Multi-process multi-target collaborative optimization control method and device for silk production line

Abstract

The invention discloses a multi-process multi-objective collaborative optimization control method and device for a silk making production line, and the method comprises the steps: obtaining process parameters, outlet moisture process values and outlet moisture content data of different process flows during silk making, constructing a moisture chain data set of the silk making production line, and enabling the process flows to comprise a main processing process and an auxiliary connection process; preprocessing the process parameters, the outlet moisture process value and the outlet moisture content data to obtain a moisture chain data set of the cut tobacco production line; based on the data set, respectively constructing a moisture prediction model of a main processing procedure and a moisture dissipation model of an auxiliary connection procedure to form a whole-line moisture chain prediction model; the method has the beneficial effects that prediction models of four main procedures of loosening and moisture regaining, leaf moistening and feeding, cut tobacco drying and blending and perfuming are connected in series with moisture dissipation models of auxiliary connection procedures such as premixing and leaf storage, and the transfer and change process of moisture in a whole production line is accurately described.

Description

Technical Field

[0001] This invention relates to the field of optimizing the moisture content at the outlet of silk production, and particularly to a multi-process, multi-objective collaborative optimization control method and device for silk production lines. Background Technology

[0002] The production process of plant fibers is a crucial step in processing plant raw materials into plant fibers suitable for textiles or other applications. This process typically includes multiple steps, such as loosening and rehydration, premixing, leaf moistening and feeding, leaf storage, shaving, drying, air separation for impurities, and scenting. Moisture content is a vital indicator for evaluating the quality of plant fibers, and it is present throughout the entire production process. According to existing process standards, the outlet moisture content of plant fibers is usually checked during four key steps: loosening and rehydration, feeding, drying, and scenting.

[0003] When examining the moisture content at the outlet of each key process, the main focus is on detecting the difference between the actual outlet moisture content and the set outlet moisture content. The closer the actual outlet moisture content is to the set outlet moisture content, the better the control effect of that process is considered. The set outlet moisture content is generally determined by the technical personnel responsible for the process after extensive theoretical analysis and experimental verification, based on factors such as raw material characteristics, process requirements, equipment capabilities, and product quality. The core logic is to "ensure that the moisture content of each process not only meets its own process requirements but also lays a stable foundation for downstream processes."

[0004] Normally, during production, operators only need to control the moisture content according to a pre-defined outlet moisture content setting to ensure the actual outlet moisture content is close to the set value. However, in actual production, raw material characteristics, equipment capacity, and environmental conditions frequently change, which may cause the pre-defined outlet moisture content to differ from actual production needs. In this case, continuing to control according to the original setting may lead to a decline in the quality of the entire batch of plant fiber leaves.

[0005] To address this issue, re-determining the outlet moisture content setpoints for each process step using traditional process parameter optimization methods typically requires significant manpower, resources, and time, which is clearly insufficient to meet production demands. Furthermore, upstream processes influence downstream processes; therefore, adjusting the outlet moisture content setpoint for a single process may negatively impact the production efficiency of other processes. Therefore, it is essential to consider the entire production chain, taking into account the interrelationships between processes, and comprehensively optimize the outlet moisture content setpoints for each step.

[0006] Limitations of existing technology:

[0007] Currently, the process setpoints for the outlet moisture content of key processes in most plant fiber and leaf production lines need to be adjusted appropriately with the season or year. Typically, these adjustments are made by process engineers setting new standard values ​​monthly, quarterly, or annually based on past control data. However, changes in environmental factors are often short-lived and unpredictable, and traditional process parameter adjustment methods are lagging and cannot quickly respond to real-time changes.

[0008] In addition, there are currently adaptive prediction and optimization methods for batch processing parameters of plant silk leaves based on reverse optimization. This method starts with the final flavoring outlet moisture setting value before the start of each batch production, and then adjusts the outlet moisture setting values ​​of each major process step by step in reverse. Although this method can optimize the moisture setting values ​​of key processes to some extent, its reverse optimization process cannot be adjusted based on real-time production data, often resulting in an inability to effectively guide subsequent processes based on the actual situation of completed processes, which may ultimately lead to significant errors. Furthermore, this optimization method typically only focuses on the moisture control of the flavoring process, ignoring the mutual influence between processes, thus easily generating local optima and ultimately leading to poor overall results. Summary of the Invention

[0009] Purpose of the invention: The purpose of this invention is to solve the technical problems in the prior art and provide a multi-process, multi-objective collaborative optimization control method and device for silk production lines.

[0010] Technical solution:

[0011] Firstly, this application proposes a multi-process, multi-objective collaborative optimization control method for a silk-making production line, including:

[0012] The process parameters, outlet moisture content, and outlet moisture content data of different processes during silk production are obtained to construct a moisture chain dataset for the silk production line. The process flow includes main processing steps and auxiliary processes.

[0013] The process parameters, outlet moisture process values, and outlet moisture content data are preprocessed to obtain a moisture chain dataset for the silk production line.

[0014] Based on the dataset, moisture prediction models for the main processing steps and moisture dissipation models for the auxiliary processes are constructed to form a moisture chain prediction model for the entire production line.

[0015] Based on the whole-line moisture chain prediction model, the process value of outlet moisture is optimized by multi-objective optimization based on evolutionary algorithm to obtain the optimized outlet moisture setpoint.

[0016] The optimized outlet moisture setting value is then sent to the stand-alone control system of each process.

[0017] Preferably, the main processing steps include loosening and rehydrating, moistening and adding materials, drying and shredding, and blending and adding flavoring.

[0018] The auxiliary processes include premixing, leaf storage, shredding, and air separation for impurity removal.

[0019] Preferably, the process parameters include adjustable key process control parameters and non-adjustable control parameters;

[0020] The adjustable key process control parameters include material flow rate, water addition, ambient temperature, ambient humidity, compensating steam valve opening, hot air velocity, drum wall temperature, exhaust damper opening, primary damper opening, secondary damper opening, inlet material flow rate, and hot air temperature for each process.

[0021] The non-adjustable control parameters include the inlet moisture content, premixing time, leaf storage time, shredding time, and air separation time for each process.

[0022] Preferably, the moisture chain dataset obtained from the silk production line includes:

[0023] Gaussian mixture model was used to remove outliers, linear interpolation was used to fill in missing values, and moving average was used to smooth the data.

[0024] The process parameters are used to obtain relevant parameters of the equipment and environmental datasets for each process.

[0025] The main processing process dataset is obtained by combining process equipment parameters and environmental datasets.

[0026] The auxiliary process dataset is obtained by combining process equipment parameters and environmental datasets.

[0027] The export moisture content data dataset was obtained by using export moisture content data.

[0028] Preferably, based on the dataset, a moisture prediction model for the main processing steps is constructed, including:

[0029] Real-time prediction models for the moisture content at the outlet of the main processing steps were constructed using multiple linear regression, gradient boosting decision tree model, and mechanistic model, respectively.

[0030] The separate moisture prediction models for each process are constructed as follows:

[0031] ;

[0032] Wherein, the Y i K represents the outlet moisture content of the main processing steps. ij C corresponds to the key adjustable process control parameters in production. i Y represents the non-regulating control parameter in production. iK ij C i The data all come from datasets of the main processing steps;

[0033] Among them, the main processing steps dataset is distinguished according to the number of the plant silk leaves produced and processed, and an export moisture content prediction model is constructed for different main processing steps with different numbers.

[0034] After integrating the prediction models for the outlet moisture content of each major processing step, it can be expressed as:

[0035] ,

[0036] in, These represent moisture prediction models for loose rehydration, leaf moistening with added ingredients, shredded processing, and blending with added flavoring, respectively.

[0037] Preferably, for the loose rehydration processing, an outlet moisture content prediction model is constructed using a multiple linear regression model.

[0038] For the leaf feeding process, a gradient boosting decision tree model is used to construct an outlet moisture content prediction model;

[0039] For the wire drying process, a mechanism model is used to construct an outlet moisture content prediction model;

[0040] For the flavoring process, a multiple linear regression model was constructed to predict the moisture content at the export point.

[0041] Preferably, based on the dataset, a moisture dissipation model for the auxiliary process is constructed, including:

[0042] The critical point at which leaf filaments and the environment reach water balance was calculated using the following formula:

[0043] ;

[0044] in, These are the characteristic constants of the filaments; different filament designations result in different characteristics. The absolute temperature of the environment. The relative humidity of the environment;

[0045] The exponential decay model is derived from the critical point at water balance, including:

[0046] ;

[0047] in, Let be the dissipation rate constant. For dissipation time; The moisture content of the silk leaves at the start of dissipation. The moisture content of the silk leaf is t after time t following the onset of dissipation;

[0048] A dissipation model for the auxiliary process is constructed using a dissipation model, including:

[0049] ;

[0050] These represent the prediction models for the premixing, leaf storage, shredding, and air separation processes, respectively.

[0051] Preferably, the optimized outlet moisture setpoint is obtained by performing multi-objective optimization based on an evolutionary algorithm on the target value of outlet moisture content using a whole-line moisture chain prediction model, including:

[0052] Construct a multi-objective optimization problem with the following fitness function:

[0053] ;

[0054] Wherein, X represents the decision variables in the main processing steps, including but not limited to the amount of water added for loosening and rehydration, the opening degree of the compensating steam valve for leaf moistening, and the hot air velocity for drying the fibers. These are the predictive export moisture models representing the loosening and rehydration, adding of ingredients, drying of silk, and flavoring processes, respectively. These represent the export moisture values ​​for loose rehydration, added ingredients, dried silk, and flavored products, respectively.

[0055] The optimization problem is solved by repeatedly applying a multi-objective optimization method based on evolutionary algorithms until convergence and the Pareto front is output.

[0056] The optimal solution is obtained from the output Pareto front, and the optimal outlet moisture setting value is obtained respectively.

[0057] Preferably, the optimization problem is solved repeatedly using a multi-objective optimization method based on evolutionary algorithms until the Pareto front is reached after a certain number of iterations or convergence, including:

[0058] Step 101: Initialize the population: Randomly generate an initial population X containing N solutions, i.e., the decision variables, forming the initial solution set. It is an individual within a population;

[0059] Step 102: Calculate the fitness of individuals in the population: Substitute each individual in the population into the fitness function to calculate the fitness, and obtain the fitness of each individual;

[0060] Step 103, Non-dominated ranking: Based on the individual fitness calculated in the previous step, individuals in the population are stratified according to dominance relationship to construct Pareto layers of different levels;

[0061] Step 104: Crowding Calculation: Calculate the crowding degree of individuals within each level to measure the density of individuals within the same level;

[0062] Step 105: Selection Operation: Select parent individuals based on different Pareto levels and crowding levels;

[0063] Step 106, Crossover Operation: Crossover and recombination are performed on the parent individuals to generate new individuals;

[0064] Step 107, Mutation Operation: Randomly mutate individuals to maintain population diversity;

[0065] Step 108: Generate a new population: Merge the parent individuals and the offspring individuals obtained after crossover and mutation, and re-select them, retaining the top few best individuals;

[0066] Step 109: If the iterative algebra or the preset convergence condition is met, stop the calculation and output the Pareto front.

[0067] Otherwise, repeat steps 102-109.

[0068] Preferably, the optimal solution is obtained from the output Pareto front to determine the optimized outlet moisture setpoint, including:

[0069] Selecting the optimal solution from the Pareto front employs a hybrid strategy, which includes a combination of at least two of the following methods: weighting method, ideal point distance method, constraint method, and priority ranking method.

[0070] The optimal export moisture setting value is obtained through the optimal solution, including:

[0071] , , , ;

[0072] in, , , , These are the optimized outlet moisture settings for loosening and rehydration, adding ingredients, drying the fibers, and adding fragrance, respectively. This is the optimal solution. The optimal solution is used to input the predicted export moisture content model representing the loose rehydration, material addition, drying, and flavoring processes.

[0073] The second part of this invention provides a multi-process, multi-objective collaborative optimization control device for a silk-making production line, comprising:

[0074] The acquisition unit is used to acquire process parameters, outlet moisture process values ​​and outlet moisture content data of different processes during silk production, and to construct a moisture chain dataset for the silk production line. The process includes main processing steps and auxiliary processes.

[0075] The processing unit is used to preprocess the process parameters, outlet moisture process value and outlet moisture content data to obtain the moisture chain dataset of the silk production line.

[0076] The generation unit is used to construct moisture prediction models for the main processing steps and moisture dissipation models for the auxiliary processing steps based on the dataset, forming a moisture chain prediction model for the entire production line.

[0077] The optimization unit is used to perform multi-objective optimization of the outlet moisture process value based on the whole line moisture chain prediction model to obtain the optimized outlet moisture setpoint.

[0078] The generating unit is used to send the optimized outlet moisture setting value to the stand-alone control system of each process.

[0079] Thirdly, embodiments of the present invention provide an electronic device, including a processor and a memory. The memory stores one or more computer programs; when the one or more computer programs stored in the memory are executed by the processor, the electronic device is able to implement any of the possible design methods described in the first aspect.

[0080] Fourthly, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method as described in any of the above embodiments.

[0081] Fifthly, embodiments of the present invention also provide a computer program product that, when run on an electronic device, causes the electronic device to perform any possible design method of any of the above aspects.

[0082] Beneficial effects:

[0083] Traditional methods control moisture independently in each process, neglecting the coupling effects between processes. This invention constructs a "moisture chain prediction model," linking prediction models for four main processes—loosening and rehydration, leaf moistening and feeding, drying, and blending and flavoring—with moisture dissipation models for auxiliary processes such as premixing and leaf storage. This accurately describes the transfer and change of moisture throughout the entire production line. Based on this, a multi-process joint optimization model (optimizing the final moisture content of the entire production line as the global objective, rather than pursuing the optimality of a single process) overcomes the drawback of "local optima, overall poor performance," significantly improving the uniformity and stability of moisture content in the finished plant fiber leaves.

[0084] Traditional cyclical parameter adjustments relying on human experience suffer from significant lag in response. A core aspect of this invention's technical solution lies in "rolling optimization before and during production." Before production, the model can utilize historical data for pre-optimization, proactively addressing known environmental changes. During production, the system can access real-time data, dynamically adjusting and optimizing setpoints to promptly respond to short-term fluctuations in raw materials, the environment, and other factors. This real-time adaptive capability enables the production system to operate under optimal or near-optimal conditions at all times, significantly improving the production line's adaptability to dynamic production conditions.

[0085] This invention abandons the traditional parameter determination method that relies on a large number of physical experiments and theoretical analysis, and instead adopts data fusion and machine learning models. Gaussian mixture models are used to handle outlier and missing data, and high-precision moisture prediction models are constructed for different process characteristics (such as multiple linear regression, GBDT, and mechanistic models), providing a reliable foundation for optimization. Subsequently, multi-objective optimization based on evolutionary algorithms automatically searches for Pareto optimal solutions in the solution space, significantly reducing the number of manual adjustments and physical experiments required to find optimal parameters, significantly saving manpower, material resources, and time costs, and improving the efficiency and accuracy of process optimization.

[0086] This technical solution goes beyond mere optimization calculations, forming a complete closed loop of "perception-prediction-optimization-control." The system directly sends the optimal outlet moisture setpoint, obtained through multi-process joint optimization, to the individual machine automatic moisture control model of each process. This allows the global optimization decision to directly drive the adaptive adjustment of the underlying equipment, ensuring the precise execution of the optimization results. Ultimately, it achieves full-process intelligence from decision-making to execution, improving the intelligent manufacturing level of the entire plant fiber and leaf production line. Attached Figure Description

[0088] Figure 1 A schematic diagram of the method framework for this invention is provided;

[0089] Figure 2 This invention provides a schematic diagram of the specific framework of the method.

[0090] Figure 3 This invention provides a schematic diagram of the method model fusion process;

[0091] Figure 4 This invention provides a schematic flowchart of a multi-objective optimization method for evolutionary algorithms;

[0092] Figure 5 This is a block diagram of a device structure provided in one embodiment of this application;

[0093] Figure 6 This is a block diagram of an electronic device structure provided in one embodiment of this application. Detailed Implementation

[0095] To make the technical solution of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0096] Example 1

[0097] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions in the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without inventive effort are within the scope of protection of this invention. Unless otherwise defined, the technical or scientific terms used herein should have the ordinary meaning understood by those skilled in the art. The terms "comprising" and similar expressions used herein mean that the element or object preceding the word covers the element or object listed following the word and its equivalents, but do not exclude other elements or objects.

[0098] In response to the problems existing in the current technology, such as Figure 1-4 As shown, a multi-process, multi-objective collaborative optimization control method for silk production lines is proposed, including:

[0099] Step 101: Obtain the process parameters, outlet moisture process value and outlet moisture content data of different processes during silk production, and construct a moisture chain dataset for the silk production line. The process includes the main processing process and the auxiliary process.

[0100] In some specific embodiments, the main processing steps include loosening and rehydrating, moistening and adding materials to the leaves, drying the shredded leaves, and blending and adding fragrance.

[0101] The auxiliary processes include premixing, leaf storage, shredding, and air separation for impurity removal.

[0102] In some specific embodiments, the process parameters include adjustable key process control parameters and non-adjustable control parameters;

[0103] The adjustable key process control parameters include material flow rate, water addition, ambient temperature, ambient humidity, compensating steam valve opening, hot air velocity, drum wall temperature, exhaust damper opening, primary damper opening, secondary damper opening, inlet material flow rate, and hot air temperature for each process.

[0104] The non-adjustable control parameters include the inlet moisture content, premixing time, leaf storage time, shredding time, and air separation time for each process.

[0105] The process names, related process parameters, and their corresponding relationships are as follows:

[0106] (1) The material flow rate X corresponding to loose rehydration 11 Inlet moisture content X 12 Water addition X 13 Hot air temperature X 14 ; Outlet moisture content Y1; Outlet moisture process value Q1;

[0107] (2) Premixing corresponds to premixing time X 21 Ambient temperature and humidity X 22 X 23 Export moisture content Y2;

[0108] (3) Material flow rate X corresponding to leaf feeding 31 Inlet moisture content X 32 Compensating steam valve opening degree X 33 Hot air temperature X 34 Export moisture content Y3; Export moisture process value Q2;

[0109] (4) Storage time X corresponding to the stored leaves 41 Ambient temperature and humidity X 42 X 43 Export moisture content Y4;

[0110] (5) Shredding time X 51 Ambient temperature and humidity X 52 X 53 Export moisture content Y5;

[0111] (6) Material flow rate X corresponding to the drying of wire 61 Inlet moisture content X 62 Hot air temperature X 63 Hot air speed X 64 Drum wall temperature X 65 Deflector opening X 66 Export moisture content Y6; Export moisture process value Q3;

[0112] (7) The ambient temperature and humidity X corresponding to air separation for impurity removal 71 X 72 Wind separation processing time X 73 Level 1 and Level 2 damper opening X 74 X 75 Export moisture content Y7;

[0113] (8) The ambient temperature and humidity X corresponding to the blending and addition of fragrance 81 X 82 Export moisture content Y8; Export moisture process value Q4.

[0114] Step 102: Preprocess the process parameters, outlet moisture process value and outlet moisture content data to obtain the moisture chain dataset of the silk production line;

[0115] In some specific embodiments, Gaussian mixture models are used to remove outlier data, linear interpolation is used to fill in missing values, and moving averages are used for data smoothing.

[0116] Gaussian Mixture Model (GMM) is a probabilistic model that assumes the data is composed of multiple Gaussian distributions. In multi-process data of yarn production, due to the complexity of the processes and the diversity of data sources (such as process parameters, environmental temperature and humidity), the data may exhibit a multimodal distribution (i.e., multiple peaks or clusters). GMM can effectively model the entire production line data, identify potential sub-distributions in the data, and thus understand the overall structure of the data.

[0117] Production data may contain outliers (such as data points caused by sensor malfunctions or operational errors), which can distort model training and lead to inaccurate predictions. In real-time acquired data, some variables may be missing due to transmission interruptions or recording errors. Missing data results in incomplete model training, affecting prediction accuracy. For time series data (such as continuously acquired process parameters), missing values ​​may appear at consecutive time points. Linear interpolation is a simple and efficient imputation method, assuming that the data changes linearly over a short period. Production data often contains random fluctuations or noise (such as sensor jitter or transient environmental interference). Smoothing aims to reduce these fluctuations and make the data more stable (i.e., statistical properties such as mean and variance are stable over time).

[0118] The process parameters are used to obtain relevant parameters of the equipment and environmental datasets for the process, including:

[0119] ;

[0120] in, Indicates the first stage of the silk production line One process, Indicates the first The first step of the process A key process parameter that can be adjusted. This represents the outlet moisture content of the i-th main processing step. This represents the process moisture content at the outlet of the i-th process.

[0121] The main processing steps dataset is obtained by combining process equipment parameters and environmental datasets, including;

[0122] ;

[0123] The auxiliary process dataset is obtained by combining process equipment-related parameters with environmental datasets, including;

[0124] ;

[0125] The export moisture content dataset was obtained by analyzing the export moisture content data. Four main processes have quality indicators (target values ​​for export moisture content), specifically for the loosening and rehydration, leaf moistening and feeding, drying, and leaf flavoring processes. The export moisture content process values ​​are expressed as follows:

[0126] ;

[0127] Step S103: Based on the dataset, construct moisture prediction models for the main processing steps and moisture dissipation models for the auxiliary processing steps respectively to form a moisture prediction model for the entire production line.

[0128] Based on the dataset, a moisture prediction model for the main processing steps is constructed, including:

[0129] Real-time prediction models for the moisture content at the outlet of the main processing steps were constructed using multiple linear regression, gradient boosting decision tree model, and mechanistic model, respectively.

[0130] The separate moisture prediction models for each process are constructed as follows:

[0131] ;

[0132] Wherein, the Y i K represents the outlet moisture content of the main processing steps. ij This corresponds to key adjustable process control parameters in production, such as material flow rate X. 11 C i This represents non-regulated control parameters in production, such as inlet moisture content X. 12 Y i K ij C i The data all come from the main processing steps dataset D1;

[0133] Among them, the main processing steps dataset is distinguished according to the number of the plant silk leaves produced and processed, and an export moisture content prediction model is constructed for different main processing steps with different numbers.

[0134] After integrating the prediction models for the outlet moisture content of each major processing step, it can be expressed as:

[0135] ;

[0136] in, These represent moisture prediction models for loose rehydration, leaf moistening with added ingredients, shredded processing, and blending with added flavoring, respectively.

[0137] In some specific embodiments, for the loose rehydration process, a multiple linear regression model is used to construct an outlet moisture content prediction model, with the outlet moisture content as the prediction index.

[0138] For the leaf feeding process, a gradient boosting decision tree model is used to construct an outlet moisture content prediction model, with the outlet moisture content as the prediction index.

[0139] For the wire drying process, a mechanism model is used to construct an outlet moisture content prediction model, with the prediction index being the outlet moisture content of the wire drying process;

[0140] For the flavoring process, a multiple linear regression model was constructed to predict the moisture content at the outlet, with the moisture content at the outlet of the dried shredded product being the prediction index.

[0141] Specifically, for a single process, the outlet moisture content is determined jointly by the process parameters of the current process and the outlet moisture content of the preceding process. Therefore, a separate moisture prediction model can be constructed for each process, and the overall prediction of moisture changes can be achieved through the mass transfer process of the moisture chain.

[0142] In some specific embodiments, based on the dataset, a moisture dissipation model for the auxiliary process is constructed, including:

[0143] The critical point at which leaf filaments and the environment reach moisture balance was calculated, and its value is determined by the ambient temperature and humidity, as shown in the following formula:

[0144] ;

[0145] in, These are the characteristic constants of the filaments; different filament designations result in different characteristics. The absolute temperature of the environment. The relative humidity is the ambient temperature, because most of the blade production environment is constant temperature and humidity. Therefore, the average ambient temperature and humidity during the dissipation process can be used to represent the ambient temperature and humidity of the entire dissipation process.

[0146] After obtaining the critical value for water balance, the change in leaf filament water content over time can be represented by an exponential decay model. This exponential decay model is derived from the critical point at water balance, including:

[0147] ;

[0148] in, Let be the dissipation rate constant. For dissipation time; The moisture content of the silk leaves at the start of dissipation. The moisture content of the silk leaf is t after time t following the onset of dissipation;

[0149] A dissipation model for the auxiliary process is constructed using a dissipation model, including:

[0150] ;

[0151] These represent prediction models for the premixing, leaf storage, shredding, and air separation processes, respectively, predicting the changes in leaf moisture content between the main processes.

[0152] Specifically, in the different processes of the entire silk production line, apart from loosening and rehydrating, moistening and adding materials, drying silk, and blending and adding fragrance, the remaining premixing, leaf storage, silk cutting, and air separation for impurity removal are collectively referred to as auxiliary processes. The moisture change process in auxiliary processes is mainly a moisture dissipation process under natural conditions. Therefore, in addition to the dissipation time, ambient temperature and humidity are also important factors affecting the moisture content in auxiliary processes.

[0153] The process of moisture dissipation in the silk leaves can be simplified as follows: "Moisture inside the silk leaves migrates to the surface through diffusion, then evaporates from the surface into the surrounding air, eventually achieving a moisture balance with the environment."

[0154] The export moisture prediction model obtained from the main processing steps and

[0155] Combined; construct a whole-line moisture chain prediction model, with the structure as follows: Figure 3 That is, each process model (f1-f8) is set in a different process.

[0156] In some embodiments, when optimizing during non-production periods, the average values ​​of the first ten production batches of the same brand are used, except for the loose rehydration water addition, loose rehydration inlet moisture, and process-related parameters for premixing, leaf storage, shredding, and air separation.

[0157] When optimizing the production time period, real-time data is used for the inlet moisture content of loose rehydration and the process-related parameters of premixing, leaf storage, shredding, and air separation for impurity removal.

[0158] By sequentially connecting the main processing steps and auxiliary steps according to the technological order, the transfer of moisture in the tobacco shreds between different steps can be achieved, integrating individual optimization tasks into a holistic multi-objective optimization task. This allows for the optimization of a single step while simultaneously considering multiple other steps, achieving synergistic optimization among different processes.

[0159] Step S104: Based on the whole line moisture chain prediction model, the outlet moisture process value is optimized by multi-objective optimization based on evolutionary algorithm to obtain the optimized outlet moisture setpoint.

[0160] Since the outlet moisture setpoints for four processes need to be calculated simultaneously, a multi-objective optimization method based on an evolutionary algorithm is adopted. Because evolutionary algorithms search for multiple solutions based on a population, they naturally allow for parallel exploration of diverse solutions. The optimization process can be represented as follows:

[0161] ;

[0162] in, It is a decision variable and belongs to the feasible region. ; Represents the fitness function. There are m objective functions, which are usually in conflict with each other. The solutions obtained in this way typically do not have an "optimal solution," but only a set of "unmanageable solutions," i.e., a "Pareto front." However, for the multi-process joint optimization problem of the silk-making production line in this application, the above problem can be defined as follows:

[0163] Construct a multi-objective optimization problem with the following fitness function:

[0164] ;

[0165] Where X is the decision variable in the main processing steps, and the decision variables include, but are not limited to, the amount of water added for loosening and rehydration. The opening degree of the compensating steam valve for leaf feeding and the hot air speed for drying the silk Wait, that is , These are the predictive export moisture models representing the loosening and rehydration, adding of ingredients, drying of silk, and flavoring processes, respectively. The process values ​​for the exit moisture content after loose rehydration, adding ingredients, drying, and adding fragrance are respectively represented; the absolute value of the difference between the predicted exit moisture content and the set exit moisture content for each major processing step is the objective function.

[0166] In actual production, operators primarily control the moisture content of loosening and rehydration, feeding, and drying by adjusting the amount of water added, the opening of the compensating steam valve for leaf moistening, and the hot air velocity for drying. Therefore, the values ​​of these three variables can be used as the final solution, as follows: Figure 4 As shown.

[0167] The optimization problem is solved by repeatedly applying a multi-objective optimization method based on evolutionary algorithms until convergence and the Pareto front is output.

[0168] The optimal solution is obtained from the output Pareto front, and the optimal outlet moisture setting value is obtained respectively.

[0169] In some specific embodiments, the optimization problem is solved repeatedly using a multi-objective optimization method based on evolutionary algorithms until the Pareto front is reached after a certain number of iterations or convergence, including:

[0170] Step 101: Initialize the population: Randomly generate an initial population X containing N solutions, i.e., the decision variables, forming the initial solution set. It is an individual within a population;

[0171] Step 102: Calculate the fitness of individuals in the population: Substitute each individual in the population into the fitness function to calculate the fitness, and obtain the fitness of each individual;

[0172] Step 103, Non-dominated ranking: Based on the individual fitness calculated in the previous step, individuals in the population are stratified according to dominance relationship to construct Pareto layers of different levels;

[0173] Step 104, Crowding Calculation: Calculate the crowding degree of individuals within each level to measure the density of individuals within the same level and ensure the diversity of solutions;

[0174] Step 105: Selection Operation: Select parent individuals based on different Pareto levels and crowding levels, usually using a competitive selection method;

[0175] Step 106, Crossover Operation: Crossover and recombination of parent individuals to generate new individuals, produce new solutions, and explore the solution space;

[0176] Step 107, Mutation Operation: Randomly mutate individuals to maintain population diversity;

[0177] Step 108: Generate a new population: Merge the parent individuals and the offspring individuals obtained after crossover and mutation, and re-select them, retaining the top few best individuals;

[0178] Step 109: If the iterative algebra or the preset convergence condition is met, stop the calculation and output the Pareto front.

[0179] Otherwise, repeat steps 102-109.

[0180] In some specific embodiments, the optimal solution is obtained from the output Pareto front to determine the optimized outlet moisture setpoint, including:

[0181] Selecting the optimal solution from the Pareto front employs a hybrid strategy, combining at least two of the following methods: weighting, ideal point distance, constraint, and priority ranking. This is because the Pareto front represents a set of solutions, and each process can only be controlled using at most one set of solutions simultaneously. When setting the selection strategy, a hybrid approach combining multiple selection methods is generally adopted. This hybrid strategy integrates the weighting, ideal point distance, constraint, and priority ranking methods, allowing for the selection of the optimal solution.

[0182] The optimal export moisture setting value is obtained through the optimal solution, including:

[0183] , , , ;

[0184] in, , , , These are the optimized outlet moisture settings for loosening and rehydration, adding ingredients, drying the fibers, and adding fragrance, respectively. This is the optimal solution. The optimal solution is used to input the predicted export moisture content model representing the loose rehydration, material addition, drying, and flavoring processes.

[0185] Specifically, the outlet moisture content values ​​for the four main production processes—loosening and rehydration, feeding, drying, and flavoring—are as follows: Generally, only the loosening and rehydration, feeding, and drying processes in production have the ability to actively regulate the outlet moisture content. The outlet moisture content of the flavoring process is affected by the combined effects of the first three processes. To ensure the actual outlet moisture content of each major process in production... All should be as close as possible to the process values. The optimal outlet moisture setpoint can be obtained by combining a moisture chain prediction model with a multi-objective optimization algorithm. Then, the optimized outlet moisture setpoint is sent to each individual control model, ultimately achieving optimized moisture control for the entire silk-making line.

[0186] Step S105: Send the optimized outlet moisture setting value to the stand-alone control system of each process.

[0187] A model for predicting export moisture content during the loosening, rehydration, adding, drying, and flavoring processes.

[0188] Combination Figure 5 This invention provides a multi-process, multi-objective collaborative optimization control device for a silk production line, comprising:

[0189] A single unit 201 is used to acquire process parameters, outlet moisture process values ​​and outlet moisture content data of different processes during silk production, and to construct a moisture chain dataset for the silk production line. The process includes main processing steps and auxiliary processes.

[0190] Processing unit 202 is used to preprocess the process parameters, outlet moisture process value and outlet moisture content data to obtain the moisture chain dataset of the silk production line.

[0191] The generation unit 203 is used to construct a moisture prediction model for the main processing steps and a moisture dissipation model for the auxiliary processing steps based on the dataset, forming a moisture chain prediction model for the entire production line.

[0192] Optimization unit 204 is used to perform multi-objective optimization of the outlet moisture process value based on the whole line moisture chain prediction model to obtain the optimized outlet moisture setpoint.

[0193] The generating unit 205 is used to send the optimized outlet moisture setting value to the stand-alone control system of each process.

[0194] In other embodiments of the present invention, an electronic device 400 is disclosed, such as... Figure 6 As shown, the electronic device may include: one or more processors 401; a memory 402; a display 403; one or more application programs (not shown); and one or more computer programs 404. These devices can be connected via one or more communication buses 405. The one or more computer programs 404 are stored in the memory 402 and configured to be executed by the one or more processors 401. The one or more computer programs 404 include instructions that can be used to perform actions such as... Figures 1 to 2 And the various steps in the corresponding embodiments.

[0195] Through the above description of the embodiments, those skilled in the art will clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device, and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0196] In the various embodiments of this invention, the functional units can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0197] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiments of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as flash memory, portable hard disk, read-only memory, random access memory, magnetic disk, or optical disk.

[0198] The above description is merely a specific implementation of the embodiments of the present invention, but the protection scope of the embodiments of the present invention is not limited thereto. Any changes or substitutions within the technical scope disclosed in the embodiments of the present invention should be covered within the protection scope of the embodiments of the present invention. Therefore, the protection scope of the embodiments of the present invention should be determined by the protection scope of the claims.

Claims

1. A multi-process multi-target collaborative optimization control method for a cut tobacco production line, characterized in that, The method comprises the following steps: acquiring process parameters, outlet moisture process values and outlet moisture content data of different process flows during silk production, and constructing a moisture chain data set of a silk production line, wherein the process flows comprise main processing processes and auxiliary continuous processes; preprocessing the process parameters, outlet moisture process values and outlet moisture content data to obtain the moisture chain data set of the silk production line; based on the data set, constructing a moisture prediction model of the main processing processes and a moisture dissipation model of the auxiliary continuous processes to form a whole-line moisture chain prediction model; based on the whole-line moisture chain prediction model, performing multi-objective optimization on the outlet moisture process values based on an evolutionary algorithm to obtain optimized outlet moisture set values; issuing the optimized outlet moisture set values to single-machine control systems of each process.

2. The multi-process multi-target collaborative optimization control method for a cut tobacco production line according to claim 1, characterized in that, The main processing processes comprise loose rehydration, leaf moistening and feeding, drying and blending and flavoring processes; The auxiliary continuous processes comprise premixing, leaf storage, cutting and air separation and impurity removal processes; The process parameters comprise adjustable key process control parameters and non-adjustable control parameters; The adjustable key process control parameters comprise material flow, water addition amount, environmental temperature, environmental humidity, compensation steam valve opening, hot air speed, drum wall temperature, moisture removal damper opening, primary air damper opening, secondary air damper opening, inlet material flow and hot air temperature of each process; The non-adjustable control parameters comprise inlet moisture content, premixing time, leaf storage time, cutting processing time and air separation processing time of each process.

3. The multi-process multi-target collaborative optimization control method for a cut tobacco production line according to claim 1, characterized in that, The method for obtaining the moisture chain data set of the silk production line comprises the following steps: performing abnormal data elimination using a Gaussian mixture model, filling missing values using a linear interpolation method, and performing data smoothing processing using a moving average method; obtaining process equipment related parameter and environmental data sets through process parameters, obtaining a main processing process data set through the process equipment related parameter and environmental data sets; obtaining an auxiliary continuous process data set through the process equipment related parameter and environmental data sets; obtaining an outlet moisture content data data set through the outlet moisture content data.

4. The multi-process multi-target collaborative optimization control method of a cut tobacco production line according to claim 3, characterized in that, Based on the data set, a moisture prediction model of the main processing processes is constructed, comprising the following steps: respectively using a multiple linear regression, a gradient boosting decision tree model and a mechanism model to construct an outlet moisture content real-time prediction model of the main processing processes; The following steps are taken to construct a separate moisture prediction model for each process: ； where Y i represents the exit moisture of the main process step, K ij corresponds to the key process control parameters adjustable in production, C i represents the non-adjustable control parameters in production, Y i , K ij , C i Data are all from the main process step data set; The main processing process data set is distinguished according to the plant silk leaf number of production and processing, and an outlet moisture content prediction model is constructed for different main processing processes of different numbers; After integrating the outlet moisture content prediction models of each main processing process, the following expression can be obtained: ； wherein, respectively represent moisture prediction models for loose re-damp, leaf conditioning, cuttage, and blending and flavoring.

5. The multi-process multi-objective collaborative optimization control method for a silk production line according to claim 4, characterized in that: for the loose rehydration processing process, a multiple linear regression model is used to construct an outlet moisture content prediction model, for the leaf moistening and feeding processing process, a gradient boosting decision tree model is used to construct an outlet moisture content prediction model; for the drying processing process, a mechanism model is used to construct an outlet moisture content prediction model; for the blending and flavoring processing process, a multiple linear regression is used to construct an outlet moisture content prediction model.

6. The multi-process multi-target collaborative optimization control method of a cut tobacco production line according to claim 3, characterized in that, Based on the data set, a moisture dissipation model of the auxiliary continuous processes is constructed, comprising the following steps: A critical point is calculated when the cut tobacco and the environment reach a moisture balance, and the formula is as follows: ； wherein, is a filament leaf property constant, different numbering of filament leaf properties are different, is the absolute temperature of the environment, is the relative humidity of the environment; An exponential decay model is obtained through the critical point when the moisture balance is reached, including: ； wherein, is the dissipation rate constant, is the dissipation time; is the moisture content of the filament leaf at the beginning of dissipation, is the moisture content of the filament leaf after t time has elapsed after the beginning of dissipation; A dissipation model is constructed for the auxiliary connecting process, including: ； Predictive models for premixing, leaf storing, shredding, and impurity removal, respectively.

7. The multi-process multi-target collaborative optimization control method of a primary processing line according to claim 3, characterized in that, Based on the whole-line moisture chain prediction model, the outlet moisture content target value is optimized by using the evolutionary algorithm-based multi-objective optimization to obtain the optimized outlet moisture setting value, including: A multi-objective optimization problem is constructed, and the fitness function is: ； Wherein, X is the decision variable in the main processing procedure, the decision variable includes but is not limited to the water adding amount of loose re-damping, the compensation steam valve opening of leaf moistening, and the hot air speed of drying, Respectively are the predicted outlet moisture models representing the loose re-damping, leaf moistening, drying, and flavoring procedures, Respectively represent the outlet moisture process values of loose re-damping, leaf moistening, drying, and flavoring. The evolutionary algorithm-based multi-objective optimization method is repeatedly used to solve the optimization problem until the Pareto front is output; The optimal solution is obtained in the output Pareto front to obtain the optimized outlet moisture setting value.

8. The multi-process multi-target collaborative optimization control method of a cut tobacco production line according to claim 7, characterized in that, The evolutionary algorithm-based multi-objective optimization method is repeatedly used to solve the optimization problem until the iteration number or the convergence of the Pareto front is reached, including: Step 101, initialize population: randomly generate an initial population X containing N solutions, i.e. decision variables, to form an initial solution set, is an individual in the population; Step 102, calculating the fitness of the population: each individual in the population is put into the fitness function to calculate the fitness, and the fitness of each individual is obtained; Step 103, non-dominated sorting: according to the individual fitness calculated in the last step, the individuals in the population are layered according to the dominance relationship, and different levels of Pareto layers are constructed; Step 104, crowding degree calculation: calculate the crowding degree of individuals in each level, measure the density of individuals in the same layer; Step 105, selection operation: select parent individuals according to different Pareto levels and crowding degrees; Step 106, crossover operation: crossover and recombination of parent individuals to generate new individuals; Step 107, mutation operation: randomly mutate the individuals to maintain the diversity of the population; Step 108, generate new population: combine the parent individuals and the offspring individuals obtained after crossover and mutation, and reselect to retain the top several optimal individuals; Step 109, if the iteration number or the preset convergence condition is reached, stop calculating and output the Pareto front; otherwise, repeat steps 102-109.

9. The multi-process multi-target collaborative optimization control method of a cut tobacco production line according to claim 7, characterized in that, The optimal solution is obtained in the output Pareto front to obtain the optimized outlet moisture setting value, including: The optimal solution is obtained by selecting the optimal solution from the Pareto front using a hybrid strategy, including a combination of at least two methods of weight method, ideal point distance method, constraint method and priority sorting method; The optimized outlet moisture setting value is obtained through the optimal solution, including: 、 、 、 ； wherein, , , , are the optimized outlet moisture set values for loose conditioning, feeding, cutting, and flavoring, respectively, is the optimal solution, are the optimal solution brought into the predicted outlet moisture models representing the loose conditioning, feeding, cutting, and flavoring processes, respectively.

10. A multi-process, multi-objective collaborative optimization control device for a silk-making production line, characterized in that, including: An acquisition unit is configured to acquire process parameters, outlet moisture process values and outlet moisture content data of different process flows during tobacco processing, and construct a tobacco processing line moisture chain dataset, wherein the process flows include main processing processes and auxiliary connecting processes; A processing unit is configured to preprocess the process parameters, outlet moisture process values and outlet moisture content data to obtain a tobacco processing line moisture chain dataset; A generation unit is configured to construct a moisture prediction model for the main processing processes and a moisture dissipation model for the auxiliary connecting processes based on the dataset, and form a whole-line moisture chain prediction model; An optimization unit is configured to perform evolutionary algorithm-based multi-objective optimization on the outlet moisture process values based on the whole-line moisture chain prediction model to obtain an optimized outlet moisture setting value; An occurrence unit is configured to send the optimized outlet moisture setting value to a single-machine control system of each process.

Images