stream. Then, the data is processed and presented synchronously. On one hand, the platform's rendering engine uses this spectral data, combined with the user's monitor's color characteristic file, to render a high-fidelity subjective visual appearance of the sample in real time for the user to observe. On the other hand, the platform simultaneously uses this spectral data to calculate and display the sample's objective colorimetric values ​​in this state, such as CIELAB values, as well as the color difference values ​​with digital standards or previous iteration results. This synchronous presentation of objective data and subjective perception provides users with a comprehensive evaluation basis.

[0095] In step S306, the user, or multiple invited collaborating users, can discuss and annotate on the remote evaluation interface based on the subjective appearance they see and the objective data they read. After a comprehensive evaluation, an authorized user makes the final decision, selecting "Approve" or "Reject," and submits the decision to the collaborative workflow management platform. The platform records the decision result and related collaborative annotation information in the digital archive of the task. If the decision is "Approve," the prototyping task process ends; if the decision is "Reject," the next step S400 is triggered.

[0096] S400: In response to the user's rejection decision, capture the hyperspectral data of the user-specified problem state with deviation; activate the intelligent correction algorithm module, calculate the correction amount of dye concentration through the formula reverse solving model according to the hyperspectral data of the problem state and the user's correction intention, so as to generate the corrected formula; based on the corrected formula, automatically generate a new digital work order and return to the execution step S200.

[0097] S401, in response to the rejection decision submitted by the user in step S300, the collaborative process management platform records the decision and related information, and activates the formula self-correction process. In this implementation, when the user submits a rejection decision, the system guides them to provide qualitative correction directions, such as selecting "needs more blue" or "needs to reduce gloss" through preset options. Simultaneously, the system captures and permanently records the "problem-awareness state" that the user specifies during the remote evaluation process as the one where the problem is most apparent. This state record is complete and specifically includes the remote probing command that generated the state. The spectral reflectance data stream acquired by the multimodal appearance remote inspection terminal under this instruction is denoted as the problem spectrum. .

[0098] S402, the intelligent correction algorithm module within the collaborative process management platform is activated, and its task is to correct the captured problem spectrum. Based on the user's qualitative correction direction, the desired target spectrum is generated. In its implementation, the system first analyzes the problem spectrum. Convert to a uniform color space, such as CIELAB, to obtain chromaticity coordinates. Then, based on the user's correction direction, the coordinate point is fine-tuned in CIELAB space. For example, if the user selects "Need more blue," then... The value is reduced by a preset step size to obtain the target chromaticity coordinates. Finally, the system uses a color space inverse transformation algorithm to convert the target chromaticity coordinates back to the spectral domain, generating the target spectrum. .

[0099] S403, in determining the target spectrum Then, the intelligent correction algorithm module calls its built-in recipe inverse solution model to calculate the solution to achieve the current problem spectrum. To the target spectrum The required dye concentration correction amount for the transformation. In one embodiment, the formulation inverse solution model is based on the spectral reflectance. With dye concentration vector A linear approximation of the nonlinear relationship between them. Spectral reflectance... In the current formula Performing a first-order Taylor expansion at the given point, we obtain:

[0100] ;

[0101] Apply this relationship at multiple sampling wavelength points Discretization allows us to construct the following matrix equation:

[0102] ;

[0103] in:

[0104] It is The spectral reflectance difference vector, whose elements are... This represents the difference between the target value and the current actual value at each wavelength. Spectral reflectance. Can be obtained from hyperspectral data Obtained through model transformation;

[0105] c is a The vector of dye concentration change to be solved is given by the following elements: , representing the The concentration of the dye needs to be adjusted;

[0106] It is The Jacobian matrix, whose elements , indicating at wavelength Spectral reflectance at position 1 to the first The sensitivity or partial derivative of a dye concentration.

[0107] The matrix can be analytically derived based on the Kubelka-Munk model established in step S102, or numerically calculated by applying a small perturbation to the concentration of each dye based on the current formulation and observing the change in spectral reflectance.

[0108] S404, due to the typical number of sampling wavelength points Much greater than the number of dye types The above matrix equations form an overdetermined system of equations. The intelligent correction algorithm module uses the least squares method to solve these equations to find the optimal solution that minimizes the sum of squared errors. c. The solution is given by the following equation:

[0109] ;

[0110] in:

[0111] Representation matrix transpose;

[0112] Representation matrix The inverse matrix.

[0113] S405, after calculating the dye concentration correction vector After step c, the system automatically generates a new, revised recipe. Subsequently, the collaborative workflow management platform automatically creates a new subtask digital work order associated with the original task, which contains the revised recipe. Including other process parameters inherited from the original task. After this new work order is generated, the system process automatically returns and executes step S200, that is, begins a new round of automated physical sample preparation. This process forms a closed-loop iteration from evaluation and decision-making to automatic correction and re-production.

[0114] The S500 stores the entire process data of each sampling cycle in a structured manner, forming a digital twin file uniquely bound to the sample; using the accumulated digital twin file, iterative optimization of the color science model and the formula reverse engineering model is performed in the background.

[0115] S501, after the completion of the entire prototyping process, regardless of whether it is successful on the first attempt or finally approved after multiple formulation self-correction iterations, the data and model management module of the collaborative process management platform will collect and structure all data generated throughout the entire lifecycle of the task to construct a complete digital twin archive uniquely bound to the physical sample. This digital twin archive may specifically include:

[0116] The target color parameter was defined when the task was initially created;

[0117] The recipe set for each iteration, including the initial recipe. and all revised formulas ;

[0118] The process parameters used in each automated physical sample preparation are as follows: ;

[0119] During each remote assessment, the logs of all user probing commands (Cmd) and the hyperspectral data cubes collected under each command are recorded. ;

[0120] The collaborative evaluation process includes annotations from all participating users and records of final decisions. All this data is linked by a unique task identifier and stored in the platform's database to ensure data integrity, consistency, and traceability.

[0121] S502, the data and model management module utilizes its continuously accumulated digital twin archives to execute a data-driven system self-optimization process in the background. This process is a continuous, iterative process based on historical experience learning, aimed at improving the performance and accuracy of the system's core algorithm model. In the implementation, this self-optimization process can specifically include two aspects:

[0122] Firstly, it optimizes the color science model in step S102. The system utilizes the "recipe-actual result" data pairs stored in the digital twin archive, that is, each recipe that has actually been executed. Its spectral reflectance obtained from the final physical measurement The system establishes a correspondence between these data and model predictions. By comparing these large amounts of actual data with the model predictions, the system can employ regression or nonlinear optimization algorithms from machine learning to determine the dye optical parameters in the Kubelka-Munk model. ) and substrate optical parameters ( The model is recalibrated and corrected to minimize the difference between the predicted spectrum and the actual measured spectrum. This allows the prediction results of the virtual sample module to get closer and closer to the physical reality as data accumulates.

[0123] Secondly, the system optimizes the reverse engineering model for the recipe in step S403. It utilizes successful amendment examples recorded in the digital twin archive, specifically from the problem spectrum... To the target spectrum The transformation was ultimately achieved through known dye concentration corrections. It is implemented in C. By analyzing a large number of such "problem-solution" data pairs, the system can empirically correct the calculation method of the Jacobian matrix J. In the specific implementation, the system can continuously update and refine each element in the matrix based on statistical regression methods. The value of this value is adjusted to more accurately reflect the sensitivity of the spectrum to dye concentration in the real physical world. This allows the intelligent correction algorithm module to calculate the compensation formula more quickly and accurately in subsequent tasks, thereby improving the success rate of correction.

Claims

1. A network-connected collaborative sampling system for small-scale dye samples, characterized in that, include: The collaborative process management platform is used to generate digital chemical orders containing formulas and send remote inspection commands to multimodal appearance remote inspection terminals. The networked automated sampling execution station is connected to the collaborative process management platform to receive the digital work order and automatically complete the preparation of physical samples according to the digital work order. The multimodal appearance remote detection terminal is communicatively connected to the collaborative process management platform. It is used to receive the remote exploration command, collect the hyperspectral data stream of the physical sample according to the remote exploration command, and transmit the hyperspectral data stream to the collaborative process management platform for users to conduct collaborative evaluation.

2. The network-connected collaborative sampling system for dye samples according to claim 1, characterized in that, The collaborative process management platform includes a virtual sample and physical rendering module, which is used to generate an interactive three-dimensional virtual sample based on the initial formula before generating the digital work order, so that the user can interactively test the initial formula in a virtual sample, and after the user confirms, the initial formula is fixed in the digital work order.

3. The network-connected collaborative sampling system for dye samples according to claim 1, characterized in that, The collaborative process management platform includes an intelligent correction algorithm module, which, in response to the user's rejection decision, captures the hyperspectral data specified by the user under the problem state, and calculates the correction amount of dye concentration based on the hyperspectral data under the problem state through a formula reverse solving model, so as to generate a corrected formula and generate a new digital work order based on the corrected formula.

4. The network-connected collaborative sampling system for dye samples according to claim 3, characterized in that, The formulation inverse solution model calculates the correction amount of the dye concentration by constructing a Jacobian matrix representing the sensitivity of spectral reflectance to dye concentration, and solving the matrix equation based on the difference between the hyperspectral data and the target spectrum under the problem state to obtain the correction amount of the dye concentration.

5. The network-connected collaborative sampling system for dye samples according to claim 1, characterized in that, The multimodal appearance remote inspection terminal includes: A programmable multi-angle light source array, used to change the type of light source and the incident angle illuminating the physical sample according to the remote exploration command; A hyperspectral imaging and robotic arm system is used to adjust the observation angle of the physical sample according to the remote exploration command and to acquire the hyperspectral data stream.

6. A network-connected collaborative sampling system for dye samples according to claim 5, characterized in that, The collaborative process management platform is configured to convert the user's interface operations into remote exploration commands in real time, so as to drive the programmable multi-angle light source array and the hyperspectral imaging and robotic arm system to respond synchronously.

7. The network-connected collaborative sampling system for dye samples according to claim 1, characterized in that, The networked automated proofing station includes: A high-precision automatic solution preparation unit is used to prepare dye solution according to the formula in the digital work order; An automated dyeing and post-processing unit is used to dye fabric samples according to the process parameters in the digital work order; An automatic sample transfer unit is used to transfer fabric samples between a high-precision automatic solution preparation unit and an automated dyeing and post-treatment unit.

8. A network-connected collaborative sampling system for dye samples according to claim 3, characterized in that, The collaborative process management platform also includes a data and model management module, which is used to structure and store the full-process data of each sampling cycle to form a digital twin archive, and to use the accumulated digital twin archive to perform background iterative optimization of the formula reverse engineering model.

9. A network-connected collaborative sampling system for dye samples according to claim 1, characterized in that, The collaborative process management platform is configured to synchronously process the received hyperspectral data stream during the collaborative evaluation to simultaneously present the subjective visual appearance of the physical sample and its corresponding objective colorimetric value on the user interface.

10. A network-connected collaborative sampling method for dye samples, applied to the network-connected collaborative sampling system for dye samples as described in any one of claims 1-9, characterized in that, Includes the following steps: An interactive 3D virtual sample is generated based on the initial formula, and a digital chemical invoice containing the initial formula is generated after user confirmation. The digital work order is sent to the networked automated sampling execution station to automatically complete the preparation of physical samples; The multimodal appearance remote inspection terminal adjusts the light source and observation angle according to the remote exploration command issued by the user, and collects the hyperspectral data stream of the physical sample for the user to conduct collaborative evaluation; In response to the user's rejection decision, the system captures hyperspectral data in the problematic state, calculates the correction amount of the dye concentration using a formulation inverse solving model to generate a corrected formulation, and returns to execute the preparation steps of the physical sample based on the corrected formulation.

Images