Intelligent control-based dyeing process optimization method and system

Abstract

The application relates to the technical field of intelligent control, and discloses a dyeing process optimization method and system based on intelligent control. The method comprises the following steps: collecting key parameters of a dyeing process to obtain an original data set; pre-processing and feature extraction are performed on the data set to obtain a feature data set; a deep learning prediction model is constructed by using the feature data set, and multi-objective optimization is performed by using a non-dominated sorting genetic algorithm to obtain an optimal process parameter combination; and scheme analysis is performed based on the optimal parameters to obtain an intelligent dyeing scheme. The application realizes multi-parameter collaborative optimization of the whole process of the dyeing process, reduces the dependence on human experience, improves the consistency of dyeing quality, has self-learning and self-adaptive capabilities, can automatically adjust and optimize strategies according to different fabric characteristics and production conditions, and thus solves problems such as large quality fluctuation, high energy consumption and poor adaptability in traditional dyeing processes.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent control, in particular to a dyeing process optimization method and system based on intelligent control. BACKGROUND

[0002] As a key link in the textile industry, the dyeing quality and process stability of dyeing process directly affect the commercial value and market competitiveness of products. Traditional dyeing process mainly relies on worker experience and fixed process formula for control, usually using single parameter control method, that is, setting temperature curve, dye ratio, additive amount and other parameters according to experience, and maintaining these parameters in the set range through simple PID controller. With the development of automation technology, intelligent control technology based on fuzzy control, expert system and other methods is introduced into the dyeing industry, realizing automatic adjustment of part of parameters such as temperature control and liquid level control. However, these methods mostly control single or a few parameters independently, lack the ability of multi-parameter collaborative optimization in the whole dyeing process, and are difficult to adapt to the changes in characteristics of different types of fabrics and dyes.

[0003] The existing dyeing process control method has obvious deficiencies: first, the traditional experience-based control method has strong dependence on worker skills, and the process parameter setting lacks scientific basis, resulting in large dyeing quality fluctuation and poor batch consistency; second, the existing automatic control system is mostly single parameter control, ignoring the complex interaction between dyeing parameters, and cannot realize global optimization; third, the existing control method lacks learning and self-adaptive ability, and it is difficult to automatically adjust and optimize parameters according to the actual dyeing effect; fourth, the dyeing process data collection and utilization degree is insufficient, and a large amount of valuable production data has not been fully mined and applied; fifth, the existing control system usually uses the same control strategy to process different types of fabrics and dye combinations, and cannot perform differentiated control according to the characteristics of specific dyeing process, reducing the adaptability and flexibility of the process. SUMMARY

[0004] The present application provides a dyeing process optimization method and system based on intelligent control, which is used to realize multi-parameter collaborative optimization in the whole dyeing process, reduce the dependence on human experience, improve the consistency of dyeing quality, and has self-learning and self-adaptive ability, which can automatically adjust and optimize the strategy according to different fabric characteristics and production conditions, thereby solving the problems of large quality fluctuation, high energy consumption and poor adaptability in traditional dyeing process.

[0005] In a first aspect, the application provides a dyeing process optimization method based on intelligent control, which comprises: collecting key parameters of a dyeing process to obtain a dyeing process original data set; performing preprocessing and feature extraction based on the dyeing process original data set to obtain a dyeing process feature data set; constructing a deep learning prediction model using the dyeing process feature data set and performing multi-objective optimization to obtain an optimal dyeing process parameter combination, wherein the multi-objective optimization adopts a non-dominated sorting genetic algorithm; and performing dyeing scheme analysis according to the optimal dyeing process parameters to obtain an intelligent dyeing scheme.

[0006] In a second aspect, the application provides a dyeing process optimization system based on intelligent control, which comprises:

[0007] a collection module configured to collect key parameters of a dyeing process to obtain a dyeing process original data set;

[0008] an extraction module configured to perform preprocessing and feature extraction based on the dyeing process original data set to obtain a dyeing process feature data set;

[0009] an optimization module configured to construct a deep learning prediction model using the dyeing process feature data set and perform multi-objective optimization to obtain an optimal dyeing process parameter combination, wherein the multi-objective optimization adopts a non-dominated sorting genetic algorithm;

[0010] an analysis module configured to perform dyeing scheme analysis according to the optimal dyeing process parameters to obtain an intelligent dyeing scheme.

[0011] In a third aspect, a dyeing process optimization device based on intelligent control is provided, which comprises: a memory and at least one processor, wherein the memory stores instructions; and the at least one processor invokes the instructions in the memory to enable the dyeing process optimization device based on intelligent control to perform the dyeing process optimization method based on intelligent control described above.

[0012] In a fourth aspect, a computer readable storage medium is provided, which stores instructions, and when the instructions are run on a computer, the computer is enabled to perform the dyeing process optimization method based on intelligent control described above.

[0013] In the technical scheme provided in the application, the key parameters of the dyeing process are collected to obtain a dyeing process original data set, the overall capture and digital expression of the data of the whole dyeing process are realized, and a data foundation is laid for subsequent intelligent analysis; the dyeing process original data set is preprocessed and feature extraction is performed to obtain a dyeing process feature data set, effectively solving the problems of large data noise and insufficient feature extraction in the traditional dyeing process, and significantly improving the data quality and feature expression capability; a deep learning prediction model is constructed using the dyeing process feature data set and multi-objective optimization is performed, not only overcoming the defect of limited expression capability of the traditional single model, but also realizing accurate prediction of the dyeing quality through the complementary advantages of processing non-time sequence features by a multi-layer perception, capturing local time sequence patterns by a convolutional neural network, and modeling long-term dependence by a recurrent neural network, and the non-dominated sorting genetic algorithm used in the multi-objective optimization can simultaneously optimize multiple targets such as color fastness, color difference, level dyeing, color mottle rate, dyeing time and energy consumption, and find the best compromise solution; a dyeing control system of a hierarchical control architecture is realized according to the optimal dyeing process parameter combination, an intelligent dyeing scheme capable of automatically adjusting process parameters is obtained, the hierarchical control architecture includes a field execution layer, a process control layer and an optimization decision layer, the control granularity is more fine, the response speed is more matched with the control demand, and the dependence on manual experience of the traditional dyeing process is greatly reduced. In particular, in the specific application field of the dyeing process, the application provides a special solution for the complex nonlinear characteristics, multi-parameter coupling problems and multi-objective optimization requirements of the dyeing process by fusing the features of multiple deep learning algorithms and the non-dominated sorting genetic algorithm, and the core contributions of the algorithm features to the scheme are as follows: the multi-structure fusion characteristics of the deep learning algorithm accurately capture the complex relationship between the static features and the dynamic time sequence features in the dyeing process; the multi-objective balancing characteristics of the non-dominated sorting genetic algorithm effectively solve the trade-off problem between the dyeing quality and the production efficiency; and the hierarchical characteristics of the hierarchical control architecture perfectly adapt to the requirements of different control granularities and time scales in the dyeing process, and realize the comprehensive optimization effect of stable dyeing quality, reduced energy consumption and improved production efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0014] In order to more clearly illustrate the technical scheme of the embodiments of the application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the application, and other drawings can be obtained by those skilled in the art without creating labor based on these drawings.

[0015] Figure 1 An embodiment of the dyeing process optimization method based on intelligent control in the embodiments of the application is shown in the figure.

[0016] Figure 2An embodiment of a system for optimizing a dyeing process based on intelligent control is shown in the figure;

[0017] Figure 3 An embodiment of a system for optimizing a dyeing process based on intelligent control is shown in the figure. DETAILED DESCRIPTION

[0018] The embodiment of the present application provides a method and system for optimizing a dyeing process based on intelligent control. The terms "first", "second", "third", "fourth" and the like (if any) in the specification and claims of the present application and the above-described drawings are used to distinguish similar objects, and do not necessarily have to be used to describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the term "comprising" or "having" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device comprising a series of steps or units does not have to be limited to only those steps or units clearly listed, but can include other steps or units that are not clearly listed or inherent to the process, method, product or device.

[0019] For ease of understanding, the specific process of the embodiment of the present application is described below. Please refer to Figure 1 An embodiment of a method for optimizing a dyeing process based on intelligent control comprises the following steps:

[0020] Step S101, collecting key parameters of a dyeing process to obtain a dyeing process original data set;

[0021] Step S102, pre-processing and feature extraction based on the dyeing process original data set to obtain a dyeing process feature data set;

[0022] Step S103, constructing a deep learning prediction model using the dyeing process feature data set and performing multi-objective optimization to obtain an optimal dyeing process parameter combination, the multi-objective optimization using a non-dominated sorting genetic algorithm;

[0023] Step S104, dyeing scheme analysis according to the optimal dyeing process parameters to obtain an intelligent dyeing scheme.

[0024] It can be understood that the execution subject of the present application can be a system for optimizing a dyeing process based on intelligent control, and can also be a terminal or a server, and the specific execution subject is not limited herein. The embodiment of the present application takes a server as an execution subject for example.

[0025] Specifically, key parameters of the dyeing process are collected to obtain a dyeing process original dataset. This step collects key parameters such as dye concentration, dyeing temperature, dyeing time, liquor ratio, pH value, additive amount, and dyeing machine speed through setting multiple sensor collection points for real-time monitoring. These sensor collection points are connected to an industrial programmable logic controller to form a field control unit, realizing preliminary processing of the data. Specifically, for dyeing temperature collection, a PT100 temperature sensor is used, with an accuracy of not less than 0.1°C; for pH value collection, a pH sensor is used, with an accuracy of not less than 0.01; and for concentration collection, a concentration sensor is used, with an accuracy of not less than 0.1%. The collected data is transmitted to a central data server through an industrial Ethernet, and a dual-machine hot backup mechanism is established to ensure data integrity and reliability. For example, in the four-in-one fiber three-bath dyeing process of silk cotton polyester nylon, the hydrogen peroxide parameters are collected at 40°C in the scouring and dyeing stage; the emulsifier and dye addition stage records the 130X40 and solvay parameters; the temperature rising stage records the temperature curve with a temperature rising rate of 1.5 degrees per minute until the temperature reaches 80°C; the special side cleaning stage records the parameters of the 80°C temperature holding state; the cooling stage records the cooling curve from 80°C to 40°C; and the deamination and water washing stages record the parameters of 40°C X20 washing water.

[0026] Based on the dyeing process original dataset, preprocessing and feature extraction are performed to obtain a dyeing process feature dataset. An adaptive anomaly detection algorithm is executed to clean the original data, identify and process missing values, outliers, and duplicate data. For missing values, linear interpolation, k-nearest neighbor interpolation, or forward filling is used for completion according to the data type; for outliers, 3σ rule or box plot rule is used for identification, and correction or rejection is performed according to the trend of the data before and after; for duplicate data, the first collected valid data point is retained and the subsequent duplicate values are deleted. After data cleaning, standardization processing is performed through nonlinear transformation, and parameters of different dimensions are uniformly converted to standard normal distribution. Then, a filter is applied to the standardized data for time series noise reduction processing. Savitzky-Golay filter is used to smooth the time series data of each parameter, with a filter window width of 15 data points and a polynomial order of 3, balancing the filtering effect and edge fidelity. Then, wavelet coefficients and statistical features are extracted from the smoothed data to form a multi-dimensional feature matrix, including mean, standard deviation, peak value, valley value, slope, skewness, and kurtosis time domain features, as well as power spectral density, main frequency component, and frequency band energy distribution obtained through fast Fourier transform. Dimensionality reduction technique is performed on the multi-dimensional feature matrix to compress the features, obtaining a low-dimensional feature set. The low-dimensional feature set is fused with the dyeing process mechanism features to generate a dyeing process feature dataset, which contains not less than 30 feature dimensions, and each dyeing batch forms a multi-dimensional feature record.

[0027] The deep learning prediction model is constructed by using the dyeing process characteristic data set and multi-objective optimization is performed to obtain the optimal dyeing process parameter combination. The dyeing process characteristic data set is divided into training set, validation set and test set according to the ratio of 7:2:1, and then the multi-structure deep learning model containing multi-layer perception, convolutional neural network and recurrent neural network is constructed based on the training set. The multi-layer perception contains 4 hidden layers, the number of neurons is 128, 256, 128 and 64 respectively, and the activation function adopts ReLU; the convolutional neural network contains 3 convolutional blocks, each convolutional block contains 2 convolutional layers and 1 maximum pooling layer, the convolution kernel size is 3×1, the initial channel number is 32, and the channel number doubles after each convolutional block; the recurrent neural network adopts a bidirectional structure, contains 2 layers of hidden state dimension of 128 gated recurrent units, and the output layer adds attention mechanism. The model is optimized through the validation set to obtain the dyeing quality prediction candidate model, and then the performance is evaluated and verified by using the test set to obtain the deep learning prediction model capable of predicting color fastness, color difference value, level dyeing and color flower rate. Based on the deep learning prediction model, the multi-objective optimization variables and constraint conditions are set, the variables include 12 key parameters such as dyeing bath temperature, dye ratio and holding time. The non-dominated sorting genetic algorithm is calculated on these variables to realize multi-objective optimization. The algorithm adopts real number coding method, the population size is set to 100, the evolution algebra is set to 500, the selection operation adopts tournament selection method, the crossover operation adopts simulated binary crossover, the crossover probability is set to 0.9, the distribution index is set to 20, the mutation operation adopts polynomial mutation, and the mutation probability is set to 1 / 12. The Pareto optimal solution set is obtained by using the optimization algorithm, and then the best compromise solution is screened out from the set by using the weighted TOPSIS method to obtain the optimal dyeing process parameter combination.

[0028] According to the optimal dyeing process parameters, the intelligent dyeing scheme is obtained by analyzing the dyeing scheme. The optimal dyeing process parameter combination is mapped and converted to obtain a standardized control parameter matrix. First, the parameters are sorted according to the process flow sequence to form a process parameter sequence, including the parameters of the dyeing bath, emulsified dye, temperature rise, temperature maintenance, temperature drop and deamination ester process. Then, the process parameter sequence is subjected to dimension normalization processing to unify the parameters of different physical quantities to the interval of 0-1 to form a unified dimension process parameter matrix. Based on the unified dimension process parameter matrix, a correlation diagram between process variables is constructed to obtain a parameter correlation network, and the Pearson correlation coefficient is used to quantify the dependence between parameters. Cluster analysis is performed on the parameter correlation network to identify highly coupled parameter groups, and joint optimization conversion is performed to obtain a decoupled parameter matrix. Then, the decoupled parameter matrix is restructured according to the control level requirements to obtain a standardized control parameter matrix. Subsequently, the standardized control parameter matrix is subjected to hierarchical analysis to obtain a field execution layer parameter set, a process control layer parameter set and an optimization decision layer parameter set. Based on the three parameter sets, a process execution control strategy is constructed to form a dynamic control execution scheme, and the scheme is applied to the dyeing process flow. Through real-time parameter monitoring and feedback adjustment, an intelligent dyeing scheme is obtained.

[0029] In the embodiment of the present application, the key parameters of the dyeing process are collected to obtain the dyeing process original data set, the overall capture and digital expression of the dyeing process data are realized, and the data foundation for subsequent intelligent analysis is laid; the dyeing process feature data set is obtained by preprocessing and feature extraction based on the dyeing process original data set, the problems of large data noise and insufficient feature extraction in the traditional dyeing process are effectively solved, and the data quality and feature expression ability are significantly improved; the deep learning prediction model is constructed by using the dyeing process feature data set, and multi-objective optimization is performed, which not only overcomes the defect of limited expression ability of the traditional single model, but also realizes the accurate prediction of the dyeing quality by the complementary advantages of the multi-layer perception machine processing non-time sequence features, the convolutional neural network capturing local time sequence mode, and the recurrent neural network modeling long-term dependence relationship, and the non-dominated sorting genetic algorithm used in multi-objective optimization can optimize multiple targets such as color fastness, color difference, uniformity, color mottle rate, dyeing time and energy consumption at the same time, and find the best compromise solution; the dyeing control system of the hierarchical control architecture is realized according to the optimal dyeing process parameter combination, the intelligent dyeing scheme capable of automatically adjusting the process parameters is obtained, the hierarchical control architecture includes the field execution layer, the process control layer and the optimization decision layer, the control granularity is more fine, the response speed is more matched with the control demand, the dependence on manual experience of the traditional dyeing process is greatly reduced, especially in the specific application field of the dyeing process, the present application provides a special solution for the complex nonlinear characteristics, multi-parameter coupling problems and multi-objective optimization requirements of the dyeing process by fusing multiple deep learning algorithms and non-dominated sorting genetic algorithm features, and the core contributions of the algorithm features to the scheme are as follows: the multi-structure fusion characteristics of the deep learning algorithm accurately capture the complex relationship between the static features and the dynamic time sequence features in the dyeing process; the multi-objective balance characteristics of the non-dominated sorting genetic algorithm effectively solve the trade-off problem between the dyeing quality and the production efficiency; the hierarchical characteristics of the hierarchical control architecture perfectly adapt to the requirements of different control granularity and time scale in the dyeing process, and realize the comprehensive optimization effect of stable dyeing quality, reduced energy consumption and improved production efficiency.

[0030] In a specific embodiment, the process of step S101 can specifically include the following steps:

[0031] A plurality of sensor collection points are set for dye concentration, dyeing liquid temperature, dyeing time, liquid ratio, pH value, additive amount and dyeing machine rotating speed to obtain a multi-dimensional parameter collection network;

[0032] An industrial programmable logic controller is connected to the multi-dimensional parameter collection network to obtain a field control unit, and the programmable logic controller is equipped with a processor with a frequency of not less than 100 MHz;

[0033] The collected data is processed by using the field control unit to execute a data filtering algorithm to obtain parameter data;

[0034] The parameter data is transmitted to a central data server through an industrial Ethernet to obtain a parameter data stream;

[0035] A dual-computer hot backup mechanism is set for the parameter data stream to obtain a data storage scheme;

[0036] Based on the data storage scheme, the parameters in the whole dyeing process are continuously collected to obtain a dyeing process original data set, which contains parameter records of all processes from the scouring and dyeing bath to the ammonia-ester water washing and rolling of cotton.

[0037] Specifically, in the dyeing process, multiple sensor collection points are set for key parameters such as dye concentration, dye temperature, dyeing time, liquid ratio, pH value, additive amount, and dyeing machine speed, etc., to construct a multi-dimensional parameter collection network. For example, in the scouring and dyeing bath stage, the temperature of hydrogen peroxide needs to be accurately monitored and kept at 40°C; in the warming-up stage, it needs to be ensured that the stable warming rate is 1.5 degrees per minute until it reaches 80°C; in the temperature maintaining stage, it needs to be monitored that the temperature is maintained at 80°C; in the cooling-down stage, it needs to be recorded that the temperature drops from 80°C to 40°C. The data collected by these sensors directly affects the dyeing quality and process control accuracy.

[0038] The multi-dimensional parameter collection network is connected to an industrial programmable logic controller (PLC) to form a field control unit. These controllers are equipped with processors with a frequency of not less than 100 MHz to ensure sufficient computing power to handle high-frequency data collection tasks. The PLC system is usually configured with not less than 8 MB of memory, not less than 32 digital quantity input and output points, and not less than 8 analog quantity input and output points to meet the control needs of complex dyeing processes. The sensors communicate with the PLC through 4-20 mA standard signal output and RS485 bus to ensure stable and reliable signal transmission. The field control unit executes a data filtering algorithm to process the original collected data and remove obvious noise and abnormal points. The data filtering algorithm includes moving average processing, median filtering, and outlier detection to preliminarily purify the original sensor data. For example, when the temperature sensor has abnormal fluctuations in a short period of time, the moving average algorithm can smooth these fluctuations; when the pH value sensor occasionally has abnormal readings, the median filtering can effectively eliminate these disturbances. After these processes, more reliable and continuous parameter data are obtained.

[0039] The preliminarily processed parameter data is transmitted to a central data server through an industrial Ethernet to form a continuous parameter data stream. The industrial Ethernet uses the OPC UA communication protocol to ensure the real-time and reliability of data transmission. The data stream is recorded with an accuracy of not less than 10 data points per second to capture subtle changes such as temperature fluctuations, chemical reaction changes, and physical state changes in the dyeing process.

[0040] A dual hot backup mechanism is set for the parameter data stream to establish a robust data storage scheme. When the main system fails, the standby system can seamlessly take over the data acquisition task within 3 seconds, avoiding data loss. The data storage uses a relational database structure to store historical data, and a real-time database to process real-time data stream. The server has a storage capacity of not less than 10 TB, which is sufficient to record long-term production data. Based on the above data storage scheme, the whole process parameters of dyeing process are continuously collected to obtain a complete dyeing process original data set. The data set includes the whole process parameter records from the beginning of the dyeing process to the end of the water washing process, including emulsifier addition, dye addition, temperature rising, temperature keeping, temperature falling, and so on. In the process of silk cotton blended fabric dyeing, tens of thousands of data points are generated in a dyeing cycle, covering the temperature curve, pH value change, dye concentration change and other key information of each stage of the process.

[0041] In a specific embodiment, the process of performing step S102 can specifically include the following steps:

[0042] Data cleaning is performed on the dyeing process original data set to obtain a complete integrity cleaned data set;

[0043] The complete integrity cleaned data set is standardized by nonlinear transformation to obtain balanced standard data;

[0044] The balanced standard data is subjected to time series noise reduction processing by applying a filter to obtain smoothed data;

[0045] Wavelet coefficients and statistical features are extracted from the smoothed data to obtain a multi-dimensional feature matrix;

[0046] Dimensionality reduction technique is performed on the multi-dimensional feature matrix to obtain a low-dimensional feature set;

[0047] The low-dimensional feature set is fused with the dyeing process mechanism features to obtain a dyeing process feature data set.

[0048] Specifically, data cleaning is performed on the original dyeing process dataset to handle missing values, outliers, and duplicate data. In the dyeing process, data missing or outliers often occur due to sensor failure or signal interference. For missing values, different completion methods are used according to the data type: for continuous parameters such as temperature, linear interpolation is used to fill in short-term missing data; for parameters such as pH value with large fluctuations, k-nearest neighbor interpolation method is used to fill in the data under similar working conditions; for parameters in stable state, forward padding method is used to replace the nearest valid value. For outliers, the 3σ rule is used to identify data points outside the normal fluctuation range, for example, if the temperature suddenly jumps from 80°C to 95°C, it will be marked as an outlier and corrected according to the process curve characteristics. For duplicate data, the first valid value is retained and the subsequent duplicate records are removed to avoid data redundancy affecting subsequent analysis.

[0049] The integrity cleaned dataset is standardized by nonlinear transformation to obtain balanced standard data. Since the dyeing process involves multiple parameters of different dimensions, such as temperature (°C), time (min), pH value, concentration (g / L), etc., standardization is needed to make each parameter comparable. For different parameter characteristics, appropriate nonlinear transformation methods are selected: for approximately normally distributed parameters, z-score standardization is used; for skewed distribution parameters such as dye concentration, log transformation is used before standardization; for parameters with clear upper and lower limits such as pH value, Min-Max normalization is used to map to the [0,1] interval. This standardization process makes parameters of different physical quantities on the same scale, facilitating subsequent model learning of the relative importance between parameters. The balanced standard data is processed by a filter for time series noise reduction to obtain smoothed data. The time series of dyeing process parameters often contains high-frequency noise, affecting data quality. Savitzky-Golay filter is applied for time series data smoothing, with window width set to 15 data points and polynomial order set to 3, effectively removing noise while preserving the original trend of the data. For example, during the process of dyeing temperature rising from 40°C to 80°C, the original data may have ±0.5°C high-frequency fluctuations, after filtering, a smoother temperature rising curve is obtained, while the key inflection point information such as the heating rate change is preserved.

[0050] The wavelet coefficients and statistical features are extracted from the smoothed data to obtain a multi-dimensional feature matrix. The wavelet transform is used to capture the features of the data at different time scales, and to identify the key patterns in the process. At the same time, statistical features including mean, standard deviation, peak value, valley value, slope, skewness and kurtosis are extracted to describe the distribution and variation characteristics of the parameters. For example, the mean and standard deviation of the temperature curve during the holding stage can quantify the stability of temperature control; the slope feature of the temperature curve during the heating stage can represent the control accuracy of the heating rate; the peak and valley values of the pH curve can reflect the key change points in the chemical reaction process. In addition, the frequency domain features including power spectral density, main frequency component and frequency band energy distribution are obtained by fast Fourier transform to reveal the periodic variation of the data.

[0051] Dimensionality reduction techniques are performed on the multi-dimensional feature matrix to obtain a low-dimensional feature set. The original features extracted during the dyeing process have high dimensionality, redundancy and noise, and need to be processed by dimensionality reduction. Principal component analysis (PCA) or nonlinear dimensionality reduction techniques such as t-SNE are applied to map high-dimensional features to low-dimensional space, preserving the data structure while reducing computational complexity. For example, for a complete dyeing process, the original extracted features may have more than 100 dimensions, which are compressed to 20-30 main features by dimensionality reduction techniques, while retaining more than 90% of the information content, effectively reducing the complexity of subsequent modeling.

[0052] The low-dimensional feature set is fused with the mechanism features of the dyeing process to obtain a dyeing process feature dataset. Although data-driven features can capture statistical patterns, mechanism features extracted in combination with process expertise can more comprehensively represent the dyeing process. Mechanism features include temperature rise rate, holding stability index, cooling rate, pH value stability time, dye adsorption rate, and other indicators that directly reflect process performance. For example, for silk cotton blended fabric dyeing, the hydrogen peroxide activity half-life feature is extracted from the scouring and dyeing bath stage; the dispersion stability index is extracted from the emulsifier stage; the deviation integral value of the actual temperature curve from the ideal curve is extracted from the heating stage; and the root mean square value of temperature fluctuation is extracted from the holding stage. These mechanism features are integrated with the aforementioned low-dimensional feature set to form a dyeing process feature dataset containing more than 30 feature dimensions, providing a comprehensive feature representation for subsequent deep learning model training.

[0053] In a specific embodiment, the process of performing step S103 can specifically include the following steps:

[0054] The dyeing process feature dataset is divided into a training set, a validation set and a test set according to a ratio of 7:2:1;

[0055] The multi-structure deep learning model including a multi-layer perception, a convolutional neural network and a recurrent neural network is constructed based on the training set to obtain a dyeing quality prediction base model. The multi-layer perception includes four hidden layers. The convolutional neural network includes three convolutional blocks. The recurrent neural network adopts a bidirectional structure.

[0056] The dyeing quality prediction base model is optimized through the validation set to obtain a dyeing quality prediction candidate model.

[0057] The dyeing quality prediction candidate model is evaluated and verified by using the test set to obtain a deep learning prediction model for predicting color fastness, color difference, level dyeing and color variation rate.

[0058] The multi-objective optimization variables and constraint conditions are set based on the deep learning prediction model to obtain dyeing process parameter optimization variables. The dyeing process parameter optimization variables include dye bath temperature, dye ratio and holding time.

[0059] The non-dominated sorting genetic algorithm is used to calculate the dyeing process parameter optimization variables to obtain a Pareto optimal solution set. The compromise solution is screened from the Pareto optimal solution set by using the weighted TOPSIS method to obtain the optimal dyeing process parameter combination.

[0060] Specifically, the dyeing process feature data set is divided into a training set, a validation set and a test set according to a ratio of 7:2:1. This division ratio ensures that there are sufficient data samples for model training, and at the same time, a certain amount of data is reserved for model verification and final performance evaluation. For dyeing data, the balance of data distribution needs to be considered to ensure that different types of dyeing process conditions and results are reasonably distributed in the three data sets, avoiding training bias. The multi-structure deep learning model including a multi-layer perception, a convolutional neural network and a recurrent neural network is constructed based on the training set to obtain a dyeing quality prediction base model. The multi-layer perception part includes four hidden layers with neuron numbers of 128, 256, 128 and 64, respectively. The activation function adopts ReLU, mainly processing non-time sequence features. The convolutional neural network part includes three convolutional blocks, each of which includes two convolutional layers and one maximum pooling layer. The convolution kernel size is 3×1, the initial channel number is 32, and the channel number doubles after each convolutional block, mainly used for extracting local feature patterns of dyeing parameters. The recurrent neural network part adopts a bidirectional structure, including two layers of hidden state dimension of 128 gate recurrent units, mainly capturing long time sequence dependence of the dyeing process. This multi-structure design fully utilizes the advantages of different neural network architectures, which can simultaneously process static features and dynamic time sequence features in the dyeing process.

[0061] The dyeing quality prediction base model is optimized by the validation set to obtain a dyeing quality prediction candidate model. In the optimization process, the Adam optimizer is used, the initial learning rate is set to 0.001, the cosine annealing scheduling is adopted, the learning rate is reduced to 0.5 of the original every 200 epochs, and the total number of training epochs is 1000. The early stopping strategy is adopted, and when the validation set loss does not improve for 50 consecutive epochs, the training is stopped. The loss function is designed as the weighted sum of mean square error and mean absolute error, and the weight ratio is 7:3 to balance the punishment degree of large and small errors. By adjusting the network structure, optimizing the hyperparameters, and applying regularization techniques such as Dropout, the performance of the model on the validation set is continuously improved, and finally the dyeing quality prediction candidate model is obtained.

[0062] The dyeing quality prediction candidate model is evaluated and verified by the test set to obtain a deep learning prediction model for predicting color fastness, color difference value, level dyeing and color variation rate. The test set is a data set completely independent of the training process, which is used to objectively evaluate the generalization ability of the model. The evaluation indicators include mean absolute error, root mean square error and R-square value. For color fastness prediction, the average absolute error of the model does not exceed 0.5 level; for color difference value prediction, the error does not exceed 0.3ΔE; for level dyeing prediction, the accuracy is not less than 92%; for color variation rate prediction, the error does not exceed 2%. The model output results include the predicted values of the four quality indicators and their 95% confidence intervals, providing a reliability estimate of the prediction results.

[0063] The multi-objective optimization variables and constraint conditions are set based on the deep learning prediction model to obtain the dyeing process parameter optimization variables. The optimization variables include 12 key process parameters such as dye bath temperature, dye ratio and holding time, and each variable is set with a reasonable value range: the dye bath temperature ranges from 30 to 90 DEG C; the dye ratio is adjusted by ± 15% according to the target formula; and the holding time ranges from 20 to 40 minutes. The constraint conditions include the physical limitations of the process parameters and the operation limitations of the dyeing equipment to ensure that the optimization results are feasible in the actual process. The non-dominated sorting genetic algorithm is used to calculate the dyeing process parameter optimization variables to obtain the Pareto optimal solution set. The non-dominated sorting genetic algorithm is an effective method for solving multi-objective optimization problems, which can simultaneously optimize color fastness, color difference, level dyeing, color mottle rate, dyeing time and energy consumption and other multiple objectives. The algorithm uses real number coding, the population size is set to 100, and the evolution number is set to 500. The tournament selection method is selected for the selection operation, and the tournament size is 3; the simulated binary crossover is used for the crossover operation, the crossover probability is set to 0.9, and the distribution index is set to 20; the polynomial mutation is used for the mutation operation, and the mutation probability is set to 1 / 12. The algorithm also introduces an adaptive population mechanism: when the number of Pareto frontier individuals exceeds 70% of the total population, the population size is increased by 25%; when the number of frontier individuals is less than 30%, the population size is reduced by 15%, but the minimum population size is not less than 60. After the algorithm converges, a series of non-dominated solutions are obtained to form the Pareto optimal solution set. Finally, the best compromise solution is selected from the Pareto optimal solution set by the weighted TOPSIS method as the optimal dyeing process parameter combination. The TOPSIS method comprehensively considers the weight proportion of the six optimization objectives, and the weight proportions of color fastness, color difference, level dyeing, color mottle rate, dyeing time and energy consumption are set to 3:2:2:1:1:1 to find the optimal parameter combination in the multi-objective trade-off.

[0064] In a specific embodiment, the process of performing the step of constructing a multi-structure deep learning model including a multi-layer perception, a convolutional neural network, and a recurrent neural network based on the training set can specifically include the following steps:

[0065] Gaussian noise and random transformation are added to the training set to obtain an expanded training data set;

[0066] The expanded training data set is calculated by a 4-layer fully connected network for forward propagation to obtain a multi-layer perception, and the number of hidden layer neurons of the 4 hidden layers is 128, 256, 128 and 64 respectively;

[0067] The expanded training data set is extracted by 3 convolutional blocks to obtain a convolutional neural network, and the convolution kernel size of the 3 convolutional blocks is 3x1, the initial channel number is 32, and the channel number is doubled after each convolutional block;

[0068] The extended training data set is sequentially modeled by a bidirectional gated recurrent unit to obtain a recurrent neural network. The bidirectional gated recurrent unit includes 2 layers, and the hidden state dimension is 128. An attention mechanism is added to the output layer.

[0069] The multi-layer perceptron, convolutional neural network and recurrent neural network are trained by using an Adam optimizer and a cosine annealing learning rate schedule to obtain a neural network model.

[0070] The neural network model is stacked and generalized to obtain a dyeing quality prediction base model.

[0071] Specifically, the training set is data enhanced by adding Gaussian noise and random transformation to obtain an extended training data set. For dyeing process data, it is particularly important. In specific implementation, Gaussian noise with a mean of 0 and a standard deviation of 0.5°C is added to the temperature parameter to simulate sensor measurement error; random time window slicing is performed on the time series data to simulate dyeing processes of different lengths; ±3% random fluctuations are added to the dye concentration parameter to simulate minor errors in the batching process. These enhancement methods enable the model to adapt to various fluctuations in actual production and improve the robustness of the model.

[0072] The extended training data set is calculated by a 4-layer fully connected network to obtain a multi-layer perceptron. The number of neurons in the 4 hidden layers of the multi-layer perceptron is set to 128, 256, 128 and 64, respectively, showing a structure of first expansion and then contraction. The first layer of 128 neurons is responsible for preliminary feature extraction; the second layer is expanded to 256 neurons to increase the network expression capability and capture more complex feature combinations; the third and fourth layers are gradually reduced to 128 and 64 neurons to compress and abstract features. The activation function uses ReLU to solve the gradient disappearance problem while maintaining computational efficiency. To prevent overfitting, a Dropout layer is added after each layer with a dropout rate of 0.3. The multi-layer perceptron mainly processes non-time sequence features in the dyeing process, such as the association between static parameters such as dye type and fabric density and dyeing quality.

[0073] The extended training data set is feature extracted by 3 convolutional blocks to obtain a convolutional neural network. Each convolutional block includes 2 convolutional layers and 1 max pooling layer, and the convolution kernel size is set to 3x1, which is suitable for capturing local patterns of time series of dyeing parameters. The initial number of channels is set to 32, and the number of channels is doubled after each convolutional block, i.e., the second convolutional block outputs 64 channels, and the third convolutional block outputs 128 channels, gradually extracting higher-level feature representations. The pooling size of the max pooling layer is 2x1, which reduces the dimension of the feature map while preserving important features. The convolutional neural network is particularly suitable for recognizing local time sequence patterns such as temperature curve shape features and heating and cooling rate change rules in the dyeing process, which are directly related to dyeing uniformity and color fastness.

[0074] The extended training data set is modeled by a bidirectional gated recurrent unit to obtain a recurrent neural network. The recurrent network adopts a bidirectional structure, including 2 layers of gated recurrent units, and the hidden state dimension is set to 128. The bidirectional structure enables the network to simultaneously consider past and future information, more comprehensively understanding the long-term dependence relationship in the time series data. The output layer adds an attention mechanism to automatically identify and focus on key time points in the dyeing process, such as temperature mutations, pH stability, and other key moments that significantly affect dyeing quality. The recurrent neural network is particularly suitable for capturing dynamic change characteristics of the entire dyeing process, such as dye adsorption process, cumulative effects of temperature fluctuations on color fastness, and other long-term time series dependence relationships.

[0075] The multi-layer perceptron, convolutional neural network, and recurrent neural network are trained using the Adam optimizer and cosine annealing learning rate scheduling to obtain the neural network model. The Adam optimizer combines the advantages of the momentum method and the adaptive learning rate method, with an initial learning rate of 0.001. The cosine annealing learning rate scheduling reduces the learning rate to 0.5 of the original every 200 epochs, effectively alleviating oscillation during the learning process and helping the model find a better local minimum. During training, a mini-batch gradient descent with a batch size of 64 is used, along with an early stopping strategy that stops training when the validation set loss does not improve for 50 consecutive epochs. The loss function uses a weighted combination of mean squared error and mean absolute error with a weight ratio of 7:3 to balance the sensitivity to large and small errors. The trained neural network model is integrated through stacking generalization technology to obtain a dyeing quality prediction base model. Stacking generalization is a powerful model integration method that uses the outputs of multiple base models as new features to train a meta-learner. In specific implementation, the output features of the multi-layer perceptron, convolutional neural network, and recurrent neural network are spliced as input features for the XGBoost model. The XGBoost model parameters include a maximum tree depth of 5, a learning rate of 0.05, a subsampling rate of 0.8, and an iteration number of 500. This integration method fully leverages the complementary advantages of each type of neural network: multi-layer perceptron is good at processing static feature relationships, convolutional network is good at capturing local time series patterns, and recurrent network is good at modeling long-term dependence relationships, resulting in more accurate dyeing quality prediction.

[0076] In a specific embodiment, the process of performing step S104 can specifically include the following steps:

[0077] According to the optimal dyeing process parameter combination, process data mapping conversion is performed to obtain a standardized control parameter matrix;

[0078] The standardized control parameter matrix is subjected to hierarchical analysis processing to obtain a field execution layer parameter set, a process control layer parameter set, and an optimization decision layer parameter set;

[0079] The process execution control strategy is constructed based on the field execution layer parameter set, the process control layer parameter set and the optimization decision layer parameter set, and a dynamic regulation and control execution scheme is obtained.

[0080] The dynamic regulation and control execution scheme is applied to the dyeing process flow, and through real-time parameter monitoring and feedback adjustment, an intelligent dyeing scheme is obtained.

[0081] Specifically, it is converted into an actually applicable intelligent dyeing scheme. First, the process data is mapped and converted according to the optimal dyeing process parameter combination to obtain a standardized control parameter matrix. In this step, the parameters are first sorted according to the process flow sequence to form a process parameter sequence, including the related parameters of the scouring and dyeing bath, emulsified dye, temperature rise, temperature maintenance, temperature drop and deamination ester process. Then, the process parameter sequence is subjected to dimension normalization processing to unify the parameters of different physical quantities to the 0-1 interval to form a unified dimension process parameter matrix. Then, a correlation diagram between process variables is constructed based on the unified dimension process parameter matrix, the Pearson correlation coefficient is used to quantify the dependence relationship between parameters, and the key parameter combination that affects each other is identified. Cluster analysis is performed on the parameter correlation network to identify highly coupled parameter groups, and joint optimization conversion is performed through principal component analysis method to convert the related parameters into orthogonal representation to obtain a decoupled parameter matrix. The decoupled parameter matrix is restructured according to the control level requirements to form a standardized control parameter matrix.

[0082] The hierarchical analysis processing is performed on the standardized control parameter matrix to obtain the field execution layer parameter set, the process control layer parameter set and the optimization decision layer parameter set. The hierarchical analysis processing is to decompose the control parameters according to the control granularity and the time scale to form a hierarchical control architecture. The field execution layer parameter set contains specific parameters for directly controlling the dyeing equipment, such as the frequency setting value of the frequency speed driver, the opening percentage of the proportional regulating valve, the power output value of the heater power controller, etc. These parameters need to be updated at a high frequency, usually on a second scale. The process control layer parameter set contains intermediate layer control parameters such as temperature control loop, pH value control loop, dye addition amount control loop, etc., which are responsible for realizing accurate control of process parameters. For the temperature control loop, a feedforward-feedback composite control strategy is adopted, and the control accuracy reaches ±0.2℃; for the pH value control loop, a fuzzy self-adaptive PID control algorithm is adopted, and the control accuracy reaches ±0.05. These parameters are optimized and adjusted on a minute scale. The optimization decision layer parameter set contains higher level process parameter optimization targets and constraint conditions, such as color fastness optimization target, uniformity requirement, energy consumption control target, etc., which are updated on a dyeing batch level. Based on the field execution layer parameter set, the process control layer parameter set and the optimization decision layer parameter set, the process execution control strategy is constructed to obtain a dynamic regulation and control execution scheme. The process execution control strategy adopts a model predictive control based architecture, uses a deep learning prediction model to predict the future process state, and calculates the optimal control trajectory. The control strategy is based on the rolling horizon optimization principle, the prediction window length is set to 120 seconds, the control window length is set to 30 seconds, and the sampling time is set to 3 seconds. In each optimization period, the system predicts the process parameter change trend for 120 seconds in the future based on the current state, calculates the optimal control action for 30 seconds in the future, and only executes the control action of the first sampling period. The dynamic regulation and control execution scheme also contains an automatic fault detection and processing mechanism. When a sensor failure, an actuator failure or a control loop anomaly is detected, the system will automatically switch to a safety control mode to ensure the reliability of the dyeing process.

[0083] The dynamic control execution scheme is applied to the dyeing process, and an intelligent dyeing scheme is obtained through real-time parameter monitoring and feedback adjustment. In actual application, first, the initial setting value of the optimal dyeing process parameter is loaded, and the dyeing process is started; during the dyeing process, the system continuously monitors the actual process parameters and compares them with the predicted values and the set values; when the deviation between the actual parameters and the predicted values exceeds the preset threshold, real-time optimization calculation is triggered, and the control strategy is dynamically adjusted. The intelligent dyeing scheme has a self-learning function, which compares the difference between the theoretical model predicted value and the actual dyeing effect, continuously updates and optimizes the parameters of the deep learning prediction model, and realizes the continuous improvement of the control performance. In the dyeing process of silk cotton blended fabric, the intelligent dyeing scheme can automatically adjust the key parameters such as the heating rate, holding time, and dye addition amount according to the characteristics of different batches of fabric, and make fine adjustments according to the real-time monitoring of the dyeing state, to ensure the consistency and stability of the dyeing quality, and optimize the energy consumption and dyeing time.

[0084] In a specific embodiment, the process of performing the process data mapping and conversion step according to the optimal dyeing process parameter combination can specifically include the following steps:

[0085] Dimensional normalization is performed on the process parameter sequence to obtain a process parameter matrix;

[0086] A correlation graph between process variables is constructed based on the process parameter matrix to obtain a parameter correlation network;

[0087] Cluster analysis is performed on the parameter correlation network to obtain a highly coupled parameter group;

[0088] Joint optimization and conversion are performed according to the highly coupled parameter group to obtain a decoupled parameter matrix;

[0089] The decoupled parameter matrix is structurally reorganized according to the control level requirements to obtain a standardized control parameter matrix. The structural reorganization assigns the parameters to the corresponding control levels and sets the priority order.

[0090] Specifically, the process parameter sequence is dimensionally normalized to obtain a process parameter matrix. In specific operations, appropriate normalization methods are adopted for parameters of different physical quantities: for temperature parameters, Min-Max normalization is used to map actual temperature values of 40-95°C to the 0-1 interval; for time parameters, the holding time of 0-60 minutes is linearly mapped to the 0-1 interval; for concentration parameters, the concentration values of various dyes and auxiliaries are divided by their maximum allowable concentration values; for pH value parameters, the actual values of 4.5-10.5 are linearly mapped to the 0-1 interval. This unified dimensional treatment eliminates the scale difference between different physical quantities. Based on the process parameter matrix, a correlation graph between process variables is constructed to obtain a parameter correlation network. By calculating the Pearson correlation coefficient between parameters, the dependency relationship between different process parameters is quantified. For example, in the dyeing process of silk cotton blended fabric, the hydrogen peroxide concentration in the scouring stage has a positive correlation of 0.72 with the addition amount of cotton dissolving agent in the emulsification stage; the temperature rising rate has a negative correlation of -0.58 with the temperature stability in the holding stage. In this way, the correlation between all process parameters is organized into a network structure, clearly showing the mutual influence relationship between parameters, and providing a basis for subsequent parameter optimization.

[0091] The parameter correlation network is subjected to cluster analysis to obtain a highly coupled parameter group. Spectral clustering algorithm is used to analyze the parameter correlation network, and parameters with an absolute correlation coefficient greater than 0.5 are classified into the same group to identify a parameter subset with strong correlation. In the dyeing process, several obvious parameter groups are usually formed: a temperature control parameter group (including temperature rising rate, holding temperature, temperature falling rate, etc.), a chemical dosage parameter group (including dye concentration, auxiliary addition amount, pH value, etc.), and a time control parameter group (including the duration of each process). According to the highly coupled parameter group, joint optimization conversion is carried out to obtain a decoupled parameter matrix. For each highly coupled parameter group, principal component analysis method is applied to convert the related parameters into orthogonal representation, reducing the redundancy and interference between parameters. For example, principal component analysis is performed on the temperature control parameter group to obtain the first principal component representing the overall level of temperature and the second principal component representing the temperature change rate; the chemical dosage parameter group is converted to separate orthogonal components representing the basic concentration of dyes, the proportion of dyes, and the auxiliary matching relationship. This joint optimization conversion reduces the coupling degree between parameters, making the parameter adjustment in the subsequent control process more independent and accurate.

[0092] The decoupled parameter matrix is restructured according to the control level requirements to obtain a standardized control parameter matrix. In the restructuring process, the parameters are allocated to different control levels according to the control characteristics and time scales of the parameters: fast-responding execution parameters (such as valve opening, heating power) are allocated to the field execution layer; medium-response-time process parameters (such as temperature set value, pH control target) are allocated to the process control layer; long-period optimization strategy parameters (such as dyeing formula optimization, energy consumption control strategy) are allocated to the optimization decision layer. At the same time, within each control level, the priority order is set according to the degree of influence of the parameters on the dyeing quality, and parameters that significantly affect the dyeing uniformity are given higher priority. Through this hierarchical parameter organization, a standardized control parameter matrix with clear structure and reasonable control granularity is formed.

[0093] The above describes the dyeing process optimization method based on intelligent control in the embodiments of the present application. The dyeing process optimization system based on intelligent control in the embodiments of the present application is described below. Please refer to Figure 2 The dyeing process optimization system based on intelligent control in the embodiments of the present application includes one embodiment:

[0094] The acquisition module 201 is configured to acquire the key parameters of the dyeing process to obtain a dyeing process original data set.

[0095] The extraction module 202 is configured to perform preprocessing and feature extraction based on the dyeing process original data set to obtain a dyeing process feature data set.

[0096] The optimization module 203 is configured to construct a deep learning prediction model using the dyeing process feature data set and perform multi-objective optimization to obtain an optimal dyeing process parameter combination. The multi-objective optimization adopts a non-dominated sorting genetic algorithm.

[0097] The analysis module 204 is configured to analyze the dyeing scheme according to the optimal dyeing process parameters to obtain an intelligent dyeing scheme.

[0098] Through the cooperation of the above various components, the dyeing process original data set is obtained by collecting the key parameters of the dyeing process, the overall capture and digital expression of the dyeing process data are realized, and the data foundation for subsequent intelligent analysis is laid; the dyeing process feature data set is obtained based on the preprocessing and feature extraction of the dyeing process original data set, effectively solving the problems of large data noise and insufficient feature extraction in the traditional dyeing process, and significantly improving the data quality and feature expression ability; the deep learning prediction model is constructed by using the dyeing process feature data set and multi-objective optimization is performed, which not only overcomes the defects of limited expression ability of traditional single model, but also realizes the accurate prediction of dyeing quality through the complementary advantages of multi-layer perception processing non-time sequence features, convolutional neural network capturing local time sequence mode and recurrent neural network modeling long-term dependence relationship, and the non-dominated sorting genetic algorithm used in multi-objective optimization can optimize multiple targets such as color fastness, color difference, uniformity, color flower rate, dyeing time and energy consumption, and find the best compromise solution; the dyeing control system of the hierarchical control architecture is realized according to the optimal dyeing process parameter combination, the intelligent dyeing scheme capable of automatically adjusting the process parameters is obtained, and the hierarchical control architecture includes a field execution layer, a process control layer and an optimization decision layer, so that the control granularity is more fine, the response speed is more matched with the control demand, and the dependence on manual experience of the traditional dyeing process is greatly reduced. In particular, in the specific application field of dyeing process, the present application provides a special solution for the complex nonlinear characteristics, multi-parameter coupling problems and multi-objective optimization requirements of the dyeing process by fusing various deep learning algorithms and non-dominated sorting genetic algorithm features. The core contribution of the algorithm features to the scheme is as follows: the multi-structure fusion characteristics of the deep learning algorithm accurately capture the complex relationship between the static features and the dynamic time sequence features in the dyeing process; the multi-objective balance characteristics of the non-dominated sorting genetic algorithm effectively solve the trade-off problem between dyeing quality and production efficiency; and the hierarchical characteristics of the hierarchical control architecture perfectly adapt to the requirements of different control granularity and time scale in the dyeing process, and realize the comprehensive optimization effect of stable dyeing quality, reduced energy consumption and improved production efficiency.

[0099] The above Figure 2 The intelligent control-based dyeing process optimization system in the embodiment of the present application is described in detail from the perspective of modular functional entities, and the intelligent control-based dyeing process optimization device in the embodiment of the present application is described in detail from the perspective of hardware processing.

[0100] Figure 3is a structural schematic view of a dyeing process optimization device based on intelligent control provided by an embodiment of the present application. The dyeing process optimization device based on intelligent control 300 can have great differences due to different configurations or performances, and can include one or more processors (central processing units, CPU) 310 (for example, one or more processors) and a memory 320, one or more storage media 330 (for example, one or more mass storage device ends) storing application programs 333 or data 332. Among them, the memory 320 and the storage medium 330 can be temporary storage or persistent storage. The programs stored in the storage medium 330 can include one or more modules (not shown in the figure), each module can include a series of instruction operations in the dyeing process optimization device based on intelligent control 300. Further, the processor 310 can be configured to communicate with the storage medium 330, execute a series of instruction operations in the storage medium 330 on the dyeing process optimization device based on intelligent control 300, so as to realize the steps of the above-mentioned dyeing process optimization method based on intelligent control.

[0101] The dyeing process optimization device based on intelligent control 300 can also include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input / output interfaces 360, and / or one or more operating systems 331, such as Windows Serve, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art can understand that, Figure 3 The structure of the dyeing process optimization device based on intelligent control shown does not constitute a limitation on the dyeing process optimization device based on intelligent control provided by the present application, and can include more or fewer components than shown, or combine certain components, or different component arrangements.

[0102] The present application also provides a computer readable storage medium, which can be a non-volatile computer readable storage medium, and can also be a volatile computer readable storage medium. The computer readable storage medium has instructions stored therein, and when the instructions are run on a computer, the computer executes the steps of the dyeing process optimization method based on intelligent control.

[0103] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system, system and unit can refer to the corresponding process in the foregoing method embodiments, which will not be described here.

[0104] The integrated unit, if implemented in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application or the entire or part of the technical solutions that essentially contribute to the prior art can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a smart control-based dyeing process optimization device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0105] The above embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for optimizing dyeing process based on intelligent control, characterized in that, The method comprises: Collecting key parameters of a dyeing process to obtain a dyeing process original data set; Performing preprocessing and feature extraction based on the dyeing process original data set to obtain a dyeing process feature data set; Utilizing the dyeing process feature data set to construct a deep learning prediction model and performing multi-objective optimization to obtain an optimal dyeing process parameter combination, the multi-objective optimization adopting a non-dominated sorting genetic algorithm, comprising: dividing the dyeing process feature data set into a training set, a validation set and a test set according to a ratio of 7:2:1; constructing a multi-structure deep learning model comprising a multi-layer perceptron, a convolutional neural network and a recurrent neural network based on the training set to obtain a dyeing quality prediction base model, the multi-layer perceptron comprising four hidden layers, the convolutional neural network comprising three convolutional blocks, and the recurrent neural network adopting a bidirectional structure; performing model optimization on the dyeing quality prediction base model through the validation set to obtain a dyeing quality prediction candidate model; performing performance evaluation and verification on the dyeing quality prediction candidate model by using the test set to obtain a deep learning prediction model for predicting color fastness, color difference value, level dyeing and color mottle rate; setting multi-objective optimization variables and constraint conditions based on the deep learning prediction model to obtain dyeing process parameter optimization variables, the dyeing process parameter optimization variables comprising dye bath temperature, dye ratio and holding time; performing non-dominated sorting genetic algorithm calculation on the dyeing process parameter optimization variables to obtain a Pareto optimal solution set, and screening a compromise solution from the Pareto optimal solution set by using a weighted TOPSIS method to obtain the optimal dyeing process parameter combination; Performing dyeing scheme analysis according to the optimal dyeing process parameters to obtain an intelligent dyeing scheme, comprising: performing process data mapping conversion according to the optimal dyeing process parameter combination to obtain a standardized control parameter matrix; performing hierarchical analysis processing on the standardized control parameter matrix to obtain a field execution layer parameter set, a process control layer parameter set and an optimization decision layer parameter set; constructing a process execution control strategy based on the field execution layer parameter set, the process control layer parameter set and the optimization decision layer parameter set to obtain a dynamic regulation and control execution scheme; applying the dynamic regulation and control execution scheme to a dyeing process flow to obtain an intelligent dyeing scheme through real-time parameter monitoring and feedback adjustment.

2. The smart control based dyeing process optimization method according to claim 1, wherein, The collecting key parameters of a dyeing process to obtain a dyeing process original data set comprises: Setting multiple sensor collection points for dye concentration, dye liquor temperature, dyeing time, liquor ratio, pH value, additive amount and dyeing machine rotating speed to obtain a multi-dimensional parameter collection network; Connecting an industrial programmable logic controller to the multi-dimensional parameter collection network to obtain a field control unit, the programmable logic controller being equipped with a processor with a frequency of not less than 100 MHz; Utilizing the field control unit to perform data filtering algorithm processing on the collected data to obtain parameter data; Transmitting the parameter data to a central data server through an industrial Ethernet to obtain a parameter data stream; Setting a dual-computer hot backup mechanism for the parameter data stream to obtain a data storage scheme; Based on the data storage scheme, the dyeing process parameters are continuously collected to obtain a dyeing process original data set, which contains parameter records of all processes from the scouring and dyeing bath to the water washing and rolling of cotton.

3. The smart control based dyeing process optimization method according to claim 1, wherein, The dyeing process original data set is preprocessed and feature extracted to obtain a dyeing process feature data set, including: The dyeing process original data set is data cleaned to obtain a completeness cleaned data set; The completeness cleaned data set is standardized by nonlinear transformation to obtain balanced standard data; The balanced standard data is time series denoised by applying a filter to obtain smoothed data; Wavelet coefficients and statistical features are extracted from the smoothed data to obtain a multi-dimensional feature matrix; The multi-dimensional feature matrix is dimension reduced to obtain a low-dimensional feature set; The low-dimensional feature set is fused with dyeing process mechanism features to obtain the dyeing process feature data set.

4. The smart control based dyeing process optimization method according to claim 1, wherein, A multi-structure deep learning model containing a multi-layer perceptron, a convolutional neural network and a recurrent neural network is constructed based on the training set to obtain a dyeing quality prediction base model, the multi-layer perceptron contains 4 hidden layers, the convolutional neural network contains 3 convolutional blocks, and the recurrent neural network adopts a bidirectional structure, including: Gaussian noise and random transformation are added to the training set to obtain an expanded training data set; The expanded training data set is calculated by a 4-layer fully connected network for forward propagation to obtain the multi-layer perceptron, the number of hidden layer neurons of the 4 hidden layers is 128, 256, 128 and 64 respectively; The expanded training data set is feature extracted by 3 convolutional blocks to obtain the convolutional neural network, the convolution kernel size of the 3 convolutional blocks is 3×1, the initial channel number is 32, and the channel number doubles after each convolutional block; The expanded training data set is sequence modeled by a bidirectional gated recurrent unit to obtain the recurrent neural network, the bidirectional gated recurrent unit contains 2 layers, the hidden state dimension is 128, and the output layer adds an attention mechanism; The multi-layer perceptron, the convolutional neural network and the recurrent neural network are trained by using an Adam optimizer and a cosine annealing learning rate schedule to obtain a neural network model; The neural network model is stacked and generalized to obtain the dyeing quality prediction base model.

5. The smart control based dyeing process optimization method according to claim 1, wherein, The optimal dyeing process parameter combination is mapped and converted to obtain a standardized control parameter matrix, including: The optimal dyeing process parameter combination is sorted according to the process flow sequence to obtain a process parameter sequence, the process parameter sequence includes scouring and dyeing bath parameters, emulsified dye parameters, temperature rising parameters, temperature holding parameters, temperature falling parameters and deamination and esterization process parameters; The process parameter sequence is dimension normalized to obtain a process parameter matrix; A parameter correlation network is constructed based on the process parameter matrix to obtain a parameter correlation network; The parameter correlation network is cluster analyzed to obtain a highly coupled parameter group; The highly coupled parameter group is jointly optimized and converted to obtain a decoupled parameter matrix; The decoupling parameter matrix is restructured according to the control level requirements to obtain a standardized control parameter matrix, and the restructuring assigns the parameters to corresponding control levels and sets a priority order.

6. A smart control based dyeing process optimization system characterized in that, The intelligent control-based dyeing process optimization system comprises: a collection module configured to collect key parameters of a dyeing process to obtain a dyeing process original data set; an extraction module configured to perform preprocessing and feature extraction based on the dyeing process original data set to obtain a dyeing process feature data set; an optimization module configured to construct a deep learning prediction model using the dyeing process feature data set and perform multi-objective optimization to obtain an optimal dyeing process parameter combination, wherein the multi-objective optimization uses a non-dominated sorting genetic algorithm; an analysis module configured to perform dyeing scheme analysis according to the optimal dyeing process parameters to obtain an intelligent dyeing scheme.

7. A smart control based dyeing process optimization apparatus, characterized by, The computer program, when executed by the processor, causes the processor to implement the intelligent control-based dyeing process optimization method according to any one of claims 1 to 5.

8. A computer-readable storage medium having stored thereon a computer program, characterized in that The computer program, when executed by the processor, causes the processor to implement the intelligent control-based dyeing process optimization method according to any one of claims 1 to 5.

Images