Lyocell fiber spinning slurry process simulation and quality prediction method and system

Abstract

The present application relates to the technical field of fiber production process simulation modeling, in particular to a lyocell fiber spinning dope process simulation and quality prediction method and system, the method comprising the following steps: using a simulation system to simulate the swelling process of cellulose in NMMO solvent, generating multiple sets of input-output data pairs to construct a data set; introducing the physical model of the simulation system as a regularization constraint in the gradient boosting tree algorithm to obtain an improved gradient boosting tree algorithm; using the data set to train the improved gradient boosting tree algorithm to predict the quality of the lyocell fiber spinning dope. The present application first proposes a dope quality prediction scheme based on simulation data and an improved gradient boosting tree algorithm, greatly reducing the dope development time and cost, and the improved gradient boosting tree algorithm incorporates the physical model of the simulation system into the model training process by adding a physical constraint regularization term to the loss function, enhancing the scientificity of the prediction results and improving the reliability of the model industrial application.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of fiber production process simulation modeling, in particular to a lyocell fiber spinning dope process simulation and quality prediction method and system. BACKGROUND

[0002] Lyocell fiber is recognized by the world as the "green fiber" of the 21st century. Its production process uses pulp as raw material, uses N-methyl morpholine-N-oxide (NMMO) as solvent, and is prepared by wet spinning technology. The core of the lyocell fiber production process is the swelling process of cellulose in NMMO solution. Swelling, as the starting stage of the whole process, plays a crucial role. The swelling of cellulose refers to the penetration of solvent molecules into the cellulose molecular chain under the action of NMMO solvent, which destroys the tight hydrogen bond network, relaxes the cellulose structure and swells, and forms the spinning dope, laying the foundation for subsequent complete dissolution.

[0003] At present, the production process parameters of the spinning dope mainly depend on the laboratory environment and the small test production line. Necessary parameters are obtained through a series of experiments, and then these parameters are applied to large-scale actual production process. This method can accurately simulate the production conditions on a small scale, so as to obtain more accurate and stable process parameters. However, the experimental process of laboratory and small test production line involves multiple links and steps, which needs to consume a certain amount of time. In an emergency, if the process parameters need to be adjusted quickly, the long experimental period may delay the progress of actual production. In addition, the construction and operation cost of laboratory and small test production line is high due to the consumption of raw materials, the purchase of equipment and the training of personnel.

[0004] At the same time, the prediction method of the quality of the spinning dope has not been systematically proposed at present, and the quality of the dope is affected by many factors, which makes the quality prediction have a certain complexity.

[0005] Therefore, in order to improve the production efficiency, save the production cost, and help the production personnel to comprehensively understand the operation skills and process flow of the spinning dope production, it is urgent to present the spinning dope production process in digital form, establish the spinning dope production process simulation model, realize the visualization and effective control of the spinning dope production. At the same time, it is also urgent to explore the prediction method of the quality of the spinning dope, so as to improve the efficiency of the quality control of the dope, provide real-time guidance for actual production, and further promote the intelligent and efficient development of the lyocell fiber production process. SUMMARY

[0006] In view of the defects in the prior art, the present application provides a lyocell fiber spinning dope process simulation and quality prediction method and system.

[0007] In order to achieve the above object, the first aspect of the present application provides a lyocell fiber spinning dope process simulation and quality prediction method, the method comprising the following steps: constructing a spinning dope production process simulation system based on the swelling mechanism of cellulose in NMMO solution, generating simulation data of lyocell fiber spinning dope, and then obtaining multiple sets of input and output data pairs to construct a data set; introducing the physical model of the spinning dope production process simulation system as a regularization constraint in the gradient boosting tree algorithm to obtain an improved gradient boosting tree algorithm; using the data set to complete the training and verification of the improved gradient boosting tree algorithm to obtain a quality prediction model for predicting the key indicators of lyocell fiber spinning dope quality. The present application first proposes a spinning dope quality prediction scheme based on simulation data and an improved gradient boosting tree algorithm, which greatly reduces the spinning dope development time and cost, and the improved gradient boosting tree algorithm integrates the physical model of the simulation system into the model training process by adding a physical constraint regularization term in the loss function, which enhances the scientificity of the prediction results and improves the reliability of the model industrial application.

[0008] Optionally, the physical model of the spinning dope production process simulation system comprises a solvent permeation model, a swelling kinetics model, a stirring action and uniformity model, and a viscosity calculation model.

[0009] Optionally, the solvent permeation model satisfies the following relationship:

[0010]

[0011] wherein, is the solvent concentration of the NMMO solution, t is the time, D is the diffusion coefficient related to the degree of polymerization of cellulose, is the spatial second-order derivative of the solvent concentration gradient.

[0012] Optionally, the swelling kinetics model satisfies the following relationship:

[0013]

[0014] wherein, is the swelling degree of cellulose in the NMMO solvent, is the swelling rate constant, is the solvent concentration, is the equilibrium solvent concentration, is the maximum swelling degree.

[0015] Optionally, the stirring action and uniformity model satisfies the following relationship:

[0016]

[0017]

[0018] wherein, is the solvent concentration, is the flow field velocity brought by stirring, representing the fluid motion caused by stirring, is the spatial first-order derivative of the solvent concentration gradient, U is the homogeneity of the lyocell fiber spinning dope, is the average solvent concentration in the NMMO solution, V is the volume of the cellulose and NMMO solution mixing system.

[0019] Optionally, the viscosity calculation model satisfies the following relationship:

[0020]

[0021] wherein, is the dope viscosity of the lyocell fiber spinning dope, is the viscosity constant, is the cellulose content, is the swelling degree, is the solvent concentration, m, n and p are respectively the influence weight of the cellulose content, the swelling degree and the solvent concentration on the dope viscosity, is the viscosity activation energy.

[0022] Optionally, the input and output data pair includes input data for inputting the mass prediction model and output data corresponding to the input data, the input data includes the cellulose polymerization degree, the solvent concentration, the cellulose content, the temperature and the stirring rate, and the output data includes the homogeneity and the dope viscosity.

[0023] Optionally, the physical model of the spinning dope production process simulation system is introduced into the gradient boosting tree algorithm as a regularization constraint, including a homogeneity constraint term and a viscosity constraint term, and the homogeneity constraint term and the viscosity constraint term satisfy the following relationships respectively:

[0024]

[0025]

[0026] wherein, is the theoretical homogeneity of the jth sample, is the influence rate of stirring intensity on the homogeneity of the dope, is the numerical value of the flow field velocity of the jth sample, is the proportional coefficient, is the stirring rate of the jth sample, is the theoretical viscosity of the jth sample, viscosity constant, m, n and p are the weight of the cellulose content, the swelling degree and the solvent concentration on the viscosity of the slurry respectively, cellulose content of the jth sample, swelling degree of the jth sample, solvent concentration of the jth sample, viscosity activation energy, R is the gas constant, temperature of the jth sample.

[0027] Optionally, the loss function of the improved gradient boosting tree algorithm satisfies the following relationship:

[0028]

[0029]

[0030] wherein, total loss, data-driven term, physical loss term, viscosity loss term, uniformity loss term, regularization weight, N is the total number of samples, prediction value of the jth sample by the algorithm, true value of the jth sample.

[0031] In a second aspect, the present application provides a lyocell fiber spinning slurry process simulation and quality prediction system, which comprises a data acquisition device, a data output device, a processor and a storage, the storage comprises a computer readable storage medium, the computer readable storage medium stores a computer program, the computer program comprises program instructions, and the program instructions enable the processor to realize the lyocell fiber spinning slurry process simulation and quality prediction method provided by the present application when executed by the processor.

[0032] In summary, the present application has at least the following beneficial effects:

[0033] 1. The present application provides a lyocell fiber spinning slurry quality systematic prediction scheme based on simulation data and an improved gradient boosting tree algorithm for the first time, fills the technical gap, provides a complete process from data generation to lyocell fiber spinning slurry quality prediction, and improves the systematicness and scientificity of slurry quality prediction.

[0034] 2、The method generates a large amount of data through a simulation system, and sufficient data can be obtained to train a prediction model without entity experiments, thereby greatly reducing research and development time and economic cost, supporting rapid process parameter optimization, and being particularly suitable for emergency adjustment scenarios.

[0035] 3、The improved gradient boosting tree algorithm of the method integrates the physical model of the spinning dope production process simulation system into the model training process by adding a physical constraint regularization term in the loss function, ensures that the prediction conforms to the physical properties of cellulose swelling, viscosity and uniformity, enhances the scientific rationality of the prediction result, avoids the possible physical deviation of the traditional data-driven model, and improves the reliability of the model in industrial application.

[0036] 4、The present application provides a system adapted to the method, which can improve the practicability of the method. BRIEF DESCRIPTION OF DRAWINGS

[0037] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiments will be briefly introduced as follows, and it should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0038] Figure 1 A process flow diagram of a lyocell fiber spinning dope process simulation and quality prediction method according to an embodiment of the present application;

[0039] Figure 2 A framework diagram of a lyocell fiber spinning dope process simulation and quality prediction system according to an embodiment of the present application. DETAILED DESCRIPTION

[0040] The specific embodiments of the present application will be described in detail below, and it should be noted that the embodiments described herein are only for illustration and do not limit the present application. In the following description, a large number of specific details are set forth in order to provide a thorough understanding of the present application. However, it is obvious to those skilled in the art that the present application does not have to be implemented with these specific details. In other examples, in order to avoid obscuring the present application, well-known circuits, software or methods are not specifically described.

[0041] Throughout this specification, the mention of “one embodiment,” “an embodiment,” “one example,” or “an example” means that a particular feature, structure, or characteristic described in connection with the embodiment or example is included in at least one embodiment of the application. The appearances of the phrases “in one embodiment,” “in an embodiment,” “one example,” or “an example” in various places in the specification are not necessarily all referring to the same embodiment or example. Furthermore, the particular features, structures, or characteristics can be combined in any suitable manner in one or more embodiments or examples. Furthermore, a person of ordinary skill in the art will understand that the drawings provided herein are for illustrative purposes only and are not necessarily drawn to scale.

[0042] It is to be understood that, in an alternative embodiment, the same symbols or letters appearing in all formulas have the same meaning and value except as otherwise specifically noted.

[0043] In an alternative embodiment, referring to Figure 1 The application provides a lyocell fiber spinning dope process simulation and quality prediction method, and the method comprises the following steps:

[0044] S1, based on the swelling mechanism of cellulose in NMMO solution, a spinning dope production process simulation system is constructed, simulation data of lyocell fiber spinning dope is generated, and then a plurality of input and output data pairs are obtained to construct a data set.

[0045] Specifically, in this embodiment, the physical model of the spinning dope production process simulation system includes a solvent permeation model, a swelling kinetics model, a stirring action and uniformity model, and a viscosity calculation model. In addition, it should be noted that the cellulose described later is lyocell cellulose.

[0046] The solvent permeation model describes the dynamic process of NMMO solvent molecules entering the porous structure of cellulose by diffusion. This is the initial step of swelling, and the penetration of solvent molecules provides the driving force for subsequent hydrogen bond breaking and cellulose swelling. The role of the solvent permeation model is to simulate how the solvent permeates into the cellulose microstructure over time, generating a concentration distribution, which provides input for the swelling kinetics model. The output of the solvent permeation model is the distribution of solvent concentration over time and space, and the remaining parameters are model inputs. The solvent permeation model satisfies the following relationship:

[0047]

[0048] wherein, is the solvent concentration of NMMO solution, representing the mass of solvent molecules per unit volume of NMMO solution, which varies with time and spatial position, and its value range is generally 400-700 kg / m3; t is time, representing the evolution of the diffusion process of solvent molecules; D is the diffusion coefficient related to the degree of polymerization of cellulose, representing the penetration ability of solvent molecules in cellulose, which is affected by temperature and cellulose structure; is the spatial second-order derivative of the solvent concentration gradient, representing the diffusion direction and intensity of solvent molecules.

[0049] The diffusion coefficient D satisfies the following relationship:

[0050]

[0051] wherein, is the pre-exponential factor of the diffusion process, representing the diffusion ability under ideal conditions, and its value range is 10 -9 ~10 -8 . is the diffusion activation energy, representing the energy required to overcome the resistance of cellulose structure, which is determined by diffusion experiments at different temperatures, and its value range is 20-40 kJ / mol, and in the subsequent simulation process, the initial value of the diffusion activation energy can be set to 30 kJ / mol; R is the gas constant; T is the temperature, and its value range is 353-393 K, and the increase of T will accelerate molecular motion, and then D will increase; DP is the degree of polymerization of cellulose, and its value range is 400-1200, and high DP will increase the length of molecular chain and the density of hydrogen bond, and then D will decrease; is the empirical exponent, representing the inhibition degree of DP on the diffusion of solvent molecules, and its value range is 0.5-1, and in the subsequent simulation process, the initial value of can be set to 0.7.

[0052] The swelling kinetics model describes the change of the swelling degree of cellulose with time, depicts the relaxation and swelling process of cellulose structure, and can simulate the dynamic behavior of the expansion of cellulose molecular chain spacing and the rupture of hydrogen bond after the penetration of solvent molecules. The role of the swelling kinetics model is to calculate the swelling degree of cellulose according to the solvent concentration, and reflect the formation process of pulp. The output of the swelling kinetics model is the change of the swelling degree with time, and the remaining parameters are the input of the model, and the swelling kinetics model satisfies the following relationship:

[0053]

[0054] wherein, is the swelling degree of cellulose in NMMO solvent, defined as the ratio of the swelling volume of cellulose to the initial volume, representing the unswollen state; is the swelling rate constant, representing the swelling speed, which is affected by temperature; To balance the solvent concentration, the solvent concentration at which saturation of swelling is reached is characterized with the cellulose content The positive correlation has a reference value range of 50-200 kg / m³, in the subsequent simulation process, The initial value of the cellulose content can be set to 100 kg / m³; The maximum swelling degree represents the limit of cellulose swelling, which is limited by and DP. The value range of the cellulose content is generally 50-150 kg / m³.

[0055] The swelling rate constant and the maximum swelling degree satisfy the following relationship respectively:

[0056]

[0057]

[0058] wherein, is a pre-exponential factor of swelling kinetics, representing the swelling rate under ideal conditions, in the subsequent simulation process, The initial value of the cellulose content can be set to 100 kg / m³; If the cellulose swelling is too slow, increase , otherwise decrease ; is a swelling activation energy, representing the energy barrier of hydrogen bond breaking, which is determined by swelling experiments at different temperatures, and the value range is 20-50 kJ / mol, in the subsequent simulation process, the initial value of the swelling activation energy can be set to 40 kJ / mol; is a constant, representing the material properties, The value range of the cellulose content is generally 5-20 kg / m³, in the subsequent simulation process, The initial value of the cellulose content can be set to 8 kg / m³; is an empirical index, representing the inhibition effect on DP, and the value range is 0.5-1, in the subsequent simulation process, The initial value of the cellulose content can be set to 0.7.

[0059] The stirring effect and uniformity model includes a stirring effect model and a uniformity model, which describes how stirring enhances the mixing of solvent and cellulose through convection, optimizes the concentration distribution, and affects the slurry uniformity. The stirring rate adjusts the distribution of the solvent through the flow field velocity, ultimately affecting the slurry quality. The stirring effect and uniformity model function is to combine diffusion and convection to update the solvent concentration distribution and calculate the uniformity of the lyocell fiber spinning slurry. The output of the stirring effect and uniformity model is the updated solvent concentration and uniformity, and the remaining parameters are the model inputs. The stirring effect model and the uniformity model satisfy the following relationship in turn:

[0060]

[0061]

[0062] wherein, is the flow field velocity caused by stirring, representing the fluid motion caused by stirring; is the spatial first-order derivative of the solvent concentration gradient, is the convection term, representing the spatial transport effect of stirring on the concentration; U is the uniformity of the lyocell fiber spinning dope, the greater the value, the more uniform the distribution of the lyocell fiber spinning dope, which is one of the key indicators of the quality of the lyocell fiber spinning dope; is the average concentration of the solvent in the NMMO solution, obtained by calculating the spatial average of the solvent concentration; V is the volume of the cellulose and NMMO solution mixing system.

[0063] The flow field velocity satisfies the following relationship:

[0064]

[0065] wherein, is the proportionality coefficient, representing the stirring equipment efficiency, and the value is 10 −4 m / rpm, determined by experiment or equipment parameters; is the stirring rate, generally in the range of 100-1000 rpm, and increasing represents the enhanced convection.

[0066] The viscosity calculation model describes the rheological properties of the lyocell fiber spinning dope, which is comprehensively affected by the cellulose content, the swelling degree, the solvent concentration and the temperature. Its role is to quantify the rheological properties of the dope and calculate the viscosity of the dope. The viscosity of the dope can describe the fluid mechanics behavior of the dope, and directly affects the feasibility of the spinning process and the product quality, so it can be used as a key indicator for quality prediction. The output of the viscosity calculation model is the viscosity of the dope, and the remaining parameters are the input. The viscosity calculation model satisfies the following relationship:

[0067]

[0068] wherein, is the viscosity of the lyocell fiber spinning dope; is the viscosity constant, representing the reference viscosity, and the value is 10 −5 obtained by experimental fitting; m, n and p are the influence weights of the cellulose content, the swelling degree and the solvent concentration on the viscosity of the dope, respectively. In this embodiment, the values of m, n and p are 1, 0.5 and -0.3, respectively; The viscosity activation energy is the energy required to activate the viscosity, and is used to characterize the temperature effect. The viscosity activation energy is usually in the range of 20-50 kJ / mol. In the subsequent simulation process, the initial value of the viscosity activation energy can be set to 30 kJ / mol.

[0069] More specifically, the construction of the spinning dope production process simulation system is completed on the Ansys fluent simulation software. In the Ansys fluent simulation software: the solvent permeation model provides the diffusion basis to drive the swelling and concentration distribution; the swelling kinetics model connects the microstructure change with the macro mass, and outputs the swelling degree; the stirring action and uniformity model optimizes the solvent distribution, and outputs the uniformity; the viscosity calculation model integrates the results of the previous models, and outputs the dope viscosity.

[0070] Further, since the value range and initial value of some parameters in each physical model of the spinning dope production process simulation system have been given in the foregoing, these parameters can be adjusted in the simulation process to generate simulation data of the lyocell fiber spinning dope, and then 10,000 groups of input-output data pairs are generated, each group of input-output data pair can be used as a sample. The input-output data pair includes input data for inputting into the quality prediction model and output data corresponding to the input data, the input data includes cellulose polymerization degree, solvent concentration, cellulose content, temperature and stirring rate, and the output data includes uniformity and dope viscosity. The method of generating a large amount of data by the simulation system can digitally present the cellulose swelling dynamic process, and sufficient data can be obtained without physical experiments to obtain the subsequent prediction model.

[0071] Further, the input-output data pairs are normalized using Min-Max normalization, and the 10,000 groups of normalized input-output data pairs are used to construct a data set.

[0072] S2, introducing the physical model of the spinning dope production process simulation system as a regularization constraint in the gradient boosting tree algorithm to obtain an improved gradient boosting tree algorithm.

[0073] Specifically, in this embodiment, the improved gradient boosting tree algorithm can also be referred to as a physically constrained and regularized improved gradient boosting tree algorithm, which introduces the physical model of the spinning dope production process simulation system as a regularization constraint on the basis of the gradient boosting tree algorithm to optimize the loss function and residual calculation. The physical model of the spinning dope production process simulation system is introduced as a regularization constraint in the gradient boosting tree algorithm, including a uniformity constraint term and a viscosity constraint term, and satisfies the following relationships, respectively:

[0074]

[0075]

[0076] wherein, is the theoretical homogeneity of the jth sample, is the rate of influence of stirring intensity on homogeneity of the slurry, is the numerical value of the flow field velocity of the jth sample, is the stirring rate of the jth sample, is the theoretical viscosity of the jth sample, is the cellulose content of the jth sample, is the swelling degree of the jth sample, is the solvent concentration of the jth sample, is the temperature of the jth sample. The value range of is generally 0.01-0.1, and the initial value can be set to 0.01.

[0077] The homogeneity constraint term adopts an exponential approximation formula of stirring action and homogeneity model for the calculation of homogeneity, rather than directly using the calculation method of homogeneity in the stirring action and homogeneity model, mainly for the following considerations:

[0078] 1. The calculation of homogeneity in the stirring action and homogeneity model involves spatial integration, which is complex and difficult to directly embed in the loss function of the machine learning model. Therefore, the homogeneity constraint term uses an exponential function to directly link homogeneity and stirring rate, simplifying the calculation and suitable for rapid iterative training.

[0079] 2. The simulation system needs to accurately simulate the dynamic distribution of solvent concentration to generate real homogeneity data. The homogeneity constraint term only needs to provide a rough but physically reasonable homogeneity estimate for the improved gradient boosting tree algorithm prediction, ensuring that the prediction result conforms to the trend of stirring action, rather than accurately reproducing the simulation.

[0080] 3. The physical constraint term needs to be calculated quickly in each training round, and integration will significantly increase the calculation cost. The homogeneity constraint term uses an exponential approximation formula that is simple and efficient, and retains the positive correlation between stirring intensity and homogeneity.

[0081] 4. The simulation data is stable or quasi-stable, and homogeneity is mainly driven by stirring intensity, while the exponential function can effectively capture this relationship, meeting the actual needs of slurry quality prediction.

[0082] Further, the loss function of the improved gradient boosting tree algorithm satisfies the following relationship:

[0083]

[0084]

[0085] wherein, is the total loss, MSE is the data-driven term, and MSE measures , and , deviation of is the predicted value of slurry viscosity by the algorithm, is the predicted value of uniformity by the algorithm, is the true value of slurry viscosity, is the true value of uniformity, is the physical loss term, measures , and , deviation of is the theoretical viscosity, is the theoretical uniformity, is the viscosity loss term, is the uniformity loss term, is the regularization weight, N is the total number of samples, is the predicted value of the jth sample by the algorithm, is the true value of the jth sample. In this embodiment, the slurry viscosity and uniformity simulated by the simulation system are taken as the true values of the slurry viscosity and uniformity.

[0086] The viscosity loss term and the uniformity loss term satisfy the following relationships respectively:

[0087]

[0088]

[0089] wherein, is the predicted value of slurry viscosity of the jth sample by the algorithm, is the predicted value of uniformity of the jth sample by the algorithm.

[0090] The residual of the improved gradient boosting tree algorithm is simultaneously driven by data and physical model, balances data fitting and physical consistency, ensures that the prediction results of slurry viscosity and uniformity are accurate and consistent with the swelling law, and the residual of the improved gradient boosting tree algorithm satisfies the following relationship:

[0091]

[0092]

[0093] wherein, is the viscosity residual of the jth sample, is the uniformity residual of the jth sample, is the true value of slurry viscosity in the jth sample, is the true value of uniformity in the jth sample. or represents data error, driving model to fit simulation data. or represents physical deviation, constraining predicted value close to theoretical value. is a regularization weight, and is a task weight, the initial value of the two task weights is 1.

[0094] More specifically, the implementation process of the improved gradient boosting tree algorithm can be divided into 5 stages, as follows:

[0095] 1. Data preparation stage: based on the content described in step S1, the data set is constructed, and the data set is divided into training set, validation set and test set according to the ratio of 8:1:1.

[0096] 2. Physical constraint calculation: the exponential approximation formula of the uniformity model in the simulation system and the viscosity calculation model are used to calculate the theoretical uniformity and theoretical viscosity corresponding to each sample as the physical constraint, that is, the theoretical uniformity and theoretical viscosity corresponding to each sample are calculated as the physical constraint according to the uniformity constraint term and the viscosity constraint term.

[0097] 3. Model initialization: first, the predicted values of uniformity and slurry viscosity are initialized, taking the mean value of the target variable, that is, , wherein, is the initialized uniformity prediction value, is the initialized slurry viscosity prediction value; then two empty decision tree sets are created for storing the trees of uniformity prediction value and slurry viscosity prediction value respectively, ensuring that each output has an independent optimization path, and the initial tree parameters include maximum depth 4~8 and minimum leaf weight 1.

[0098] 4. Iterative training: the adjusted residual error is calculated for each sample by combining data error and physical deviation, one decision tree is trained for slurry viscosity and uniformity respectively, the corresponding residual error is fitted, and the prediction value is updated according to the step size of 0.05 to ensure smooth updating; continuously iterate until the preset tree number or the validation set error converges, and output the final prediction value of slurry viscosity and uniformity.

[0099] 5. Model training, evaluation and testing: use the data set to complete the training, validation and testing of the improved gradient boosting tree algorithm, which is used to predict the key indicators of lyocell fiber spinning slurry quality.

[0100] From the above, the improved gradient boosting tree algorithm embeds the physical model of the simulation system in the loss function, not only retains the strong fitting ability of gradient boosting tree for nonlinear relationship and feature interaction, but also ensures that the prediction result conforms to the physical law of the swelling process of cellulose in NMMO solvent, thereby significantly improving the physical consistency, prediction accuracy and adaptability of the model to simulation data, especially suitable for the industrial scene of pulp quality prediction. The traditional gradient boosting tree only relies on data-driven and lacks constraints on physical mechanisms, making it difficult to meet this specific demand.

[0101] S3, using the data set to complete the training and verification of the improved gradient boosting tree algorithm, and obtaining a quality prediction model for predicting the key indicators of lyocell fiber spinning pulp quality.

[0102] Specifically, in this embodiment, the data set obtained in step S1 is divided into a training set, a validation set and a test set according to the ratio of 8:1:1, so as to complete the training, verification and testing of the improved gradient boosting tree algorithm, and finally obtain a quality prediction model. Then, by adjusting the input of the quality prediction model, the corresponding key indicators of lyocell fiber spinning pulp quality can be obtained.

[0103] More specifically, a plurality of quality prediction models can be trained by adjusting the parameters such as the number of trees, the maximum depth, the learning rate and the regularization weight in the improved gradient boosting tree algorithm. Then, each quality prediction model is tested on the validation set, and then the root mean square error and the determination coefficient of each quality prediction model are calculated to ensure that the root mean square error of the quality prediction model is less than 5% and the determination coefficient is higher than 0.9, so as to select the best quality prediction model.

[0104] This embodiment uses simulation data to train the improved gradient boosting tree algorithm to construct a quality prediction model for predicting the key indicators of lyocell fiber spinning pulp quality, which can greatly reduce the research and development time and economic cost, and the prediction result conforms to the physical law, supports rapid process parameter optimization, and is especially suitable for emergency adjustment scene.

[0105] It should be noted that in some cases, the actions described in the specification can be performed in different orders and still achieve the desired results, and in this embodiment, the order of the steps given is only to make the embodiment look clearer and more understandable, and is not a limitation.

[0106] In an alternative embodiment, see Figure 2In order to improve the practicability of the method, the application further provides a lyocell fiber spinning slurry process simulation and quality prediction system, the lyocell fiber spinning slurry process simulation and quality prediction system comprises a data acquisition device 1, a data output device 2, a processor 3 and a storage 4, the storage 4 comprises a computer readable storage medium, the computer readable storage medium stores a computer program, the computer program comprises program instructions, and the program instructions enable the processor 3 to realize the content in steps S1 to S3 when the program instructions are executed by the processor 3.

[0107] In summary, the application has at least the following beneficial effects:

[0108] 1. The prior art lacks a systematic prediction method for slurry quality, and only relies on experimental data for indirect derivation. The present method first proposes a lyocell fiber spinning slurry quality systematic prediction scheme based on simulation data and an improved gradient boosting tree algorithm, fills the technical gap, provides a complete process from data generation to lyocell fiber spinning slurry quality prediction, and improves the systematicness and scientificity of slurry quality prediction.

[0109] 2. The prior art adjusts process parameters to obtain spinning slurry meeting the quality by relying on laboratory and small-scale experiments, which is time-consuming and costly. The present method generates a large amount of data through a simulation system, and can obtain sufficient data to train a prediction model without physical experiments, greatly reducing the research and development time and economic cost, supporting rapid process parameter optimization, and being particularly suitable for emergency adjustment scenarios.

[0110] 3. The prior art does not systematically utilize the physical laws of cellulose swelling process in NMMO solvent, and the slurry quality prediction result may not conform to the actual mechanism. The improved gradient boosting tree algorithm of the present method incorporates the physical model of the spinning slurry production process simulation system into the model training process by adding a physical constraint regularization term in the loss function, ensures that the prediction conforms to the physical properties of cellulose swelling, viscosity and uniformity, enhances the scientific rationality of the prediction result, avoids the possible physical deviation of traditional data-driven models, and improves the reliability of the model in industrial application.

[0111] 4. The present application provides a system adapted to the method, which can improve the practicability of the method.

[0112] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and are not intended to limit the present application; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that the technical solutions recorded in the foregoing embodiments can still be modified, or some or all of the technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application, and they should be covered in the scope of the claims and the description of the present application.

Claims

1. A method for simulating and predicting the quality of lyocell fiber spinning sizing process, characterized in that, Includes the following steps: A simulation system for the production process of spinning pulp was constructed based on the swelling mechanism of cellulose in NMMO solution. Simulation data of lyocell fiber spinning pulp was generated, and multiple sets of input-output data pairs were obtained to construct a dataset. An improved gradient boosting tree algorithm is obtained by incorporating the physical model of the spinning pulp production process simulation system as a regularization constraint into the gradient boosting tree algorithm. The physical models of the spinning pulp production process simulation system include solvent penetration model, swelling kinetics model, stirring and uniformity model, and viscosity calculation model. The loss function of the improved gradient boosting tree algorithm satisfies the following relationship: ， ， in, For the total loss, For data-driven items, For physical loss items, This is the viscosity loss term. For uniformity loss term, Here, N represents the regularization weights, and N is the total number of samples. Let be the algorithm's predicted value for the j-th sample. This represents the true value of the j-th sample. The improved gradient boosting tree algorithm was trained and validated using the dataset to obtain a quality prediction model, which is used to predict key indicators of the quality of lyocell fiber spinning pulp.

2. The method for simulation and quality prediction of lyocell fiber spinning sizing process according to claim 1, characterized in that, The solvent permeation model satisfies the following relationship: ， in, denoted as the solvent concentration of the NMMO solution, t as time, and D as the diffusion coefficient related to the degree of polymerization of cellulose. This is the spatial second derivative of the solvent concentration gradient.

3. The method for simulation and quality prediction of lyocell fiber spinning sizing process according to claim 2, characterized in that, The swelling kinetics model satisfies the following relationship: ， in, The degree of swelling of cellulose in NMMO solvent. The swelling rate constant is The solvent concentration is... To balance the solvent concentration, This represents the maximum degree of swelling.

4. The method for simulation and quality prediction of lyocell fiber spinning sizing process according to claim 3, characterized in that, The stirring effect and the homogeneity model satisfy the following relationship: ， ， in, The solvent concentration is... The velocity of the flow field caused by stirring characterizes the fluid motion induced by stirring. Let be the spatial first derivative of the solvent concentration gradient, and U be the uniformity of the sizing paste used in Lyocell fiber spinning. V represents the average solvent concentration in the NMMO solution, and V is the volume of the mixture of cellulose and NMMO solution.

5. The method for simulation and quality prediction of lyocell fiber spinning sizing process according to claim 4, characterized in that, The viscosity calculation model satisfies the following relationship: ， in, The viscosity of the slurry for spinning lyocell fibers. Where is the viscosity constant. The content is cellulose. The degree of swelling, The solvent concentration is given, and m, n, and p are the weights of the effects of the cellulose content, the degree of swelling, and the solvent concentration on the viscosity of the porridge, respectively. This is the viscosity activation energy.

6. The method for simulation and quality prediction of lyocell fiber spinning sizing process according to claim 5, characterized in that: The input-output data pair includes input data for inputting the quality prediction model and output data corresponding to the input data. The input data includes the degree of polymerization of cellulose, the solvent concentration, the cellulose content, the temperature, and the stirring rate. The output data includes the uniformity and the viscosity of the porridge.

7. The method for simulation and quality prediction of lyocell fiber spinning sizing process according to claim 5, characterized in that, The physical model of the spinning slurry production process simulation system is introduced into the gradient boosting tree algorithm as a regularization constraint, including a uniformity constraint term and a viscosity constraint term. The uniformity constraint term and the viscosity constraint term satisfy the following relationships respectively: ， ， in, For the theoretical uniformity of the j-th sample, The rate at which stirring intensity affects the uniformity of the porridge. Let be the numerical value of the flow field velocity for the j-th sample. This is the proportionality coefficient. Let j be the stirring rate of the j-th sample. Let j be the theoretical viscosity of the j-th sample. Let m be the viscosity constant, and m, n, and p be the weights of the effects of the cellulose content, the degree of swelling, and the solvent concentration on the viscosity of the porridge, respectively. Let be the cellulose content of the j-th sample. The degree of swelling of the j-th sample, The solvent concentration for the j-th sample is... Where is the viscosity activation energy, and R is the gas constant. Let be the temperature of the j-th sample.

8. A simulation and quality prediction system for lyocell fiber spinning sizing process, characterized in that, The lyocell fiber spinning sizing process simulation and quality prediction system includes: a data acquisition device, a data output device, a processor, and a storage device. The storage device includes a computer-readable storage medium storing a computer program. The computer program includes program instructions, which, when executed by the processor, cause the processor to implement the lyocell fiber spinning sizing process simulation and quality prediction method as described in any one of claims 1-7.

Images