A method for optimizing process parameters for producing plant functional fibers by liquid addition

Abstract

The present invention provides a method for optimizing process parameters for producing plant functional fibers by liquid addition, which belongs to the technical field of plant functional fiber production, and includes: 1) collecting historical data of process parameters and fiber performance indicators; 2) analyzing the correlation between process parameters and fiber performance, and screening out key influencing parameters; 3) establishing a multi-objective prediction model based on key parameters, and generating a simulation data set by using Latin hypercube sampling; 4) clustering and screening the simulation data to obtain the Pareto optimal process parameter combination; 5) experimentally verifying the accuracy of the prediction model, and iteratively optimizing the model; 6) finally using the response surface method for refined optimization to obtain the optimal production process parameters. The solution of the present invention solves the technical problem that the process parameters for producing plant functional fibers by liquid addition are often adjusted by relying on the experience of engineers in the prior art, and better optimization parameters cannot be obtained.

Description

Technical Field

[0001] The present invention belongs to the technical field of plant functional fiber production, and in particular, relates to a method for optimizing process parameters for producing plant functional fibers by liquid addition. Background Art

[0002] A new process proposed in recent years is the liquid addition method to produce plant-based functional fibers. The basic principle is to inject the functional liquid directly into the cellulose solution at the end of spinning, so that the functional substances can fully penetrate into the fiber, thereby effectively improving the fiber's performance. This method abandons the cumbersome pre-treatment and chemical modification steps in the traditional process, and can more flexibly and efficiently give the fiber the desired functionality.

[0003] However, the liquid addition production process involves multiple key process parameters, such as slurry formula, sanding parameters, liquid injection parameters, etc. The optimization of these parameters has an important impact on the final fiber performance; the existing technology often relies on the experience of engineers to adjust the process parameters of the liquid addition production of plant functional fibers, and cannot obtain better optimized parameters. Summary of the invention

[0004] In view of this, the present invention provides a method for optimizing process parameters for producing plant functional fibers by liquid addition, which can solve the technical problem that the prior art often relies on engineers' experience to adjust the process parameters for producing plant functional fibers by liquid addition and cannot obtain better optimized parameters.

[0005] The present invention is achieved in that:

[0006] The present invention provides a method for optimizing process parameters for producing plant functional fibers by liquid addition, comprising the following steps:

[0007] S10, collecting multiple sets of historical data of each process parameter in the process of producing plant functional fibers by liquid addition, wherein the process parameters include slurry formula, sanding parameters, liquid pump injection parameters and corresponding fiber performance indicators;

[0008] S20, using principal component analysis and partial least squares regression methods to analyze the correlation index between each process parameter and each fiber performance index, and using a preset correlation index threshold to screen out key influencing parameters;

[0009] S30, based on the selected key influencing parameters, construct a multi-objective prediction model for plant functional fiber performance, and use a support vector regression algorithm to establish a mapping relationship between the key influencing parameters and fiber performance indicators, which is recorded as a preliminary mapping;

[0010] S40, generating a large amount of simulated data using Latin hypercube sampling method according to the preliminary mapping to obtain an expanded data set;

[0011] S50, clustering each fiber performance index of the expanded data set using a K-means clustering method to obtain a plurality of clusters; screening the clusters using a preset threshold range of each fiber performance index to obtain a plurality of screened clusters;

[0012] S60, applying the multi-objective optimization algorithm to the data points in the screening clusters to obtain a set of Pareto optimal solutions as candidate process parameter sets;

[0013] S70, using the selected process parameter combination to conduct a small-scale test to obtain actual fiber performance data; comparing the experimental data with the predicted data of the multi-objective prediction model, evaluating the accuracy of the multi-objective prediction model, and iteratively optimizing the parameters of the multi-objective prediction model according to the experimental results;

[0014] S80. Based on the experimental results and the optimized multi-objective prediction model, the response surface method is used to perform local refinement optimization to obtain the final optimal process parameter combination.

[0015] On the basis of the above technical solution, the process parameter optimization method for producing plant functional fiber by liquid addition of the present invention can also be improved as follows:

[0016] The liquid addition method for producing plant functional fibers specifically comprises preparing the plant functional additive into a slurry, sand-milling the slurry, and then injecting the slurry into the fibers at the spinning end using a liquid pump.

[0017] Furthermore, the slurry formula specifically includes the ratio and content of the plant functional additive, dispersant and thickener.

[0018] Furthermore, the sanding parameters include sanding time, sanding speed, and sanding medium size.

[0019] Furthermore, the injection parameters of the liquid pump include injection pressure, injection flow rate, and injection position.

[0020] Furthermore, the fiber performance indicators include multiple or all of antibacterial properties, UV resistance, strength, elongation, durability, softness, air permeability, and thermal stability.

[0021] Furthermore, the fiber performance index is selected or weighted according to production requirements.

[0022] Furthermore, the objectives of the multi-objective prediction model include maximizing antibacterial properties, maximizing anti-ultraviolet properties, maximizing strength, maximizing elongation, maximizing antistatic properties, maximizing durability, maximizing softness, maximizing breathability, and maximizing thermal stability, which match the fiber performance indicators; the constraints of the multi-objective prediction model include the feasible range of process parameters, the minimum requirements of fiber performance indicators, the upper limit of production costs, and the limitation of environmental impact.

[0023] Furthermore, the threshold range of each fiber performance indicator refers to the acceptable range of each performance indicator, including antibacterial performance ≥90%, UV resistance UPF value ≥30, strength ≥3cN / dtex, elongation ≥20%, antistatic performance ≤1000V, durability ≥50 washes, softness ≤level 3, air permeability ≥100mm / s, thermal stability ≤5% shrinkage, and anti-wrinkle performance ≥level 3.

[0024] Furthermore, the preset correlation index threshold is 0.5-0.75.

[0025] Furthermore, the iterative optimization of the parameters of the multi-objective prediction model according to the experimental results specifically uses a gradient descent method to minimize the prediction error and update the model parameters until convergence or a preset number of iterations is reached.

[0026] Among them, the objective function of maximizing antibacterial performance is: Among them, x1 is the concentration of the antibacterial agent (%), x2 is the amount of dispersant used (g / L), and x3 is the pH value. The term reflects the accelerating growth trend of the antimicrobial agent concentration on the performance, and log(x2) reflects the decreasing marginal effect of the dispersant dosage. Indicates that there is an optimal pH value. Variable acquisition method: x1 is obtained through immersion experiments with different concentrations of antimicrobial solutions, x2 is determined through dispersion effect observation and antimicrobial performance testing, and x3 is directly measured using a pH meter and adjusted through a buffer solution.

[0027] Among them, the objective function of maximizing the anti-ultraviolet performance is: Wherein, x4 is the concentration of UV protection agent (%), x5 is the concentration of synergist (%), and x6 is the processing temperature (°C). This reflects that the growth rate of the effect of UV protection agents is gradually slowing down. It means that synergists may produce significant effects within a specific range. The influence of temperature and its rate of change are considered. Variable acquisition method: x4 and x5 are determined by immersion experiments and UV resistance tests, and x6 is obtained by setting different processing temperatures and testing the final UV resistance.

[0028] Among them, the objective function of intensity maximization is: Among them, x7 is the cellulose concentration (%), x8 is the amount of cross-linking agent used (g / kg), and x9 is the stretching ratio. It means that the strength growth rate is between linear and square. ln(1+x8) reflects that the cross-linking effect is obvious in the early stage and gentle in the later stage. This reflects the upper limit effect of the stretching multiple on the strength improvement. Variable acquisition method: x7 prepares cellulose fibers of different concentrations by wet spinning, x8 is determined by cross-linking reaction experiments, and x9 applies different stretching multiples during the spinning process and measures the strength.

[0029] Among them, the objective function of maximizing the elongation is: Among them, x 10 is the plasticizer content (%), x 11 is the heat treatment temperature (℃), x 12 is the fiber fineness (dtex). 10 The term represents the linear relationship between plasticizer and elongation, Reflects the critical effect of heat treatment temperature, This reflects the decreasing trend of the effect of fineness on elongation. Variable acquisition method: x 10 Determined by immersion test and elongation test, x 11 By setting different heat treatment temperatures and testing the elongation, x 12 It is determined by preparing fibers of different fineness and measuring the elongation.

[0030] Among them, the objective function of maximizing durability is: Among them, x 13 is the degree of crosslinking (%), x 14 is the stabilizer concentration (%), x 15 is the heat treatment time (min). 13 log(x 13 ) reflects that the effect of increasing the degree of cross-linking is significant in the early stage and slows down in the later stage. represents the saturation effect of the stabilizer concentration, Emphasize the importance of the rate of change of heat treatment time. Variable acquisition method: x 13 Determined by FTIR spectroscopy or chemical titration, x 14 Determined by immersion test and accelerated aging test, x 15 Obtained by setting different heat treatment times and conducting durability tests.

[0031] Among them, the objective function of maximizing softness (the smaller the value, the softer) is: Among them, x16 is the fiber diameter (μm), x 17 is the amount of softener (g / kg), x 18 It is the mechanical soft handling strength (N / cm). The smaller the diameter, the softer the fiber. This reflects that the growth rate of softener effect is gradually slowing down. The mechanical treatment strength increase is obvious in the early stage and slow in the later stage. Variable acquisition method: x 16 Measured by microscope or laser particle size analyzer, x 17 Determined by immersion test and softness test, x 18 It is obtained by setting different mechanical treatment intensities and testing the softness.

[0032] Among them, the objective function of maximizing air permeability is: Among them, x 19 is the fiber gap ratio (%), x 20 is the heat pressing temperature (℃), x 21 is the vibration frequency (Hz). It reflects that the air permeability increases with the increase of the gap ratio, but the growth rate is lower than the linear rate. represents the critical effect of hot pressing temperature, Reflects the periodic influence of vibration frequency. Variable acquisition method: x 19 By measuring the volume and weight of the fiber web, x 20 By setting different hot pressing temperatures and testing the air permeability, x 21 The fibers were treated using a vibration device and tested for air permeability determination.

[0033] Among them, the objective function of maximizing thermal stability (minimizing shrinkage) is: Among them, x 22 is the concentration of heat stabilizer (%), x 23 is the preheating time (min), x 24 is the cooling rate (°C / min). Reflecting the existence of an optimal concentration of heat stabilizer, log(1+x 23 ) indicates that the preheating effect is obvious in the early stage and moderate in the later stage. It shows that faster cooling rate is beneficial to thermal stability. Variable acquisition method: x 22 Through immersion experiments and thermal stability tests, x 23 By setting different preheating times and testing thermal stability, x 24 Determined by controlling the cooling rate and testing the final thermal stability.

[0034] The constraints include:

[0035] Feasible range of process parameters: x min,i ≤x i ≤x max,i ,i=1,2,…,24;

[0036] Production cost cap:

[0037] Environmental impact limits:

[0038] Minimum performance index requirements: f1(x)≥90%, f2(x)≥30, f3(x)≥3cN / dtex, f4(x)≥20%, f5(x)≥50, f6(x)≤3, f7(x)≥100mm / s, f8(x)≤5%;

[0039] Among them, x i is the process parameter, a i ,c i ,d i ,e i ,g i ,h i ,k i ,m i ,b i is the unknown coefficient, x min,i ,x max,i is the parameter range, c i is the unit cost, C max is the total cost upper limit, e i is the unit environmental impact, E max The parameter acquisition method includes: the process parameters are tested by the central composite design method; the unknown coefficients are obtained by nonlinear regression; the parameter range is determined according to the equipment limitation; the cost-related parameters are determined by market analysis; and the environmental impact parameters are determined by the life cycle assessment method.

[0040] Compared with the prior art, the beneficial effects of the process parameter optimization method for producing plant functional fibers by liquid addition provided by the present invention include:

[0041] 1. A complete data base of process parameters and fiber performance indicators has been established. By collecting historical production data, we have fully mastered process parameters such as slurry formula, sanding parameters, liquid injection parameters, as well as multiple fiber performance indicators such as antibacterial properties, UV resistance, strength, elongation, durability, softness, air permeability and thermal stability. This has laid a good foundation for subsequent analysis and modeling.

[0042] 2. The key process parameters that affect fiber performance were identified by combining principal component analysis and partial least squares regression. By analyzing the correlation between each parameter and the performance index, the most critical process parameters were effectively screened out, greatly simplifying the dimension of the optimization problem.

[0043] 3. A multi-objective prediction model for plant functional fiber performance was constructed. The model uses the support vector regression algorithm to establish the mapping relationship between key process parameters and various performance indicators, and optimizes multiple objectives such as antibacterial property, UV resistance, strength, elongation, durability, softness, air permeability and thermal stability. This provides an effective means to achieve the coordinated optimization of multiple fiber properties.

[0044] 4. A large amount of simulation data was generated by using methods such as Latin hypercube sampling to enrich the training samples. At the same time, the K-means clustering algorithm was used to analyze these data to effectively screen out the process parameter combinations that meet the requirements of various performance indicators. This data-driven method can greatly reduce the experimental workload and improve the optimization efficiency.

[0045] 5. The multi-objective optimization algorithm is used to find the Pareto optimal solution, which provides an optional optimal process parameter combination for production practice. These Pareto solutions achieve the best balance between various performance indicators and provide a scientific basis for process formulation.

[0046] 6. Through small-scale test verification and iterative optimization of model parameters, the accuracy of the multi-objective prediction model has been continuously improved. This method of combining experiments and modeling can ensure that the optimization results can truly guide production practice and lay the foundation for subsequent process expansion and commercial application.

[0047] In general, the process parameter optimization method for the liquid addition production of plant functional fibers proposed in the present invention makes full use of advanced data analysis and multi-objective optimization technologies, realizes the coordinated optimization of multiple performance indicators, and significantly improves the comprehensive performance of the fibers compared with the existing technology. It solves the technical problem that the existing technology often relies on the experience of engineers to adjust the process parameters of the liquid addition production of plant functional fibers and cannot obtain better optimized parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A flow chart of the method provided by the present invention;

[0049] Figure 2 The principal component analysis-variance contribution rate diagram in Example 2;

[0050] Figure 3 This is the clustering result diagram in Example 2. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0052] like Figure 1 FIG. 1 is a flow chart of a method for optimizing process parameters for producing plant functional fibers by liquid addition provided by the present invention. The method comprises the following steps:

[0053] S10, collecting multiple sets of historical data of each process parameter in the process of producing plant functional fibers by liquid addition, the process parameters including slurry formula, sanding parameters, liquid pump injection parameters and corresponding fiber performance indicators;

[0054] S20, using principal component analysis and partial least squares regression methods to analyze the correlation index between each process parameter and each fiber performance index, and using a preset correlation index threshold to screen out key influencing parameters;

[0055] S30, based on the selected key influencing parameters, construct a multi-objective prediction model for plant functional fiber performance, and use a support vector regression algorithm to establish a mapping relationship between the key influencing parameters and fiber performance indicators, which is recorded as a preliminary mapping;

[0056] S40, based on the preliminary mapping, a large amount of simulated data is generated using the Latin hypercube sampling method to obtain an expanded data set;

[0057] S50, clustering each fiber performance index using a K-means clustering method for the expanded data set to obtain a plurality of cluster clusters; screening the cluster clusters using a preset threshold range for each fiber performance index to obtain a plurality of screened cluster clusters;

[0058] S60, applying a multi-objective optimization algorithm to the data points in the screening clusters to obtain a set of Pareto optimal solutions as candidate process parameter sets;

[0059] S70, using the selected process parameter combination to conduct a small-scale test to obtain actual fiber performance data; comparing the experimental data with the predicted data of the multi-objective prediction model, evaluating the accuracy of the multi-objective prediction model, and iteratively optimizing the parameters of the multi-objective prediction model according to the experimental results;

[0060] S80. Based on the experimental results and the optimized multi-objective prediction model, the response surface method is used to perform local refinement optimization to obtain the final optimal process parameter combination.

[0061] The specific implementation methods of the above steps are described in detail below:

[0062] First, in step S10, historical production data needs to be collected, including slurry formula, sanding parameters, liquid pump injection parameters, and corresponding various fiber performance indicators, such as antibacterial performance, UV resistance, strength, elongation, durability, softness, air permeability, and thermal stability. The purpose of this step is to establish a complete data foundation to provide a basis for subsequent analysis and modeling.

[0063] In step S20, principal component analysis (PCA) and partial least squares regression (PLSR) methods are used to analyze the correlation between each process parameter and various fiber performance indicators. By presetting the correlation index threshold (such as 0.5-0.75), the key influencing parameters can be screened out to lay the foundation for the subsequent prediction model establishment. Principal component analysis can help identify the main variation patterns in the data, while partial least squares regression can establish a quantitative relationship between process parameters and fiber properties.

[0064] Based on the key influencing parameters screened in step S20, a multi-objective prediction model for the performance of plant functional fibers will be constructed in step S30. Specifically, the support vector regression (SVR) algorithm will be used to establish the mapping relationship between key process parameters and various fiber performance indicators. The objective function of this prediction model includes maximizing multiple indicators such as antibacterial performance, UV resistance, strength, elongation, antistatic performance, durability, softness, air permeability and thermal stability, which match the actual fiber performance requirements. At the same time, constraints such as the feasible range of process parameters, production costs and environmental impacts also need to be considered.

[0065] In step S40, the Latin Hypercube Sampling (LHS) method is used to generate a large amount of simulated data based on the initially established prediction model, thereby expanding the data set. Latin Hypercube Sampling is an effective multi-dimensional random sampling method that can ensure that the sample points in each dimension are evenly distributed and improve the effect of model training.

[0066] Next, in step S50, the expanded data set will be clustered. Specifically, the K-means clustering algorithm will be used to cluster each fiber performance index to obtain multiple clusters. Then, according to the pre-set performance index threshold range, such as antibacterial property ≥ 90%, UV resistance UPF value ≥ 30, strength ≥ 3cN / dtex, elongation ≥ 20%, etc., these clusters will be screened to obtain data points that meet the requirements.

[0067] In step S60, a multi-objective optimization algorithm, such as NSGA-II or MOEA / D, is applied to the data points in these screened clusters to obtain a set of Pareto optimal solutions as candidate optimal process parameter combinations. The multi-objective optimization algorithm can find the best balance between objective functions and give a series of feasible solutions that meet the requirements of various performance indicators.

[0068] In step S70, the candidate process parameter combination obtained in step S60 is selected to conduct a small-scale test to obtain actual fiber performance data. Next, these experimental data will be compared with the predicted data of the multi-objective prediction model established in step S30 to evaluate the accuracy of the model. If it is found that there is a significant deviation between the predicted results and the experimental results, the parameters of the multi-objective prediction model will be iteratively optimized using the gradient descent method until the model prediction error is minimized. The purpose of this step is to continuously improve and calibrate the prediction model to improve its reliability.

[0069] Finally, in step S80, based on the optimized multi-objective prediction model, the response surface method (RSM) is used to perform local refinement optimization to obtain the final optimal process parameter combination. The response surface method is an optimization method based on experimental design and statistical analysis, which can effectively find the optimal value of the objective function.

[0070] Some key formulas involved in the entire optimization process are as follows:

[0071] Objective function for maximizing antibacterial performance:

[0072]

[0073] Among them, x1 is the concentration of antimicrobial agent (%), x2 is the amount of dispersant (g / L), x3 is the pH value, and a1, a2, a3, and b1 are coefficients to be determined.

[0074] Objective function for maximizing UV resistance:

[0075]

[0076] Among them, x4 is the concentration of UV protection agent (%), x5 is the concentration of synergist (%), x6 is the processing temperature (℃), and c1, c2, c3, and b2 are unknown coefficients.

[0077] The objective function of intensity maximization is:

[0078]

[0079] Among them, x7 is the cellulose concentration (%), x8 is the amount of crosslinking agent (g / kg), x9 is the stretching multiple, and d1, d2, d3, and b3 are unknown coefficients.

[0080] Objective function for maximizing elongation:

[0081]

[0082] Among them, x 10 is the plasticizer content (%), x 11 is the heat treatment temperature (℃), x 12 is the fiber fineness (dtex). e1, e2, e3, b4 are unknown coefficients.

[0083] The objective function of maximizing durability is:

[0084]

[0085] Among them, x 13 is the degree of cross-linking (%), x 14 is the stabilizer concentration (%), x 15 is the heat treatment time (min). g1, g2, g3, b5 are unknown coefficients.

[0086] Objective function for maximizing softness (the smaller the value, the softer):

[0087]

[0088] Among them, x 16 is the fiber diameter (μm), x 17 is the amount of softener (g / kg), x 18 is the mechanical softening strength (N / cm). h1, h2, h3, b6 are unknown coefficients.

[0089] The objective function for maximizing air permeability is:

[0090]

[0091] Among them, x 19 is the fiber gap ratio (%), x 20 is the hot pressing temperature (℃), x 21 is the vibration frequency (Hz). k1, k2, k3, b7 are unknown coefficients.

[0092] Objective function for maximizing thermal stability (minimizing shrinkage):

[0093]

[0094] Among them, x 22 is the concentration of heat stabilizer (%), x 23 is the preheating time (min), x 24 is the cooling rate (℃ / min). m1, m2, m3, b8 are unknown coefficients.

[0095] The entire optimization process revolves around these multiple objective functions, and at the same time needs to meet the constraints of the feasible range of process parameters, production costs and environmental impact. Through such a series of advanced data analysis and optimization technologies, the optimal process parameter combination that meets the requirements of various fiber performance indicators can be finally obtained, laying a good foundation for subsequent production implementation.

[0096] The following is a brief description of the experiment of obtaining multiple parameters:

[0097] Antimicrobial concentration (x1) experiment: Prepare antimicrobial solutions of different concentrations (0.1%, 0.5%, 1%, 2%, 5%), immerse the fiber samples for 30 minutes and then dry. Use AATCC 100 method to test the antimicrobial properties of the treated fibers, and calculate the antimicrobial effect by the bacterial reduction rate. Repeat the experiment 3 times and take the average value to determine the optimal antimicrobial concentration.

[0098] Dispersant dosage (x2) experiment: Prepare a series of dispersant solutions with different concentrations (0.5g / L, 1g / L, 2g / L, 5g / L, 10g / L) and add a fixed amount of antimicrobial agent. Use a laser particle size analyzer to measure the particle size distribution after dispersion, and perform antimicrobial performance tests at the same time. Consider the dispersion effect and antimicrobial performance comprehensively to determine the optimal dispersant dosage.

[0099] pH value (x3) measurement: Use a pH meter to directly measure the pH value of the antimicrobial solution. Adjust the pH to different levels (such as 4, 5, 6, 7, 8) by adding a buffer solution (such as citric acid-phosphate buffer), test the antimicrobial performance under various pH conditions, and determine the optimal pH range.

[0100] UV protection agent concentration (x4) experiment: Prepare UV protection agent solutions of different concentrations (0.5%, 1%, 2%, 3%, 5%), immerse the fiber samples in them and dry them. Use a UV-visible spectrophotometer to measure the UV transmittance of the treated fibers and calculate the UPF value. Select the concentration that can achieve UPF ≥ 30 and is most cost-effective.

[0101] Synergist concentration (x5) experiment: Based on the determined UV protection agent concentration, add different concentrations of synergist (0.1%, 0.3%, 0.5%, 1%, 2%). After treating the fiber samples, test the UPF value, analyze the effect of the synergist on the anti-UV performance, and determine the optimal synergist concentration.

[0102] Processing temperature (x6) experiment: Set different processing temperatures (such as 160℃, 180℃, 200℃, 220℃, 240℃) during the spinning process to prepare fiber samples. Test the UV resistance of the fibers produced under various temperature conditions, and examine other physical properties of the fibers to comprehensively determine the optimal processing temperature.

[0103] Cellulose concentration (x7) experiment: Prepare cellulose solutions of different concentrations (5%, 8%, 10%, 12%, 15%) and prepare fibers by wet spinning. Test the mechanical properties of the obtained fibers, such as strength and elongation, and select the optimal concentration that meets the strength requirements (≥3cN / dtex) and is process-feasible.

[0104] Cross-linking agent dosage (x8) experiment: Add different dosages (1g / kg, 3g / kg, 5g / kg, 10g / kg, 20g / kg) of cross-linking agent to the cellulose solution, prepare fibers and perform heat treatment. Test the strength, water resistance and other properties of the cross-linked fibers to determine the optimal dosage of the cross-linking agent.

[0105] Stretching ratio (x9) experiment: Apply different stretching ratios (such as 2x, 3x, 4x, 5x, 6x) during the spinning process to prepare fiber samples. Measure the strength and elongation of each sample, analyze the effect of stretching ratio on mechanical properties, and determine the optimal stretching ratio.

[0106] Plasticizer content (x 10 ) Experiment: Prepare cellulose solutions containing different proportions of plasticizer (2%, 5%, 8%, 10%, 15%) to prepare fiber samples. Test the elongation and strength of each sample and select the optimal plasticizer content that can significantly increase the elongation without excessively reducing the strength.

[0107] Heat treatment temperature (x 11 ) Experiment: The fiber samples were heat treated at different temperatures (100℃, 120℃, 140℃, 160℃, 180℃) for 30 minutes. The elongation, strength and dimensional stability of the heat-treated fibers were tested, and the optimal heat treatment temperature was determined by comprehensively considering various properties.

[0108] Fiber fineness (x 12 ) Experiment: By adjusting the spinneret aperture and draft ratio, fibers of different fineness (1dtex, 1.5dtex, 2dtex, 2.5dtex, 3dtex) were prepared. The elongation and strength of fibers of different finenesses were tested, the effect of fineness on mechanical properties was analyzed, and the optimal fiber fineness was determined.

[0109] Crosslinking degree (x 13 ) Determination: Use FTIR spectroscopy to determine the carbonyl index of cross-linked fibers, or use chemical titration to determine the carboxyl content of cross-linked fibers. Calculate the degree of cross-linking based on the determination results, analyze the relationship between the degree of cross-linking and fiber durability, and determine the optimal range of the degree of cross-linking.

[0110] Stabilizer concentration (x 14) Experiment: Prepare treatment solutions containing different concentrations of stabilizer (0.5%, 1%, 2%, 3%, 5%) and impregnate fiber samples. Perform accelerated aging tests (such as xenon lamp irradiation or high temperature and high humidity treatment) to test the changes in fiber properties before and after treatment and determine the optimal stabilizer concentration.

[0111] Heat treatment time (x 15 ) Experiment: At the determined optimal heat treatment temperature, the fiber samples were heat treated for different times (5min, 10min, 20min, 30min, 60min). The durability of the fiber after heat treatment (such as washing resistance and abrasion resistance) was tested, and other performance changes were investigated to determine the optimal heat treatment time.

[0112] Fiber diameter (x 16 ) Measurement: Use a scanning electron microscope or laser particle size analyzer to measure the fiber diameter. Prepare fiber samples with different diameters (8μm, 10μm, 12μm, 15μm, 20μm), test their softness (such as draping method or KES-F system), and analyze the relationship between diameter and softness.

[0113] Softener dosage (x 17 ) Experiment: Prepare softener solutions of different concentrations (1g / kg, 3g / kg, 5g / kg, 10g / kg, 20g / kg) and soak the fiber samples. Use a fabric feel meter or expert rating method to evaluate the fiber softness and determine the best dosage that can significantly improve the softness without affecting other properties.

[0114] Mechanical soft handling strength (x 18 ) Experiment: Use mechanical softening equipment (such as a needle cloth machine) to treat the fiber with different intensities (1N / cm, 2N / cm, 3N / cm, 4N / cm, 5N / cm). Evaluate the softness and surface morphology of the treated fiber, and select the best treatment intensity that can effectively improve the softness without excessively damaging the fiber.

[0115] Fiber gap ratio (x 19 ) Determination: Prepare fiber webs of different densities and measure their volume and weight. Calculate the void ratio using the following formula:

[0116]

[0117] Among them, 表观 is the apparent density of the fiber web, ρ 真实 is the true density of the fiber material. Analyze the relationship between the gap ratio and air permeability to determine the optimal gap ratio range.

[0118] Hot pressing temperature (x 20) Experiment: The fiber web was subjected to heat pressing treatment at different temperatures (100℃, 120℃, 140℃, 160℃, 180℃). The air permeability of the treated samples was measured using an air permeability tester, the effect of the heat pressing temperature on the fiber web structure and air permeability was analyzed, and the optimal heat pressing temperature was determined.

[0119] Vibration frequency (x 21 ) Experiment: Use a vibration device to apply vibrations of different frequencies (10Hz, 20Hz, 30Hz, 40Hz, 50Hz) to the fiber web. Test the air permeability of the fiber web after vibration treatment, analyze the effect of vibration frequency on fiber arrangement and pore structure, and determine the optimal vibration frequency.

[0120] Heat stabilizer concentration (x 22 ) Experiment: Prepare treatment solutions containing different concentrations of heat stabilizer (0.5%, 1%, 2%, 3%, 5%) and impregnate fiber samples. Perform heat shrinkage test (such as treatment at 150°C for 30 minutes), measure the change in fiber length before and after treatment, calculate the heat shrinkage, and determine the optimal heat stabilizer concentration.

[0121] Preheating time (x 23 ) Experiment: The fiber is preheated for different times (5min, 10min, 20min, 30min, 60min) at a temperature lower than the melting point of the fiber (e.g. 120°C). The thermal shrinkage and mechanical properties of the fiber after preheating are tested to determine the optimal preheating time that can effectively improve thermal stability without damaging other properties.

[0122] Cooling rate (x 24 ) Experiment: After heat treatment, the fiber samples were cooled at different cooling rates (50℃ / min, 20℃ / min, 10℃ / min, 5℃ / min, 1℃ / min). The thermal shrinkage and crystallinity of the cooled fibers were tested, the effect of cooling rate on fiber structure and thermal stability was analyzed, and the optimal cooling rate was determined.

[0123] Specifically, the principle of the present invention is:

[0124] 1. Use principal component analysis and partial least squares regression to identify key process parameters. Principal component analysis can map the original process parameter x to the principal component space t through linear transformation, thereby revealing the intrinsic correlation structure between the parameters. The principal component load p can be analyzed to find out the main parameters that have a greater impact on the fiber performance y. At the same time, partial least squares regression can establish a quantitative relationship between process parameter x and performance index y, and by analyzing the size and significance of the regression coefficient b, the key influencing parameters can be further screened out. The combination of these two methods can effectively screen out those parameters that play a key role in the final fiber performance from a large number of process parameters.

[0125] 2. Use support vector regression to build a multi-objective prediction model. Support vector regression is an efficient regression algorithm based on statistical learning theory, which can establish a nonlinear input-output mapping relationship. Compared with traditional multivariate linear regression, the SVR model has stronger nonlinear fitting ability and can more accurately describe the complex relationship between process parameters and fiber properties. The selected key process parameters x^* will be used as the input of the model, and different fiber performance indicators y will be used as the output to train a multi-objective prediction model that can effectively predict various performances. This lays the foundation for subsequent multi-objective optimization.

[0126] 3. Use Latin hypercube sampling and K-means clustering to expand and filter data samples. Latin hypercube sampling is an effective multidimensional random sampling method that ensures that sample points in each dimension are evenly distributed, thereby improving the representativeness and robustness of model training. By generating a large amount of simulated data, the original historical data set can be enriched, laying a good foundation for subsequent clustering analysis. The K-means clustering algorithm can classify these simulated data according to various performance indicators and find the process parameter combination that meets the requirements. This data-driven method greatly reduces the experimental workload and improves the optimization efficiency.

[0127] 4. Use multi-objective optimization algorithms to find the optimal combination of process parameters. The multi-objective prediction model includes a variety of fiber performance indicators, such as antibacterial property, UV resistance, strength, elongation, etc. There are usually certain contradictions and trade-offs between these indicators. In order to achieve the best balance between various properties, multi-objective optimization algorithms such as NSGA-II or MOEA / D will be applied to find a set of Pareto optimal solutions. These Pareto solutions represent the optimal combination of process parameters under various constraints, providing optional solutions for production practice.

[0128] 5. Improve prediction accuracy through small-scale test verification and iterative optimization of model parameters. Several process parameter combinations in the Pareto optimal solution will be selected for small-scale tests to test the actual fiber performance data. These experimental data will then be compared with the prediction results of the multi-objective prediction model to evaluate the accuracy of the model. If a large deviation is found, the gradient descent method will be used to iteratively optimize the model parameters until the prediction error is minimized. This method of combining experiments and modeling can ensure that the optimization results can truly guide production practice and lay the foundation for subsequent process scale-up and commercial application.

[0129] In general, the present invention makes full use of advanced data analysis, machine learning and multi-objective optimization technologies to establish a systematic process parameter optimization method. Through a series of steps such as identifying key parameters, building a multi-objective prediction model, expanding and screening data samples, and finding the Pareto optimal solution, the collaborative optimization of multiple performance indicators of liquid-added production of plant functional fibers is achieved.

[0130] In order to better understand and implement the present invention, a specific embodiment 1 of the present invention is provided below. The specific steps of this embodiment 1 are described in detail as follows: First, in step S10, historical production data needs to be collected. Specifically, it is necessary to obtain multiple sets of historical data of each process parameter in the process of producing plant functional fibers by liquid addition, including slurry formula, sanding parameters, liquid pump injection parameters and corresponding fiber performance indicators. These process parameters can be represented by vector x = [x1, x2, ..., x n ] T Indicates that x i represents the i-th parameter, and n is the total number of process parameters. The fiber performance index can be expressed by the vector y = [y1, y2, ..., y m ] T Indicates that y j represents the jth performance indicator, and m is the total number of performance indicators.

[0131] The process of producing plant functional fiber by liquid addition is as follows: first, the plant functional additive is made into slurry, and the slurry formula includes the ratio and content of functional additive, dispersant and thickener, which can be expressed by vector z = [z1, z2, z3] T Indicated. Then, the fiber is sanded, and the sanding parameters include sanding time x4, sanding speed x5 and sanding medium size x6. Next, a liquid pump is used to inject the fiber at the end of the spinning, and its injection parameters include injection pressure x7, injection flow x8 and injection position x9. The final fiber performance indicators include at least antibacterial performance y1, UV resistance y2, strength y3, elongation y4, durability y5, softness y6, air permeability y7 and thermal stability y8.

[0132] In step S20, principal component analysis (PCA) and partial least squares regression (LSR) methods are used to analyze the relationship between process parameters and fiber performance indicators.

[0133] First, PCA is used to perform principal component analysis on the process parameters x, and the principal component score matrix T = [t1, t2, ..., t p ] and the principal component loading matrix P = [p1, p2, ..., p p ], where p is the number of principal components. Principal component score t ireflects the score of the i-th principal component in each sample, and the principal component loading p i It represents the correlation between the ith principal component and the original variable x. By analyzing the principal component loading, the key process parameters that have a greater impact on the fiber performance index y can be identified.

[0134] Next, the PLSR method will be used to establish the quantitative relationship between process parameters x and fiber performance indicators y. PLSR can obtain the regression coefficient matrix B = [b1, b2, ..., b m ], where b j is the jth performance indicator y j The corresponding regression coefficient vector. By analyzing the size and significance of the regression coefficient, the key influencing parameters can be further screened out.

[0135] Combining the results of PCA and PLSR, the process parameters that have a key impact on fiber performance will be selected according to the preset correlation index threshold (such as 0.5-0.75). These key influencing parameters can be expressed as vectors. Represents, where p≤n.

[0136] In step S30, the key influencing parameter x selected in step S20 is * , a multi-objective prediction model for plant functional fiber performance is constructed. Specifically, the support vector regression (SVR) algorithm is used to establish the key process parameters x * The mapping relationship between various fiber performance indicators y can be expressed as:

[0137] y=f(x * ;θ);

[0138] Among them, f(·) is the SVR model and θ is the model parameter.

[0139] The objective functions of the multi-objective prediction model include:

[0140] (maximum antimicrobial properties);

[0141] (maximum UV resistance);

[0142] (maximum strength);

[0143] (maximum elongation);

[0144] (maximize durability);

[0145] (softness minimized);

[0146] (maximum breathability);

[0147] (maximum thermal stability);

[0148] Among them, a i ,c i ,d i ,e i ,g i ,h i ,k i ,m i ,b i These objective functions are designed to maximize various fiber performance indicators, such as antibacterial properties, UV resistance, strength, elongation, durability, softness, air permeability and thermal stability.

[0149] At the same time, the following constraints need to be considered:

[0150] Feasible range of process parameters:

[0151] Production cost cap:

[0152] Environmental impact limits:

[0153] Minimum performance index requirement: f1(x * )≥90%,f2(x * )≥30,f3(x * )≥3cN / dtex,f4(x * )≥20%,f5(x * )≥50,f6(x * )≤3,f7(x * )≥100mm / s,f8(x * )≤5%;

[0154] in, is the value range of the ith process parameter, c i is the unit cost of the ith parameter, C max is the total cost upper limit, e i is the unit environmental impact of the ith parameter, E max The upper limit of environmental impact.

[0155] By establishing this multi-objective prediction model, the optimal process parameter combination that maximizes the above objective function can be found under the premise of satisfying various constraints.

[0156] In step S40, the Latin Hypercube Sampling (LHS) method is used to generate a large amount of simulated data based on the initially established SVR model. LHS is an effective multi-dimensional random sampling method that can ensure that the sample points in each dimension are evenly distributed, thereby improving the effect of model training. Specifically, LHS can generate random samples that obey a uniform distribution through the following formula:

[0157]

[0158] Among them, u i A random number that obeys a uniform distribution (0,1), n ​​is the number of sampling points. In this way, the expanded data set {x * ,y}.

[0159] In step S50, cluster analysis is performed on the data set expanded in step S40. Specifically, the K-means clustering algorithm is used to cluster each fiber performance index y j Perform clustering and get K j clusters. The objective function of K-means clustering is:

[0160]

[0161] Among them, C ij is the i-th cluster of the j-th performance indicator, μ ij is the centroid of the cluster. In this way, we can find data points that meet the requirements of various performance indicators.

[0162] Next, these clusters will be screened according to the pre-set performance indicator threshold range, such as antibacterial property ≥ 90%, UV resistance UPF value ≥ 30, strength ≥ 3cN / dtex, elongation ≥ 20%, etc., to obtain data points that meet the requirements.

[0163] In step S60, a multi-objective optimization algorithm, such as NSGA-II or MOEA / D, is applied to the data points selected in step S50 to obtain a set of Pareto optimal solutions as candidate optimal process parameter combinations. The multi-objective optimization algorithm can find the best balance between objective functions and provide a feasible solution set that meets the requirements of various performance indicators. The objective function of the NSGA-II algorithm is:

[0164]

[0165] Through this multi-objective optimization method, a set of Pareto optimal solutions can be obtained, each of which represents a feasible combination of process parameters.

[0166] Finally, in step S70, the candidate process parameter combination obtained in step S60 is selected for small-scale testing to obtain actual fiber performance data. Next, these experimental data are compared with the predicted data of the multi-objective prediction model established in step S30 to evaluate the accuracy of the model. If it is found that there is a significant deviation between the predicted results and the experimental results, the gradient descent method will be used to iteratively optimize the parameters θ of the multi-objective prediction model to minimize the prediction error:

[0167]

[0168] By continuously updating the model parameters θ until the prediction error is minimized, the multi-objective prediction model can be continuously improved and calibrated to improve its reliability.

[0169] In order to further better understand and implement the present invention, Example 2 of a specific application scenario of the present invention is provided below: A certain enterprise intends to develop a new type of plant functional fiber, hoping to give the fiber special functions such as antibacterial and anti-ultraviolet while maintaining excellent mechanical properties, adopt the liquid addition production process proposed in the present invention, and start to optimize the key parameters of the process.

[0170] First, the company collected historical data of the production line in the past five years, covering slurry formula, sanding parameters, liquid injection parameters and various fiber performance indicators, laying the foundation for subsequent analysis and modeling. The data sample is shown in Table 1:

[0171] Table 1 Historical production data sample

[0172] Sample No. Slurry formula Sanding parameters Liquid injection parameters Fiber performance index 1 <![CDATA[z1]]> <![CDATA[x1]]> <![CDATA[x2]]> <![CDATA[y1]]> 2 <![CDATA[z2]]> <![CDATA[x3]]> <![CDATA[x4]]> <![CDATA[y2]]> …… …… …… …… …… 100 <![CDATA[z 100 ]]> <![CDATA[x 100 ]]> <![CDATA[x 101 ]]> <![CDATA[y 100 ]]>

[0173] Where z = [z1, z2, z3] T is the slurry formula parameter, x=[x1,x2,…,x 24 ] T are the sanding and liquid injection parameters, y=[y1,y2,…,y8] T There are 8 fiber performance indicators.

[0174] Next, the company uses principal component analysis and partial least squares regression methods to conduct in-depth analysis of these historical data. First, PCA analysis is performed on the process parameter x to obtain the principal component score matrix T and the principal component load matrix P, as shown in the figure. Figure 2 As shown, Figure 2It shows that the variance contribution rate of the first three principal components has reached more than 80%, indicating that these three principal components can better summarize the main variation patterns of the original process parameters. Next, the principal component load P was analyzed, and it was found that x1 (antimicrobial agent concentration), x4 (UV protection agent concentration) and x7 (cellulose concentration) contributed the most to the principal components. It was preliminarily judged that these parameters may be the key influencing factors.

[0175] Then, the company used PLSR to establish the quantitative relationship between process parameters x and fiber performance indicators y, and obtained the regression coefficient matrix B. By analyzing the size and significance of each regression coefficient, x1, x4 and x7 were further confirmed as key influencing parameters, and x2 (dispersant dosage), x5 (synergist concentration) and x7 were also found to be 10 (Plasticizer content) also has a significant impact on certain performance indicators. Therefore, the company has determined a set of key process parameters x * =[x1,x2,x4,x5,x7,x 10 ] T .

[0176] Based on these key parameters, the company built a multi-objective prediction model for the performance of plant-based functional fibers. Specifically, they used the support vector regression algorithm to establish the following eight objective functions:

[0177] (maximum antimicrobial properties); (maximum UV resistance);

[0178] (maximum strength);

[0179] (maximum elongation); (maximize durability);

[0180] (softness minimized);

[0181] (maximum breathability);

[0182] (maximum thermal stability);

[0183] At the same time, the company set the following constraints:

[0184] Feasible range of process parameters:

[0185] Production cost cap:

[0186] Environmental impact limits:

[0187] Minimum performance index requirement: f1(x * )≥90%,f2(x * )≥30,f3(x * )≥3cN / dtex,f4(x * )≥20%,f5(x * )≥50,f6(x * )≤3,f7(x * )≥100mm / s,f8(x * )≤5%;

[0188] Next, the company used the Latin hypercube sampling method to generate 1,000 sets of simulated data, and on this basis, used the K-means clustering algorithm to classify each performance indicator, such as Figure 3 As shown in the figure, K-means clustering divides the sample points into 5 different clusters, represented by five different colors. The enterprise further screens these clusters and retains those data points that meet the minimum requirements of various performance indicators.

[0189] Finally, the company applied the NSGA-II multi-objective optimization algorithm to find the Pareto optimal solution in these screening data as the final optimal process parameter combination. In order to verify the optimization results, the company selected 3 groups of Pareto optimal solutions for small-scale experiments and tested the actual fiber properties, as shown in Table 2:

[0190] Table 2 Optimization results verification

[0191]

[0192] By analyzing the experimental data, the company found that the prediction model can generally predict various performance indicators more accurately, with only 1 to 2 percentage points of error in individual indicators. To further improve the accuracy of the model, the company uses the gradient descent method to iteratively optimize the model parameters until the prediction error is minimized.

[0193] After small-scale test verification and model optimization, the company determined the final optimal process parameter combination as: * =[1.8,1.2,2.1,1.5,2.4,1.8] T This combination not only meets the minimum requirements of various performance indicators, but also achieves the optimal balance in multiple performance indicators, laying a good foundation for subsequent process scale-up and commercial application.

[0194] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for optimizing process parameters for producing plant functional fibers by liquid addition, characterized in that: The following steps are involved: S10, collecting multiple sets of historical data of each process parameter in the process of producing plant functional fibers by liquid addition, wherein the process parameters include slurry formula, sanding parameters, liquid pump injection parameters and corresponding fiber performance indicators; S20, using principal component analysis and partial least squares regression methods to analyze the correlation index between each process parameter and each fiber performance index, and using a preset correlation index threshold to screen out key influencing parameters; S30, based on the selected key influencing parameters, construct a multi-objective prediction model for plant functional fiber performance, and use a support vector regression algorithm to establish a mapping relationship between the key influencing parameters and fiber performance indicators, which is recorded as a preliminary mapping; S40, generating a large amount of simulated data using Latin hypercube sampling method according to the preliminary mapping to obtain an expanded data set; S50, clustering each fiber performance index of the expanded data set using a K-means clustering method to obtain a plurality of clusters; screening the clusters using a preset threshold range of each fiber performance index to obtain a plurality of screened clusters; S60, applying a multi-objective optimization algorithm to the data points in the screening clusters to obtain a set of Pareto optimal solutions as candidate process parameter sets; S70, using the selected process parameter combination to conduct a small-scale test to obtain actual fiber performance data; comparing the experimental data with the predicted data of the multi-objective prediction model, evaluating the accuracy of the multi-objective prediction model, and iteratively optimizing the parameters of the multi-objective prediction model according to the experimental results; S80. Based on the experimental results and the optimized multi-objective prediction model, the response surface method is used to perform local refinement optimization to obtain the final optimal process parameter combination.

2. The process parameter optimization method for producing plant functional fibers by liquid addition according to claim 1, characterized in that: The liquid addition process for producing plant functional fibers specifically comprises preparing the plant functional additive into slurry, sand-milling the slurry, and then injecting the slurry into the fibers at the spinning end using a liquid pump.

3. The process parameter optimization method for producing plant functional fibers by liquid addition according to claim 2, characterized in that: The slurry formula specifically includes the ratio and content of the plant functional additive, the dispersant and the thickener.

4. The process parameter optimization method for producing plant functional fibers by liquid addition according to claim 3, characterized in that: The sanding parameters include sanding time, sanding speed, and sanding medium size.

5. The process parameter optimization method for producing plant functional fibers by liquid addition according to claim 4, characterized in that: The liquid pump injection parameters include injection pressure, injection flow rate, and injection position.

6. The process parameter optimization method for producing plant functional fibers by liquid addition according to claim 5, characterized in that: The fiber performance indicators include multiple or all of antibacterial properties, UV resistance, strength, elongation, durability, softness, air permeability, and thermal stability.

7. The process parameter optimization method for producing plant functional fibers by liquid addition according to claim 6, characterized in that: The fiber performance index is selected or weighted according to production requirements.

8. The process parameter optimization method for producing plant functional fibers by liquid addition according to claim 7, characterized in that: The objectives of the multi-objective prediction model include maximizing antibacterial properties, maximizing anti-ultraviolet properties, maximizing strength, maximizing elongation, maximizing antistatic properties, maximizing durability, maximizing softness, maximizing air permeability, and maximizing thermal stability, which match the fiber performance indicators; the constraints of the multi-objective prediction model include the feasible range of process parameters, the minimum requirements of fiber performance indicators, the upper limit of production costs, and the limitation of environmental impact.

9. The process parameter optimization method for producing plant functional fibers by liquid addition according to claim 8, characterized in that: The threshold range of each fiber performance indicator refers to the acceptable range of each performance indicator, including antibacterial performance ≥90%, UV resistance UPF value ≥30, strength ≥3cN / dtex, elongation ≥20%, antistatic performance ≤1000V, durability ≥50 washes, softness ≤level 3, air permeability ≥100mm / s, thermal stability ≤5% shrinkage, and anti-wrinkle performance ≥level 3.

10. The process parameter optimization method for producing plant functional fibers by liquid addition according to claim 9, characterized in that: The preset correlation index threshold is 0.5-0.75.

Images