Automatic feeding and proportioning control system for blue dyeing process

Abstract

According to the automatic feeding and proportioning control system for the blue dyeing process, full-automatic feeding and accurate proportioning in the blue dyeing process are achieved through the indigo concentration real-time monitoring module, the intelligent proportioning and accurate feeding module, the reduction potential closed-loop adjustment module, the oxidation degree dynamic control module and the like. The system adopts methods such as near infrared spectrum analysis, a micro-fluidic technology and gas-liquid mixing micro-jet flow, so that accurate addition and uniform mixing of raw materials are ensured. Meanwhile, the oxidation process and the temperature distribution are accurately controlled by utilizing a multi-stage programmable gas diffusion and partitioned electromagnetic induction heating technology, and the dyeing quality and the production efficiency are improved.

Description

Technical Field

[0001] The invention belongs to the technical field of textile dyeing, and in particular relates to an automatic feeding and proportioning control system for a blue dyeing process. Background Art

[0002] Although the traditional blue dyeing process has a long history, it faces many technical challenges in modern applications. Among them, the instability of dye ratio is a significant problem, which not only affects the consistency of dyeing quality, but also increases production costs. In addition, the reduction and oxidation processes in the blue dyeing process are difficult to control. These two steps have a direct impact on the color fastness of the final product, and unstable control often leads to uneven color and low color fastness of the finished product, which cannot meet the high standards of the market.

[0003] The current indigo dyeing technology on the market still needs to be improved in terms of automation. Existing technologies mostly rely on manual operations, which makes it difficult to achieve standardization and automation of large-scale production, which limits the production capacity and efficiency of the indigo dyeing process. At the same time, the lack of intelligent and precise control of the entire dyeing process leads to serious waste of resources in the production process and a greater impact on the environment.

[0004] Therefore, there is an urgent need for a new technology that can solve the above problems. This technology should be able to achieve precise control of the dye ratio, optimize the reduction and oxidation processes, and improve the automation and intelligence level of the process to meet the needs of modern production. Summary of the invention

[0005] The present invention provides an automatic feeding and proportioning control system for the blue dyeing process, aiming to solve the technical problems existing in the traditional blue dyeing process, such as unstable dye proportions, difficult to control reduction and oxidation processes, color fastness problems, and low automation. The system integrates a number of advanced detection and control technologies to achieve intelligent and precise control of the entire blue dyeing process.

[0006] The system comprises the following components:

[0007] 1. Indigo concentration real-time monitoring module: This module uses near-infrared spectroscopy analysis technology to continuously monitor the indigo concentration in the dye solution. It contains a self-cleaning eddy current sampling device, which can effectively prevent the influence of dye deposition on the measurement accuracy. The module also integrates temperature and pH sensors, and automatically corrects the influence of these factors on the measurement results through machine learning algorithms. In addition, the module is also equipped with a data preprocessing unit, which can filter out signal noise and improve the stability and reliability of the measurement.

[0008] 2. Intelligent Ratio Precision Feeding Control Unit: This unit is responsible for accurately controlling the addition of various raw materials according to the recipe requirements and real-time monitoring data. It is equipped with a high-precision peristaltic pump and solenoid valve array to control the feeding of multiple raw materials at the same time. In order to ensure the accurate addition of small doses of additives, the unit also integrates microfluidics technology. In addition, the unit uses an adaptive fuzzy control algorithm that can dynamically adjust the feeding rate and ratio according to real-time parameters such as indigo concentration and pH value. The unit also has a self-learning function that can optimize the feeding strategy based on historical data.

[0009] 3. Reduction potential closed-loop regulation module: The main function of this module is to accurately control the reduction potential of the dye solution to ensure the stability of the indigo reduction process. It uses platinum-palladium alloy electrodes to improve the accuracy and stability of reduction potential measurement. The module integrates a gas-liquid mixing micro-jet device to greatly improve the dispersion efficiency of the reducing agent. The module uses a model predictive control (MPC) algorithm to achieve feedforward-feedback composite control of the reduction potential. In addition, the module also has a self-diagnosis function that can promptly detect and report abnormal conditions such as electrode failure.

[0010] 4. Dynamic control unit for oxidation degree: This unit is responsible for precisely controlling the oxidation process of the fabric to ensure dyeing uniformity and color stability. It is designed with a multi-level programmable gas diffusion device to achieve segmented precise control of the oxidation process. The unit integrates a high-speed camera and image processing system to analyze fabric color changes in real time. Using a deep reinforcement learning algorithm, the unit can dynamically optimize oxidation parameters. In addition, the unit is also equipped with a gas flow precision regulating valve to accurately control the gas supply during the oxidation process as needed.

[0011] 5. Temperature gradient precision adjustment module: This module realizes precise control and uniform distribution of temperature in the dye vat. It adopts zoned electromagnetic induction heating technology to independently control the temperature of different areas of the dye vat. The module integrates a thermal imager array to build a three-dimensional temperature field model inside the dye vat. By applying the computational fluid dynamics (CFD) algorithm, the module can optimize the heat flow distribution and effectively eliminate hot spots and cold areas. In addition, the module also has an intelligent energy-saving function, which can automatically adjust the heating strategy according to process requirements to maximize energy efficiency.

[0012] 6. Dye solution pH automatic balancing unit: This unit is responsible for real-time adjustment and stabilization of the pH value of the dye solution to optimize the dyeing effect. It uses ion-selective electrodes to improve the anti-interference ability of pH measurement. The unit is designed with a microinjection system to achieve accurate addition of acid and alkali. Using a neural network PID control algorithm, the unit significantly improves the response speed and stability of pH adjustment. In addition, the unit also has a prediction function that can predict changes in pH based on historical data and current trends, achieve advance adjustment, and further improve control accuracy.

[0013] 7. Fabric dyeing uniformity monitoring module: This module monitors the dyeing status of the fabric in the dye solution in real time to ensure dyeing uniformity. It uses ultrasonic array technology to scan the position and posture of the fabric in the dye solution in real time. The module integrates electromagnetic induction sensors to monitor the degree of dye absorption by the fabric. By applying machine vision algorithms, the module can analyze the color distribution on the fabric surface and detect and correct unevenness in time. In addition, the module is also equipped with an automatic guidance system that can adjust the movement path of the fabric in the dye vat according to the monitoring results to further improve dyeing uniformity.

[0014] Furthermore, the machine learning algorithm in the indigo concentration real-time monitoring module is a support vector regression (SVR) algorithm, which establishes a multidimensional feature space and comprehensively considers the influence of factors such as temperature and pH value on the indigo concentration measurement to achieve automatic correction of the measurement results.

[0015] Furthermore, the adaptive fuzzy control algorithm adopted by the intelligent proportioning and precise feeding control unit is based on the Mamdani inference model. By setting parameters such as indigo concentration, reduction potential and pH value as input variables and feeding rate and ratio as output variables, a fuzzy rule base is constructed to achieve dynamic optimization of the feeding process.

[0016] Furthermore, the model predictive control (MPC) algorithm in the reduction potential closed-loop regulation module adopts a dynamic matrix control (DMC) method, which predicts the reduction potential changes in the future period by establishing a dynamic model between the reduction potential and the amount of reducing agent added, and optimizes control actions to achieve precise regulation of the reduction potential.

[0017] Furthermore, the deep reinforcement learning algorithm in the oxidation degree dynamic control unit adopts a double-delay deep Q network (Double DQN) structure, takes parameters such as the fabric color change rate and uniformity as state inputs, and the oxidant addition rate, stirring speed, etc. as action outputs, and continuously optimizes the strategy through interaction with the environment to achieve adaptive control of the oxidation process.

[0018] Furthermore, the computational fluid dynamics (CFD) algorithm in the temperature gradient precise adjustment module is based on the finite volume method, adopts the RNG k-ε turbulence model, and combines the multiphase flow model to simulate the flow and heat transfer process of the dye solution, optimize the control strategy of the heating element, and achieve the uniformity of the temperature field in the dye vat.

[0019] Furthermore, the neural network PID control algorithm in the dye solution pH automatic balancing unit adopts BP neural network, optimizes PID parameters through online learning, and improves the response speed and stability of pH value regulation. The algorithm takes pH value deviation and its change rate as input, outputs optimized PID parameters, and realizes adaptive regulation of pH value control.

[0020] Furthermore, the machine vision algorithm in the fabric dyeing uniformity monitoring module adopts a convolutional neural network (CNN) structure, which identifies the unevenness of color distribution through real-time analysis of the fabric surface image, and feeds back the analysis results to the intelligent proportioning and precise feeding control unit to form a closed-loop control to ensure the uniformity of the dyeing process.

[0021] Through the above technical scheme, the present invention realizes intelligent control of the whole process of blue dyeing process, significantly improves dyeing uniformity, indigo utilization, reduction stability, oxidation accuracy, temperature uniformity and pH stability, and at the same time greatly improves production efficiency and resource utilization, providing strong support for the modernization and sustainable development of blue dyeing process.

[0022] The present invention is beneficial in that:

[0023] 1. Improve the control accuracy of indigo concentration: The real-time monitoring module of indigo concentration adopts near-infrared spectroscopy analysis technology and self-cleaning eddy current sampling device to improve the measurement accuracy of indigo concentration. This helps to improve the utilization rate of dyes and reduce the residual dyes in wastewater.

[0024] 2. Enhanced ratio stability: The intelligent ratio precision feeding control unit adopts high-precision peristaltic pump and microfluidics technology to improve the accuracy of adding small doses of additives. This is conducive to improving the color consistency between batches.

[0025] 3. Optimize reduction process control: The reduction potential closed-loop regulation module improves the control stability of the reduction potential through the platinum-palladium alloy electrode and the gas-liquid mixing microfluidic device. This can improve the utilization efficiency of the reducing agent.

[0026] 4. Improve the accuracy of the oxidation process: The multi-level programmable gas diffusion device and deep reinforcement learning algorithm of the oxidation degree dynamic control unit improve the control accuracy of the oxidation process. This helps to improve the color fastness of the finished fabric.

[0027] 5. Improve the uniformity of temperature field: The temperature gradient precision adjustment module improves the uniformity of temperature distribution in the dyeing tank through the partitioned electromagnetic induction heating technology and thermal imager array. This is conducive to improving dyeing uniformity.

[0028] 6. Enhanced pH stability: The dye solution pH automatic balance unit uses ion selective electrodes and a microinjection system to improve the pH control stability. This can improve the reproducibility of dyeing.

[0029] 7. Improve dyeing uniformity: The fabric dyeing uniformity monitoring module improves the monitoring accuracy of the fabric dyeing status through ultrasonic array technology and electromagnetic induction sensors. This helps to improve the consistency of product quality.

[0030] 8. Improve the level of automation: The integrated automation design of the system reduces manual intervention, can reduce human errors and improve process stability.

[0031] 9. Enhanced process adaptability: The system adopts machine learning, fuzzy control, deep reinforcement learning, etc., which enables it to better adapt to different types of fibers and dyes, helping to improve process flexibility. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 A schematic diagram of the system architecture of the present invention is shown;

[0033] Figure 2 A flow chart of the steps of the present invention is shown. DETAILED DESCRIPTION

[0034] Combination Figure 1 , the system includes the following components:

[0035] 1. Indigo concentration real-time monitoring module: This module uses near-infrared spectroscopy analysis technology to continuously monitor the indigo concentration in the dye solution. It contains a self-cleaning eddy current sampling device, which can effectively prevent the influence of dye deposition on the measurement accuracy. The module also integrates temperature and pH sensors, and automatically corrects the influence of these factors on the measurement results through machine learning algorithms. In addition, the module also has a data preprocessing function, which can filter out signal noise and improve the stability and reliability of the measurement.

[0036] 2. Intelligent Ratio Precision Feeding Control Unit: This unit is responsible for accurately controlling the addition of various raw materials according to the recipe requirements and real-time monitoring data. It is equipped with a high-precision peristaltic pump and solenoid valve array to control the feeding of multiple raw materials at the same time. In order to ensure the accurate addition of small doses of additives, the unit also integrates microfluidics technology. In addition, the unit uses an adaptive fuzzy control algorithm that can dynamically adjust the feeding rate and ratio according to real-time parameters such as indigo concentration and pH value. The unit also has a self-learning function that can optimize the feeding strategy based on historical data.

[0037] 3. Reduction potential closed-loop regulation module: The main function of this module is to accurately control the reduction potential of the dye solution to ensure the stability of the indigo reduction process. It uses platinum-palladium alloy electrodes to improve the accuracy and stability of reduction potential measurement. The module integrates a gas-liquid mixing micro-jet device to greatly improve the dispersion efficiency of the reducing agent. The module uses a model predictive control (MPC) algorithm to achieve feedforward-feedback composite control of the reduction potential. In addition, the module also has a self-diagnosis function that can promptly detect and report abnormal conditions such as electrode failure.

[0038] 4. Dynamic control unit for oxidation degree: This unit is responsible for precisely controlling the oxidation process of the fabric to ensure dyeing uniformity and color stability. It is designed with a multi-level programmable gas diffusion device to achieve segmented precise control of the oxidation process. The unit integrates a high-speed camera and image processing system to analyze fabric color changes in real time. Using a deep reinforcement learning algorithm, the unit can dynamically optimize oxidation parameters. In addition, the unit is also equipped with a gas flow precision regulating valve to accurately control the gas supply during the oxidation process as needed.

[0039] 5. Temperature gradient precision adjustment module: This module realizes precise control and uniform distribution of temperature in the dye vat. It adopts zoned electromagnetic induction heating technology to independently control the temperature of different areas of the dye vat. The module integrates a thermal imager array to build a three-dimensional temperature field model inside the dye vat. By applying the computational fluid dynamics (CFD) algorithm, the module can optimize the heat flow distribution and effectively eliminate hot spots and cold areas. In addition, the module also has an intelligent energy-saving function, which can automatically adjust the heating strategy according to process requirements to maximize energy efficiency.

[0040] 6. Dye solution pH automatic balancing unit: This unit is responsible for real-time adjustment and stabilization of the pH value of the dye solution to optimize the dyeing effect. It uses ion-selective electrodes to improve the anti-interference ability of pH measurement. The unit is designed with a microinjection system to achieve accurate addition of acid and alkali. Using a neural network PID control algorithm, the unit significantly improves the response speed and stability of pH adjustment. In addition, the unit also has a prediction function that can predict changes in pH based on historical data and current trends, achieve advance adjustment, and further improve control accuracy.

[0041] 7. Fabric dyeing uniformity monitoring module: This module monitors the dyeing status of the fabric in the dye solution in real time to ensure dyeing uniformity. It uses ultrasonic array technology to scan the position and posture of the fabric in the dye solution in real time. The module integrates electromagnetic induction sensors to monitor the degree of dye absorption by the fabric. By applying machine vision algorithms, the module can analyze the color distribution on the fabric surface and detect and correct unevenness in time. In addition, the module is also equipped with an automatic guidance system that can adjust the movement path of the fabric in the dye vat according to the monitoring results to further improve dyeing uniformity.

[0042] Furthermore, the machine learning algorithm in the indigo concentration real-time monitoring module is a support vector regression (SVR) algorithm, which establishes a multidimensional feature space and comprehensively considers the influence of factors such as temperature and pH value on the indigo concentration measurement to achieve automatic correction of the measurement results.

[0043] Specifically, the application of the support vector regression is as follows:

[0044] A). Feature selection and extraction:

[0045] In near-infrared spectroscopy, the absorption peak at a specific wavelength is correlated with the indigo concentration. By analyzing historical experimental data, the characteristic wavelengths (such as 660nm, 720nm, etc.) that are highly correlated with the indigo concentration are determined.

[0046] Temperature and pH were selected as environmental characteristic parameters because they have the greatest impact on the spectrum. These parameters were collected in real time to provide a dynamic correction for the indigo concentration measurement.

[0047] B). Model training:

[0048] Data collection: Spectral data sets at different temperatures and pH values ​​were collected through experiments. These data include near-infrared spectral curves of multiple groups of samples, corresponding temperatures, pH values, and actual measured indigo concentrations.

[0049] Data annotation: Annotate the data set. The annotation items include spectral feature vector (absorption peak, wavelength, etc.), temperature, pH value and corresponding indigo concentration.

[0050] Training process: The model is trained using the support vector regression (SVR) algorithm, with spectral features and environmental features (temperature and pH) as input and indigo concentration as output. SVR uses kernel functions (such as RBF kernels) to construct nonlinear relationships in high-dimensional space and find the optimal mapping relationship between spectral features and indigo concentration.

[0051] C). Calibration and prediction:

[0052] Real-time data input: In actual operation, the system collects spectral data, temperature and pH value in real time.

[0053] Concentration prediction: These real-time data are input into the trained SVR model, and the model outputs the predicted indigo concentration value. Through the multi-dimensional feature space mapping capability of SVR, the spectral deviation caused by temperature and pH value can be corrected to achieve accurate indigo concentration prediction.

[0054] Specifically, the data preprocessing method involved in the real-time monitoring module for indigo concentration includes but is not limited to:

[0055] A). Signal smoothing:

[0056] Moving average filtering: Smooth the spectral signal by taking the average value within a certain range, thereby reducing high-frequency noise. Set a fixed window (such as 5 points or 7 points) and calculate the moving average of the spectral data point by point to eliminate rapid fluctuations.

[0057] Gaussian smoothing filter: Use the Gaussian function as the weight to smooth the signal. Compared with moving average, Gaussian smoothing can more naturally reduce the noise far away from the center point according to the weight of the Gaussian distribution.

[0058] B). Baseline correction:

[0059] Polynomial Fitting: Identify and remove baseline drift in spectral signals by fitting a polynomial curve. Select the appropriate polynomial order (such as second or third order), fit the baseline portion of the spectrum and subtract it.

[0060] Wavelet transform: Through wavelet transform, the spectral signal is decomposed into wavelet coefficients of different scales. The baseline signal usually appears in the low-frequency part. By removing the low-frequency coefficients, the baseline drift can be effectively removed.

[0061] C). De-noising:

[0062] Wavelet denoising: Use wavelet transform to separate the noise components in the spectral signal. By setting a suitable threshold, high-frequency noise can be removed and valid signals can be retained.

[0063] Principal Component Analysis (PCA): Reduce the dimensionality of spectral data, identify the main change patterns, and filter out noise components that are not related to indigo concentration. Through PCA, the principal components that reflect the main information can be extracted and the noise can be eliminated.

[0064] D). Standard Normal Variable Transformation (SNV) or Multivariate Scatter Correction (MSC):

[0065] These methods are used to correct for scattering effects in spectral data. The standard normal variate transformation normalizes the data by subtracting each spectral signal from its mean and dividing by the standard deviation to reduce the interference of scattering on concentration measurements.

[0066] Multivariate scatter correction corrects the spectral data by modeling the relationship between the scattering effect and the actual concentration in the sample, making the spectral characteristics of different samples more comparable.

[0067] Furthermore, the adaptive fuzzy control algorithm adopted by the intelligent proportioning and precise feeding control unit is based on the Mamdani inference model. By setting parameters such as indigo concentration, reduction potential and pH value as input variables and feeding rate and ratio as output variables, a fuzzy rule base is constructed to achieve dynamic optimization of the feeding process.

[0068] Specifically, the adaptive fuzzy control algorithm is applied as follows:

[0069] A). Initial stage: Establishing fuzzy rule base

[0070] In the initial stage of the system, the fuzzy rule base is mainly built based on expert experience and process requirements. These rules are based on the Mamdani inference model and include fuzzification of input variables (such as indigo concentration, pH value, reduction potential), as well as control decisions for output variables (such as feed rate and ratio). The initial fuzzy rule base may be as follows:

[0071] Rule 1: If the indigo concentration is "low" and the pH is "alkaline", then the dye feed rate is "increase".

[0072] Rule 2: If the reduction potential is "strong" and the pH is "neutral", then the reductant feed rate remains "unchanged".

[0073] Rule 3: If the indigo concentration is "high" and the pH is "acidic", then the dye feed rate is "reduced".

[0074] These rules are constructed through expert knowledge and preliminary experimental data, providing a basic control logic for the system.

[0075] B). Data collection and real-time monitoring

[0076] During the actual production process, the system continuously collects real-time data, including:

[0077] Process parameters such as indigo concentration, pH value, reduction potential, etc.

[0078] Actual addition of dyes and auxiliaries (feeding rates and ratios).

[0079] Quality indicators of the final dyeing effect, such as color uniformity and color fastness.

[0080] These data are recorded in real time and stored in the system's database, providing a basis for subsequent analysis.

[0081] C). Fuzzy reasoning and control

[0082] During the production process, the system uses the Mamdani reasoning model to perform fuzzy reasoning based on real-time monitoring data input:

[0083] Input fuzzification: Convert the real-time monitored indigo concentration, pH value and other data into fuzzy sets. For example, fuzzify the current indigo concentration value into the "medium" level.

[0084] Rule matching and reasoning: The system matches the appropriate rules in the fuzzy rule base according to the current input fuzzy set and makes inferences. For example, if the current indigo concentration is "low" and the pH value is "alkaline", the system will match Rule 1 and decide to increase the dye feed rate.

[0085] Output defuzzification: Through a defuzzification method (such as the centroid method), the exact value of the output is calculated, such as increasing the dye feed rate by 10%.

[0086] D) Application of self-learning mechanism

[0087] The key to the self-learning function is that the system can continuously optimize and adjust the fuzzy rules and control strategies based on production results and historical data. The self-learning process is as follows:

[0088] Real-time feedback and effect evaluation: After executing the fuzzy control strategy, the system continuously monitors the dyeing effect. By testing the quality indicators of the final product (such as color difference and color fastness), the system evaluates the effectiveness of the current control strategy.

[0089] Data storage and analysis: All production data (input parameters, control strategies, dyeing results) will be stored. By analyzing these historical data, the system can identify the relationship between parameter changes and dyeing results. For example, through data analysis, the system may find that when the indigo concentration is "low" and the pH value is "neutral", increasing the feed rate of the reducing agent can significantly improve the color uniformity.

[0090] Update fuzzy rule base: Based on the analysis results of historical data, the self-learning mechanism automatically updates the fuzzy rule base. New rules may be added, and the priority or conditions of old rules may be adjusted. For example, the system may introduce new rules:

[0091] New rule: If indigo concentration is "low" and pH is "neutral", then increase the reducing agent feed rate.

[0092] Optimize control parameters: Not only are the fuzzy rules updated, but the self-learning mechanism can also adjust the defuzzified output range and control parameters to make the system more sensitive or stable in response to input changes.

[0093] E). Continuous optimization and iteration

[0094] Self-learning is an ongoing, iterative process. The system continuously uses new data and feedback information to optimize the control strategy:

[0095] Feedback loop: Data analysis after each production run may bring new insights and improvement opportunities. These improvements are reflected in the control strategy for the next production run.

[0096] Long-term optimization: As time goes by, the system accumulates a large amount of operation data and effect feedback, gradually forming a more accurate and comprehensive fuzzy rule base. The control effect of the system is continuously optimized, and it can better cope with the uncertainty and changes in production.

[0097] Furthermore, the model predictive control (MPC) algorithm in the reduction potential closed-loop regulation module adopts a dynamic matrix control (DMC) method, which predicts the reduction potential changes in the future period of time by establishing a dynamic model between the reduction potential and the amount of reducing agent added, and optimizes the control action to achieve precise regulation of the reduction potential.

[0098] Specifically, the dynamic matrix control (DMC) method is applied as follows:

[0099] A). Establish dynamic response model

[0100] The DMC algorithm relies on the impulse response model of the system, which is used to describe the dynamic response of the reduction potential to the amount of reducing agent added. Assume that the response of the reduction potential V to the reducing agent amount Q can be expressed by the impulse response coefficients h1, h2...h N Description, then:

[0101]

[0102] in:

[0103] V t+k To predict the reduction potential for the next k steps;

[0104] V base is the baseline reduction potential, which represents the natural state of the system without any control action;

[0105] ΔQ t+k-i is the incremental amount of reducing agent added at the ki-th moment;

[0106] h i is the impulse response coefficient, which is obtained by fitting the experimental data.

[0107] B). Feedforward control

[0108] External perturbation modeling: Considering the effect of changes in dye flow rate F and temperature T on the reduction potential, these perturbations are modeled as external inputs. The response coefficients f of these perturbations are obtained experimentally. i and t i :

[0109]

[0110] in:

[0111] ΔF t+k-i is the change in dye flow rate;

[0112] ΔT t+k-i is the change in dye solution temperature.

[0113] Feedforward control quantity calculation: According to the disturbance response model, the impact of the expected disturbance on the reduction potential is calculated and pre-compensated. For example, it is expected that an increase in flow rate may lead to a decrease in the reduction potential, and the feedforward control quantity ΔQ ff The calculation method is:

[0114]

[0115] C) Feedback control

[0116] Error calculation: real-time measurement of reduction potential V actual and the target potential V set The deviation e t =V set -V actual ;

[0117] Feedback correction: DMC solves the control input for the next N steps through optimization to minimize the output error. The optimization goal is:

[0118]

[0119] in:

[0120] V t+j The predicted reduction potential for the future moment is calculated by the model and used to predict the future state of the system and optimize the control action based on this prediction;

[0121] ΔQ t+j-1 is the change in the amount of reducing agent added at the future j-1 moment, which controls the actual amount of addition applied in the system and is used to smooth the control action to prevent overreaction or oscillation; λ is the penalty weight of the control input change, which is used to balance the system response speed and control stability.

[0122] D). Calculation and execution of comprehensive control quantities

[0123] Total control amount synthesis: Combine the feedforward and feedback control amounts to get the actual amount of reducing agent added:

[0124] ΔQ t =ΔQ ff +ΔQ fb

[0125] Among them, ΔQ fb is the feedback correction amount.

[0126] Control action execution: Real-time adjustment of the amount of reducing agent added Q through the control system t , ensuring that the reduction potential is maintained within the target range. The gas-liquid mixing microfluidic device accurately adds the reducing agent based on the calculated results.

[0127] Specifically, the self-diagnosis method of the reduction potential closed-loop regulation module is as follows:

[0128] A). Real-time signal monitoring:

[0129] Signal characteristics: The system continuously monitors the potential V output by the electrode t , Noise N t , response time T res .

[0130] Baseline parameter setting: Set the signal characteristic baseline under normal working conditions, for example, the normal potential range V n is [-0.65v,-0.60v], the noise threshold N th =5mV.

[0131] B). Anomaly detection and identification:

[0132] Signal anomaly detection: The system uses standard deviation and mean calculation to detect potential fluctuations. If the standard deviation of the potential signal exceeds the normal threshold or the mean deviates from the baseline, it is judged as abnormal.

[0133] Response time anomaly detection: If the system response time to the control signal is T res Exceeding the set threshold T th = 0.2 seconds, it is considered as an abnormal response.

[0134] C). Fault diagnosis and treatment:

[0135] Abnormal pattern classification: Based on abnormal characteristics (such as high noise, signal distortion), the system uses preset classification rules to determine the fault type, such as electrode contamination, poor electrode contact, etc.

[0136] Alarm and automatic processing: After detecting an abnormality, the system will immediately alarm and prompt the operator to check. For minor abnormalities, the system automatically adjusts the signal processing algorithm; for serious faults, the system switches to the backup electrode.

[0137] D). Redundant design and automatic correction:

[0138] Redundant electrode design: The system uses two sets of electrodes that work independently of each other and cross-validate the results. When an abnormal signal is detected, the system switches to the backup electrodes to ensure measurement continuity.

[0139] Automatic correction: For correctable anomalies (such as high signal noise), the system automatically corrects the signal through a filtering algorithm to reduce the impact of the anomaly on the control system.

[0140] Furthermore, the deep reinforcement learning algorithm in the oxidation degree dynamic control unit adopts a double-delay deep Q network (Double DQN) structure, takes parameters such as the fabric color change rate and uniformity as state inputs, and the oxidant addition rate, stirring speed, etc. as action outputs, and continuously optimizes the strategy through interaction with the environment to achieve adaptive control of the oxidation process.

[0141] Specifically, the double-delay deep Q network is applied as follows:

[0142] A). State space definition:

[0143] State parameters: define the state space S of the system, including:

[0144] Fabric color change rate C rate :Real-time monitoring of the rate of fabric color change, reflecting the speed of oxidation reaction;

[0145] Color uniformity U: Analyze the color distribution on the fabric surface through the image processing system to measure the dyeing uniformity;

[0146] Oxidant concentration conc : Real-time monitoring of oxidant concentration to ensure it is within the optimal range;

[0147] Oxygen flow rate G flow : Current oxygen flow rate;

[0148] Temperature T: temperature of the dye solution.

[0149] State space representation: S t =[C rate ,U,O conc ,G flow ,T], the system is based on the current state S t Select the optimal control action.

[0150] B). Action space definition:

[0151] Action parameters: define the action space A of the system, including:

[0152] Oxidant addition rate A add : Control the rate of oxidant addition;

[0153] Stirring speed A stir : Control the speed of the stirring equipment to ensure uniform mixing of the dye solution;

[0154] Gas flow regulation A gas : Adjust the oxygen flow into the dye solution.

[0155] Expression of action space: A t =[A add,A stir ,A gas ], according to the current state S t , the system selects one or more actions A t to execute.

[0156] C) Reward function design:

[0157] Reward function R(S t ,A t ): Design a reward function to measure the effect of each action, with the goal of maximizing the long-term cumulative reward.

[0158] Positive reward: If the oxidation process stabilizes the color change rate within the target range and the color uniformity is improved, the system will give a positive reward.

[0159] Negative reward: If the oxidation process causes the color to change too quickly or too slowly, resulting in reduced uniformity, the system will give a negative reward.

[0160] R(S t ,A t )=α(U target -U t )-β|C rate_target -C rate |-γ(O conc -O target )

[0161] in:

[0162] α, β, and γ are weight parameters that control the importance of different objectives;

[0163] U target is the target color uniformity;

[0164] U t is the color uniformity at time t;

[0165] C rate_target is the color change rate of the target fabric;

[0166] C rate is the fabric color change rate;

[0167] O conc is the oxidant concentration;

[0168] O target is the target oxidant concentration.

[0169] D) Deep Q-Network and Dual Delay Mechanism

[0170] Input Layer:

[0171] Contains all parameters in the state space: C rate,U,O conc , G flow , T; used to obtain the current system state as the input of the neural network.

[0172] Hidden layer: Multiple hidden layers, the number of layers and the number of neurons in each layer are designed according to the specific situation (such as 3 layers, 128 neurons in each layer); increase the expressive power of the network and handle complex nonlinear relationships.

[0173] Output layer: Corresponding to each possible action in the action space, output the Q value of each action.

[0174] The Q-value of each action represents the expected long-term reward of choosing that action in the current state.

[0175] Target network and evaluation network:

[0176] Target network Q': used to calculate the target Q value, updated regularly to avoid overfitting.

[0177] Evaluation network Q: used to select the optimal action and directly affects the current strategy.

[0178] Every certain number of steps, the weights of the evaluation network are copied to the target network.

[0179] Target Q value calculation:

[0180] Q target =R(S t ,A t )+γmaxQ'(S t+1 ,A t+1 )

[0181] Among them, γ is a discount factor, which usually ranges from 0.9 to 0.99 and is used to weigh immediate rewards and future rewards.

[0182] E) Experience replay for training and strategy optimization: The system uses the experience replay mechanism to randomly sample from stored past experience for training, avoiding correlation between samples and improving training efficiency and stability.

[0183] Policy update: At each step, the system selects actions based on the current policy and updates the policy by observing the execution results. Policy update is achieved by minimizing the following loss function:

[0184] L(θ)=E[(Q target -Q(S t ,A t ;θ)) 2 ]

[0185] in:

[0186] L(θ) is the loss function, which reflects the error between the Q value predicted by the model and the target Q value; E is the expected value operator, which represents the overall average effect of the loss function;

[0187] Q target is the target Q value, which is calculated through the target network to ensure the stability of learning; Q(S t ,A t ; θ) is the Q value predicted by the current network (evaluation network), according to the current state S t and action A t Calculation, the parameter θ represents the weight of the neural network.

[0188] Furthermore, the computational fluid dynamics (CFD) algorithm in the temperature gradient precise adjustment module is based on the finite volume method, adopts the RNG k-ε turbulence model, and combines the multiphase flow model to simulate the flow and heat transfer process of the dye solution, optimize the control strategy of the heating element, and achieve the uniformity of the temperature field in the dye vat.

[0189] Specifically, the RNG k-ε turbulence model is applied as follows:

[0190] A). Establish temperature field model

[0191] Finite Volume Method: CFD uses the finite volume method to divide the space inside the dye vat into many small control volumes. The energy conservation equations within each control volume are used to calculate heat transfer and flow.

[0192] RNG k-ε turbulence model: used to simulate the turbulent behavior of dye liquor. The k-ε model can effectively capture the turbulent eddies in the dye liquor flow, especially in the heat exchange process between different temperature layers. These small-scale flows are crucial for accurate control of temperature distribution.

[0193] Multiphase flow model: simulates the flow and interaction of different phases (such as liquid, bubbles, etc.) that may exist in the dye solution. This is very important for accurately simulating the real flow conditions of the dye solution.

[0194] B). Simulation and calculation:

[0195] Input parameters: including heater position and power, dye flow rate, initial temperature distribution, etc.

[0196] Boundary conditions: Set the temperature and flow conditions of the dye vat wall, heating element surface, and dye liquid surface. The CFD model uses these conditions to simulate the heat flow distribution in the dye vat.

[0197] Solution: Use CFD software to solve the control equations, obtain the temperature, velocity, and pressure distribution of each point in the dye vat, and form a three-dimensional temperature field.

[0198] C). Optimize heat flow distribution:

[0199] Hot spot elimination: Reduce local overheating by adjusting the output power and position of the heating elements. CFD simulation helps identify possible hot spots and eliminate them by optimizing the heating strategy.

[0200] Cold zone compensation: By changing the power distribution of the heater or increasing local stirring, the temperature of the lower temperature area is increased to achieve uniformity of the overall temperature field.

[0201] D). Feedback and Adjustment:

[0202] Real-time adjustment: The thermal imaging camera array provides real-time temperature data, and the CFD model dynamically simulates based on this data to provide the best heating strategy. The system continuously adjusts the power and position of the heating elements to maintain the ideal temperature distribution.

[0203] Multiple Iterations: Repeatedly iterate the CFD model based on real-time data to continuously optimize heat flow distribution and ensure temperature uniformity and stability.

[0204] Furthermore, the neural network PID control algorithm in the dye solution pH automatic balancing unit adopts BP neural network, optimizes PID parameters through online learning, and improves the response speed and stability of pH value regulation. The algorithm takes pH value deviation and its change rate as input, outputs optimized PID parameters, and realizes adaptive regulation of pH value control.

[0205] Specifically, the BP neural network control algorithm is applied as follows:

[0206] A). Real-time data collection and status monitoring

[0207] The pH value of the dye solution is continuously monitored using an ion selective electrode. Data is collected every second to ensure real-time information on pH changes.

[0208] At the same time, the temperature of the dye solution, dye concentration, acid and alkali addition and other parameters are monitored, and these data serve as the basic input for control and prediction.

[0209] B) Simple prediction and early warning

[0210] Historical data recording: The system continuously collects and records historical data of dye solution pH value and related parameters to form a data record library.

[0211] Simple moving average prediction: Using historical data, the system can calculate a simple moving average or weighted average to predict future pH trend changes. For example, using the data from the last 5 minutes to calculate the average pH change rate as a simple prediction of the short-term trend.

[0212] Early warning mechanism: If the simple prediction results show that the future pH value may exceed the set safety range (such as the deviation exceeds ±0.1), the system will issue a warning signal in advance and prepare to make adjustments.

[0213] C). Real-time adjustment of neural network PID control

[0214] Input layer: The input of neural network PID control includes the current pH value deviation e(t) (i.e. the difference between the current pH value and the set value), the deviation change rate de(t) / dt, and simple predicted future deviation information.

[0215] Hidden layer and output layer: BP neural network processes input data, analyzes current status and future trends, and outputs optimized PID parameters K p ,K i ,K d These parameters are used to adjust the control strategy of acid and alkali addition in real time.

[0216] Real-time regulation: Based on optimized PID parameters, the system quickly adjusts the injection amount of acid or base. Through the microinjection system, acid or base is accurately added to quickly respond to current pH changes and predicted trends. The neural network PID controller updates parameters in real time to ensure that the system responds quickly without introducing oscillations or overshoot.

[0217] D) Feedback adjustment and adaptive learning

[0218] Feedback mechanism: The system adjusts the control strategy based on the difference between the real-time monitored pH value and the set value. After each adjustment, the actual pH value change is recorded and used to further optimize the neural network.

[0219] Online learning: The BP neural network continuously adjusts its weights and parameters through online learning to adapt to changes in process conditions. The system optimizes control parameters based on feedback data to improve control accuracy and robustness.

[0220] E). Continuous optimization and stable control

[0221] Comprehensive optimization: Combining real-time pH adjustment and simple prediction, the system can dynamically optimize the amount of acid and alkali added, avoid over-adjustment or delayed response, and keep the pH value stable in the ideal range.

[0222] Control stability and resource utilization: Through early prediction and timely adjustment, the system reduces the excessive use of acid and alkali, improves resource utilization efficiency, and reduces the problem of uneven dyeing caused by pH fluctuations.

[0223] Furthermore, the machine vision algorithm in the fabric dyeing uniformity monitoring module adopts a convolutional neural network (CNN) structure, which identifies the unevenness of color distribution through real-time analysis of the fabric surface image, and feeds back the analysis results to the intelligent proportioning and precise feeding control unit to form a closed-loop control to ensure the uniformity of the dyeing process.

[0224] Specifically, the application of the convolutional neural network (CNN) is as follows:

[0225] A). Initial status monitoring and preparation

[0226] a). Configuration of ultrasonic array

[0227] Before dyeing begins, the system activates the ultrasonic array. Sensors are evenly distributed above and on the sides of the dye vat, and each sensor emits ultrasonic pulses at a fixed time. These pulses are reflected back after encountering the fabric surface.

[0228] The system measures the reflection time and intensity of the ultrasonic waves to calculate the height, immersion depth and tilt angle of the fabric in real time. These data are used to build a three-dimensional position model of the fabric in the dye vat.

[0229] b). Ensure even soaking

[0230] The initial monitoring data from the ultrasonic sensor is used to confirm the immersion status of the fabric in the dye solution. The system checks whether the height and tilt angle of the fabric are within the preset range. If it detects that a part of the fabric is floating or sinking, the system will sound an alarm to indicate that adjustment is needed.

[0231] Through these data, the system ensures that the fabric is evenly immersed in the dye liquor, laying the foundation for the subsequent dyeing process.

[0232] B). Image acquisition and color distribution analysis

[0233] a). Real-time image acquisition

[0234] After the dyeing process begins, the system continuously captures the surface image of the fabric through cameras installed above and on the sides of the dyeing tank. The camera positions are designed to ensure that the entire fabric surface is covered, including the edges and center areas of the dyeing tank.

[0235] The image acquisition frequency is set at 2 frames per second to ensure that the color changes on the fabric surface can be reflected in time. The acquired image data is transmitted to the central processing unit in real time.

[0236] b). Image preprocessing

[0237] The acquired images first go through a series of preprocessing steps to ensure the accuracy of the analysis:

[0238] Denoising: Use a Gaussian filter to remove random noise from the image.

[0239] Contrast enhancement: Improve the contrast of the image through histogram equalization, making the color changes more obvious.

[0240] Color correction: Adjust image colors based on a standard white balance algorithm to make image colors consistent under different lighting conditions.

[0241] c). Model architecture

[0242] The system uses a pre-trained deep convolutional neural network (CNN) model to analyze images. The CNN model includes multiple convolutional layers and pooling layers to extract features from the image.

[0243] Convolutional layers: Use 3x3 and 5x5 convolution kernels to extract local features in the image. Low-level convolutional layers recognize simple features, such as color boundaries, and high-level convolutional layers recognize more complex features, such as color spots and gradients.

[0244] Pooling layer: The maximum pooling operation reduces the dimension of the data while retaining key feature information. The pooled data is input into the fully connected layer to generate the final uniformity score.

[0245] d). Model training

[0246] The CNN model was trained during the development phase using a large dataset of dyed fabric images that included annotations of different dyeing conditions and unevenness. Through repeated training, the CNN model learned to recognize and distinguish the features of uniform and uneven dyeing.

[0247] e). Real-time analysis and scoring

[0248] In practical applications, the CNN model processes the preprocessed image and outputs a score about the color uniformity of the fabric. The score is between 0 and 1, and the higher the score, the more uniform the dyeing. The system sets a threshold (such as 0.8), and scores below the threshold will be marked as potential uneven dyeing areas.

[0249] C) Collaboration and feedback mechanism

[0250] a). Combination of ultrasonic and visual data

[0251] During the dyeing process, the system continuously acquires data from both the ultrasonic array and machine vision. Ultrasonic waves provide the three-dimensional position and soaking status of the fabric, while machine vision provides information on the color distribution on the fabric surface.

[0252] The system combines the positional anomalies detected by ultrasound (such as fabric floating or sinking) with the color unevenness information detected by machine vision. Through this combination, the system can confirm which uneven dyeing phenomena may be caused by positional anomalies.

[0253] b). Feedback and adjustment

[0254] The system generates feedback signals based on the combined analysis of ultrasound and machine vision. If uneven dyeing is detected, the system locates the specific area and problem and generates adjustment instructions.

[0255] The system automatically adjusts the path and speed of the fabric in the dye vat to ensure that these problem areas are corrected. Adjustment strategies include changing the direction of the fabric's movement, adjusting the soaking time, or adjusting the fabric's position in the dye bath.

[0256] D. Continuous monitoring and closed-loop control

[0257] a). Continuous monitoring

[0258] Throughout the dyeing process, the system continuously collects new images and ultrasonic data. Through continuous monitoring, the system can quickly detect new unevenness and make timely adjustments.

[0259] b). Closed-loop control and optimization

[0260] The system uses real-time feedback information to form a closed-loop control mechanism. After each adjustment, the system immediately collects and analyzes new data to verify the adjustment effect. If further unevenness is detected, the system will continue to optimize the adjustment strategy.

[0261] Combination Figure 2 , the system workflow is as follows:

[0262] S1. Dye solution preparation:

[0263] After the system is started, the dye liquor is initially prepared in the dye vat, mixing indigo dye, reducing agent and other necessary auxiliaries.

[0264] S2. Indigo concentration monitoring:

[0265] The real-time monitoring module for indigo concentration is started, using near-infrared spectroscopy analysis technology to collect dye liquor samples every 30 seconds through a self-cleaning eddy current sampling device to continuously monitor the indigo concentration.

[0266] S3. pH value monitoring:

[0267] At the same time, pH monitoring is carried out. The ion-selective electrode of the dye solution pH automatic balancing unit measures the pH value every 10 seconds to ensure the chemical stability of the dye solution.

[0268] S4. Intelligent ratio adjustment:

[0269] The intelligent proportioning and precise feeding control unit receives real-time data from indigo concentration monitoring and pH monitoring. Using adaptive fuzzy control algorithms, the system calculates the amount of various raw materials that need to be adjusted to achieve the preset indigo concentration and pH value targets.

[0270] S5. Automatic feeding:

[0271] Based on the calculation results of intelligent ratio adjustment, the system adds indigo dye through a high-precision peristaltic pump (accuracy ±0.1%) and adds small doses of additives such as reducing agent Na2S2O4 through a microfluidic device (minimum metering 0.1mL).

[0272] S6. Reduction potential control:

[0273] The platinum-palladium alloy electrode of the reduction potential closed-loop regulation module continuously monitors the reduction potential of the dye solution, with a target value of -650mV (vs. Ag / AgCl). The gas-liquid mixing microfluidic device (working pressure 0.3MPa) accurately adds the reducing agent according to the detection results. The model predictive control (MPC) algorithm adjusts the amount of reducing agent added in real time, with the prediction time domain set to 5 minutes and the control time domain to 1 minute.

[0274] S7. Fabric dyeing:

[0275] Fabric (such as 100% cotton, weight 200g / m 2 ) enters the dyeing tank at a preset speed (such as 2m / min) to start the dyeing process. The system controls the immersion speed and dyeing time of the fabric to ensure a full and uniform dyeing effect.

[0276] S8. Oxidation process control:

[0277] After the dyeing is completed (about 30 minutes), the fabric enters the oxidation stage. The multi-stage programmable gas diffusion device of the oxidation degree dynamic control unit is activated, and the initial oxygen flow rate is set to 10L / min. The high-speed camera system (1000fps) captures the color changes of the fabric in real time. The deep reinforcement learning algorithm (based on the DQN architecture) updates the control strategy every 5 seconds and dynamically adjusts the oxidation parameters.

[0278] S9.Quality Assessment:

[0279] The system comprehensively analyzes the data of each module, including the final dyeing uniformity, color fastness and other indicators. A comprehensive quality assessment is carried out every 5 minutes to detect whether the preset quality standards (such as color difference ΔE*ab<0.5) are met. If any abnormality is found, the system automatically adjusts the relevant parameters.

[0280] S10. Complete the dyeing process:

[0281] When the system confirms that the dyeing process has reached the preset quality standard, the dyeing process ends. The system records all the parameters of this dyeing, including the final indigo concentration, pH value, temperature curve, reduction potential change, oxidation time, etc., to provide reference for subsequent batches.

[0282] S11. Temperature gradient adjustment:

[0283] The temperature gradient precision adjustment module works continuously throughout the process. The zoned electromagnetic induction heating device (power 30kW) maintains the temperature of the dye vat at a preset level (e.g. 50℃±0.5℃). The thermal imager array (resolution 640x480, temperature resolution 0.05℃) updates the temperature field model every 5 seconds. The computational fluid dynamics (CFD) algorithm continuously optimizes the heat flow distribution to ensure the uniformity of the temperature field.

[0284] S12. Fabric dyeing uniformity monitoring:

[0285] The fabric dyeing uniformity monitoring module works continuously during the fabric dyeing, oxidation process control and quality assessment stages. The ultrasonic array (frequency 1MHz) scans the fabric position 100 times per second. The electromagnetic induction sensor (sensitivity 0.1g / L) monitors the fabric's absorption of dye in real time. The machine vision system (resolution 1920x1080, frame rate 60fps) analyzes the color distribution on the fabric surface and promptly detects and corrects unevenness.

[0286] Through the above implementation, the automatic feeding and proportioning control system of the blue dyeing process of the present invention realizes intelligent and precise control of the whole process of blue dyeing by integrating modules such as real-time monitoring of indigo concentration, intelligent proportioning and precise feeding, closed-loop regulation of reduction potential, dynamic control of oxidation degree, precise regulation of temperature gradient, automatic balance of pH value and monitoring of fabric dyeing uniformity. The system significantly improves dyeing uniformity, indigo utilization and production efficiency, while reducing resource consumption, providing strong support for the modernization and sustainable development of blue dyeing technology.

[0287] The above description of the present invention is not intended to limit the present invention. Under the guidance of the present invention, those skilled in the art may make various changes and improvements without departing from the scope of protection of the present invention and the claims, which all belong to the protection scope of the present invention. The technologies not described in detail in the present invention are all prior art.

[0288] It will be appreciated by those skilled in the art that various modifications and variations may be made to the present invention without departing from the spirit and scope of the present invention. These modifications and variations all fall within the scope of protection of the present invention. The scope of protection of the present invention is defined in the appended claims.

[0289] The protection scope of the present invention is not limited to the above embodiments. Any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention fall within the protection scope of the present invention.

Claims

1. An automatic feeding and proportioning control system for blue dyeing process, characterized in that: include: Indigo concentration real-time monitoring module, used to continuously monitor the concentration of indigo in the dye solution; Intelligent proportioning and precise feeding control unit, used to accurately control the addition of various raw materials according to formula requirements and real-time monitoring data; Reduction potential closed-loop regulation module, used to accurately control the reduction potential of the dye solution; Dynamic oxidation control unit for precise control of the oxidation process of fabrics; Temperature gradient precise adjustment module, used to achieve precise control and uniform distribution of temperature in the dyeing tank; Dye solution pH automatic balancing unit, used to adjust and stabilize the pH value of the dye solution in real time; Fabric dyeing uniformity monitoring module, used to monitor the dyeing status of fabrics in dyeing liquid in real time; The indigo concentration real-time monitoring module is connected to the intelligent proportioning and precise feeding control unit to provide it with real-time indigo concentration data; The intelligent proportioning and precise feeding control unit is connected to the reduction potential closed-loop adjustment module to adjust the addition of the reducing agent according to the proportioning result; The reduction potential closed-loop regulation module is connected to the oxidation degree dynamic control unit to provide reduction state information thereto; The temperature gradient precise adjustment module is respectively connected to the intelligent proportioning precise feeding control unit, the reduction potential closed-loop adjustment module and the oxidation degree dynamic control unit to adjust the temperature of each process; The dye liquor pH value automatic balancing unit is connected to the intelligent proportioning and precise feeding control unit to provide real-time pH value information; The fabric dyeing uniformity monitoring module is respectively connected to the intelligent proportioning and precise feeding control unit and the oxidation degree dynamic control unit to provide fabric dyeing status information.

2. The system according to claim 1, characterized in that The indigo concentration real-time monitoring module comprises: Near-infrared spectroscopy is used to measure the concentration of indigo in the dye solution; Self-cleaning eddy current sampling device is used to prevent dye deposition from affecting measurement accuracy.

3. The system according to claim 1, characterized in that The intelligent proportioning and precise feeding control unit comprises: High-precision peristaltic pump and solenoid valve array to control the feeding of multiple raw materials; Microfluidics technology to ensure precise addition of small doses of additives; Adaptive fuzzy control algorithm for dynamic adjustment of feed rate and ratio.

4. The system according to claim 1, characterized in that The reduction potential closed-loop regulation module comprises: Platinum-palladium alloy electrode, used to measure reduction potential; A gas-liquid mixing microfluidizer is used to improve the dispersion efficiency of the reducing agent; Model predictive control (MPC) algorithm is used to realize feedforward-feedback composite control of reduction potential.

5. The system according to claim 1, characterized in that The oxidation degree dynamic control unit comprises: Multi-stage programmable gas diffusion device, used to achieve segmented precise control of the oxidation process; High-speed camera and image processing system for real-time analysis of fabric color changes; Deep reinforcement learning algorithm for dynamic optimization of oxidation parameters.

6. The system according to claim 1, characterized in that The temperature gradient precise adjustment module comprises: Partitioned electromagnetic induction heating technology is used to achieve precise temperature control in the dye vat; Thermal imager array, used to construct a three-dimensional temperature field model inside the dye vat; Computational fluid dynamics (CFD) algorithms for optimizing heat flow distribution.

7. The system according to claim 1, characterized in that The dye liquor pH value automatic balancing unit comprises: Ion-selective electrodes for measuring pH; Microinjection system for precise addition of acid and alkali; Neural network PID control algorithm is used to improve the response speed and stability of pH value regulation.

8. The system according to claim 1, characterized in that The fabric dyeing uniformity monitoring module comprises: Ultrasonic array technology, used to scan the position and posture of the fabric in the dye solution in real time; Electromagnetic induction sensors to monitor the degree of dye absorption by fabrics; Machine vision algorithms to analyze color distribution on fabric surfaces.

Images