Intelligent Management and Control System and Method for Textile Production Based on Deep Learning

Abstract

The present invention belongs to the field of intelligent production technology, and provides an intelligent control system and method for textile production based on deep learning. The method includes: obtaining the opening cotton process data of a spinning mill that sends a ginning task, and matching the cotton knot content according to the opening cotton process data; performing similarity calculations on the cotton knot content and the type of lint cotton picking in the ginning process database to obtain a number of first ginning process data; using a joint simulation analysis model to perform joint simulation analysis on the cotton knot content and each first ginning process data to obtain the optimal second ginning process data. Relying on the production management platform and using deep learning technology, the present invention realizes the intelligent coordination of the production processes of the spinning mill and the ginning mill, accurately matches the opening cotton and ginning process parameters, and effectively avoids the generation of cotton knots due to excessive cleaning in the ginning mill. Compared with the traditional production method, it reduces the impurity removal cost and energy consumption of the spinning mill, improves production efficiency, and ensures the quality stability of the yarn.

Description

Technical Field

[0001] The present invention relates to the field of smart production technology, and specifically to a deep learning-based intelligent management and control system and method for textile production. Background Art

[0002] In the modernization of the textile industry, ginning mills and spinning mills, as key upstream and downstream links in the supply chain, have a crucial impact on the final quality and production efficiency of textiles. As market demands for yarn quality continue to rise, effectively coordinating the production process parameters of these two mills has become a critical issue that the industry urgently needs to address.

[0003] Currently, significant process integration flaws exist between cotton ginning and spinning mills. To meet traditional lint appearance grading standards, aiming for white, clean, and free of small impurities, cotton ginning mills often over-clean the lint during production, frequently using equipment like airflow lint cleaning. This causes excessive fiber tumbling and a sharp increase in the number of neps. While spinning mills have strengthened their cleaning process by increasing the number of pre-opening and fine-opening equipment, the large number of neps generated by ginning still requires additional manpower and material resources to handle. This not only increases production costs but also reduces production efficiency and affects yarn quality stability.

[0004] Furthermore, existing textile production control systems rely heavily on manual experience or simple rule-based settings, lacking precise control over the complex relationships between ginning and spinning mill process parameters. This makes it impossible to dynamically adjust ginning mill process parameters based on the actual needs of the spinning mill, making it difficult to achieve refined and intelligent management of the production process.

[0005] Therefore, there is an urgent need for a method that can intelligently control the production process parameters of the cotton ginning mill based on the process requirements of the spinning mill to solve the problems existing in the collaborative production of the two. Summary of the Invention

[0006] In this regard, the present invention provides a method, system, electronic device, computer storage medium and computer program product for intelligent management and control of textile production based on deep learning to solve at least one of the above technical problems.

[0007] In a first aspect, the present invention provides an intelligent management and control method for textile production based on deep learning, comprising the following method steps: obtaining the cotton opening process data of the spinning mill that sends the ginning task, and obtaining the cotton nep content according to the matching of the cotton opening process data; performing similarity calculations in the ginning process database according to the cotton nep content and the lint cotton picking type to obtain a number of first ginning process data; wherein the picking types include machine picking and manual picking; using a joint simulation analysis model to perform a joint simulation analysis on the cotton nep content and each of the first ginning process data to obtain the optimal second ginning process data.

[0008] In the second aspect of the present invention, an intelligent management and control system for textile production based on deep learning is provided, characterized in that the system includes a processing module and a storage module, and computer program code is stored in the storage module, and the computer program code is called and executed by the processing module to implement the following steps: obtaining the opening process data of a spinning mill that sends a ginning task, and matching the cotton knot content according to the opening process data; performing similarity calculation in a ginning process database according to the cotton knot content and the type of cotton picking, and obtaining a number of first ginning process data; wherein, the picking types include machine picking and manual picking; using a joint simulation analysis model to perform joint simulation analysis on the cotton knot content and each of the first ginning process data to obtain optimal second ginning process data.

[0009] In the third aspect of the present invention, an electronic device is provided, which includes: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, and when the computer program is executed by the processor, it implements the method described in any one of the foregoing items.

[0010] In the fourth aspect of the present invention, a computer storage medium is provided, which stores a computer program that can be executed by a processor to implement the method described in any one of the foregoing items.

[0011] In the fifth aspect of the present invention, a computer program product is provided, which includes a computer program that can be executed by a processor to implement the method described in any one of the foregoing items.

[0012] Relying on the production management platform and using deep learning technology, the present invention realizes the intelligent coordination of the production processes of spinning mills and ginning mills, accurately matches the opening and ginning process parameters, and effectively avoids the generation of cotton knots due to excessive cleaning in the ginning mill. Compared with the traditional production method, it reduces the impurity removal cost and energy consumption of the spinning mill, improves the production efficiency, and ensures the quality stability of the yarn. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the present invention, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0014] Figure 1 It is a schematic flowchart of an intelligent management and control method for textile production based on deep learning disclosed in an embodiment of the present invention.

[0015] Figure 2It is a schematic flowchart of generating the second ginning process data disclosed in the embodiments of the present invention.

[0016] Figure 3 It is a schematic structural diagram of an intelligent textile production control system based on deep learning disclosed in the embodiments of the present invention. Detailed implementation manners

[0017] The following specific embodiments illustrate the implementation manners of the present application. Those skilled in the art can easily understand other advantages and effects of the present application from the content disclosed in this specification. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0018] In addition, the technical features involved in different implementation manners of the present application described below can be combined with each other as long as they do not conflict with each other.

[0019] As Figure 1 shown, the embodiments of the present invention disclose an intelligent textile production control method based on deep learning, including the following method steps: S10, obtaining the opening process data of the spinning mill that sends the ginning task, and matching the nep content according to the opening process data.

[0020] As Figure 3 shown, the solution of the present invention is applied to a production management platform, which simultaneously accesses the ginning mill and the spinning mill. When there is a demand for cotton raw materials, the spinning mill sends a ginning task to the platform, and the ginning task is configured with the opening process data currently adopted by the spinning mill. The opening process data includes but is not limited to the number of pre-opening channels, the number of fine-opening channels, the rotational speed of the opening machine, the distance between the beater and the dust rod, the air flow pressure parameter, etc.

[0021] It should be noted that the main task of the spinning mill's opening is to remove leaf debris and broken seeds. The opening task includes pre-opening and fine-opening arranged in one or two channels; for the nep problem, as long as the number of neps generated by the ginning mill is not too high, the carding machine of the spinning mill can effectively remove neps through card clothing carding.

[0022] The production management platform matches the nep content according to the received opening process data. Among them, the nep content is mainly determined according to the number of pre-opening and fine-opening channels adopted during the opening process. For example, if one pre-opening and one fine-opening are adopted, a lower nep content is required, while if two pre-openings and two fine-openings are adopted, a higher nep content is allowed.

[0023] S20. Perform similarity calculations in the ginning process database based on the nep content and the type of unginned cotton picking, and obtain a number of first ginning process data. Among them, the picking types include machine picking and manual picking.

[0024] The initial impurity content, fiber state and other characteristics of unginned cotton vary with different picking types (machine picking or manual picking), which will affect the selection of the ginning process. Based on the known nep content requirements and the type of unginned cotton picking, similarity calculations are performed in the ginning process database. The ginning process database stores multiple ginning process data, and each ginning process data corresponds to different picking types, standard ginning process parameters, and their corresponding nep content, impurity removal effect and other data.

[0025] Through similarity calculation algorithms, such as the cosine similarity algorithm, Euclidean distance algorithm, etc., search for multiple ginning process data in the database that are closest to the current nep content requirements and picking types, which are the first ginning process data.

[0026] S30. Use the joint simulation analysis model to perform joint simulation analysis on the nep content and each of the first ginning process data, and obtain the optimal second ginning process data.

[0027] In order to determine the process parameters that best suit the current requirements from a number of first ginning process data, a joint simulation analysis is performed using the joint simulation analysis model, so as to obtain the optimal second ginning process data. The joint simulation analysis model is constructed based on deep learning technology and can comprehensively consider the complex interaction relationships among various factors such as nep content requirements and ginning equipment parameters (such as the rotation speed of the gin, the working intensity of the lint cleaner, etc.) and process operation parameters (such as processing time, air flow control, etc.) included in each of the first ginning process data.

[0028] By simulating and analyzing the above factors, evaluate the impact of different first ginning process data on other quality indicators of unginned cotton (such as fiber damage degree, impurity removal efficiency, etc.) while meeting the nep content requirements. After multiple rounds of simulation and optimization calculations, finally, select the second ginning process data from each of the first ginning process data that can not only meet the nep content requirements of the spinning mill but also ensure the optimal comprehensive quality of unginned cotton, or fit new ginning process parameters based on each of the first ginning process data.

[0029] S40. Control the ginning factory to process the unginned cotton according to the second ginning process data.

[0030] After determining the optimal second ginning process data, the production management platform forwards the second ginning process data to the production equipment and management system of the ginning factory. The equipment in the ginning factory automatically adjusts the operating parameters according to the received second ginning process data, and the operators will also perform production operations according to the new process requirements under the corresponding prompts, so as to achieve precise control of the production process in the ginning factory and ensure that the ginning factory processes the lint according to the process standards that meet the needs of the spinning mill. In this way, the problem of excessive neps caused by excessive cleaning of the lint in the ginning factory can be avoided.

[0031] Relying on the production management platform and using deep learning technology, the present invention realizes the intelligent coordination of the production processes of the spinning mill and the ginning factory, precisely matches the opening and ginning process parameters, and effectively avoids the generation of neps due to excessive cleaning in the ginning factory. Compared with the traditional production method, it reduces the impurity removal cost and energy consumption of the spinning mill, improves the production efficiency, and ensures the quality stability of the yarn.

[0032] As an example, the matching of the nep content according to the opening process data includes: retrieving a mapping model, which is trained by a deep learning algorithm based on the historical production data of the spinning mill; wherein, the historical production data includes the number of pre-opening passes, the number of fine-opening passes, the rotational speed of the opening machine, the interval between the beater and the dust bar, the air flow pressure parameters and their corresponding nep detection data under different opening process configurations; inputting the opening process data into the mapping model to map and obtain the predicted value of the corresponding nep content; wherein, when the opening process data includes two pre-opening passes and two fine-opening passes, the predicted value of the nep content output by the mapping model is higher than that when it includes one pre-opening pass and one fine-opening pass, and the predicted value does not exceed the nep removal threshold of the carding process in the spinning mill.

[0033] The present invention has pre-trained a mapping model based on the historical production data of the spinning mill and a deep learning algorithm. The historical production data includes the number of pre-opening passes, the number of fine-opening passes, the rotational speed of the opening machine, the interval between the beater and the dust bar, the air flow pressure parameters and their corresponding nep detection data under different opening process configurations. After training, this mapping model integrates the above multi-dimensional parameters such as the number of pre-opening passes, the number of fine-opening passes, and the beater speed with the nep detection data, and can capture the non-linear correlation relationship between the above complex parameters. It can be understood that the mapping model is preferably constructed using long short-term memory network LSTM, gradient boosting tree (XGBoost / LightGBM), and the specific construction and training process will not be elaborated here.

[0034] After receiving the opening process data, the opening process data is input into the mapping model, and the mapping model maps the corresponding predicted value of the neps content according to the process configuration. Through the above method of this embodiment, it is possible to reduce the generation of neps and energy consumption caused by excessive cleaning in the ginning mill, and also reduce the load of the carding process in the spinning mill, realizing the coordinated optimization of production efficiency, yarn quality and cost.

[0035] As an example, as Figure 2 shown, the combined simulation analysis model is used to perform combined simulation analysis on the neps content and each of the first ginning process data to obtain the optimal second ginning process data, including: generating a random noise signal and adding it to the first ginning process data corresponding to the manual picking type; the standard deviation of the random noise signal is dynamically adjusted according to the fluctuation range of the impurities in the manually picked cotton in this ginning cycle; through Monte Carlo simulation, random perturbation tests are performed on each of the first ginning process data for more than a preset number of times to generate a parameter space sample, and for each parameter space sample, the corresponding predicted value of the neps content is calculated based on the neps prediction model; clustering analysis is performed on each of the parameter space samples, and the parameter combination with the highest frequency of occurrence and a predicted value of the neps content lower than the threshold is extracted as the initial solution, and an optimization algorithm is used to iteratively optimize the initial solution to output the optimal second ginning process data that meets the end condition.

[0036] Although the ginning process data stored in the database is obtained through the analysis of actual production records, the relevant parameters in the production records may be distorted due to factors such as the fluctuations of the ginning equipment itself and the characteristics of the lint itself (such as cleanliness, fiber quality, etc.). Therefore, in the foregoing step S20, the several similar first ginning process data obtained from the database actually cannot perfectly match the current ginning task, and there may even be a large deviation. Therefore, in this embodiment, these similar first ginning process data are re-fitted and analyzed to obtain the second ginning process data that is most likely to perfectly match the current ginning task.

[0037] At the same time, due to the lack of standardization in the picking process of manually picked cotton, the impurity content fluctuates greatly (the standard deviation can usually reach 2-3 times that of machine-picked cotton), and traditional deterministic process parameter optimization methods are difficult to cope with this uncertainty, easily leading to excessive neps or excessive fiber damage under excessive cleaning.

[0038] For this reason, the present invention first calculates the fluctuation range (expressed by the standard deviation or the range ) of the impurity detection data of the manually picked cotton in this ginning cycle (such as one week, one month, etc.), and generates a random noise signal that conforms to the Gaussian distribution or ( (where the scaling factor is), and superimposed on the first ginning process data to simulate the impurity fluctuations in actual production.

[0039] Next, using Monte Carlo simulation, a random perturbation test of ≥ 1000 times is performed on each piece of the first ginning process data to generate a parameter space sample covering possible parameter combinations and capture the non-linear effects between parameters. Then, a neps prediction model constructed using algorithms such as random forest, neural network, etc. is used to calculate the predicted neps content corresponding to each parameter space sample, that is, a quantifiable quality index of the parameter space sample is obtained. Next, algorithms such as DBSCAN or K-means are used to cluster the parameter space samples, and the parameter combination with the highest frequency of occurrence and a predicted neps content lower than the threshold is extracted as the initial solution. In this way, the parameter combination selection strategy can avoid falling into a local optimum. Finally, heuristic algorithms such as genetic algorithm, particle swarm algorithm, etc. are used to iteratively optimize the above initial solution until the change rate of the objective function value is < 0.1% for 10 consecutive generations, at which point it is determined that the end condition is satisfied, the iteration is terminated, and the optimal parameter combination at this time, that is, the second ginning process data, is output.

[0040] As an example, the random perturbation test of each of the first ginning process data for more than a preset number of times by Monte Carlo simulation to generate a parameter space sample includes: determining the perturbation amplitude and sampling density of the random perturbation according to the impurity fluctuation amplitude of handpicked cotton, and configuring an additional amplitude for the highly sensitive parameters in the first ginning process data on the basis of the perturbation amplitude to obtain the target perturbation amplitude; wherein, the perturbation amplitude is positively correlated with the impurity fluctuation amplitude of handpicked cotton, and the additional amplitude is a preset value; and performing a random perturbation of each of the first ginning process data for a preset number of times according to the perturbation amplitude, the target perturbation amplitude, and the sampling density to generate a parameter space sample.

[0041] Existing methods all use fixed perturbation (for example, the fixed perturbation amplitude is ±5% and the sampling density is 500 times) to perform random perturbation tests. When the impurities fluctuate highly, the fixed small-amplitude perturbation may not cover the effective adjustment range of the parameters, resulting in a conservative optimization result (for example, missing the key solution of "substantially increasing the air flow intensity for high impurities"); while when the impurities fluctuate slightly, the fixed large-amplitude perturbation will generate a large number of invalid samples (for example, the parameters deviate too much from the actual working conditions), wasting computing resources.

[0042] Meanwhile, due to differences in picking batches, cotton field areas, weather conditions, etc. for manually picked cotton, there are significant fluctuations in impurity content (such as leaves, boll shells, dust). For example, cotton picked during the rainy season may contain more mud and impurities due to moisture (large fluctuation range); the impurity content in different plots of the same cotton field may vary by 5% - 10% (local fluctuation). Moreover, when there are many impurities (such as impurity content > 3%), parameters such as the air flow intensity and beater speed of the gin need to be adjusted more precisely, otherwise it is easy to cause a sharp increase in cotton knots; when there are few impurities (such as impurity content < 1.5%), the tolerance space for parameter adjustment is larger, and a small perturbation can cover the effective solution.

[0043] In response to this, the present invention sets the perturbation amplitude of the random perturbation to be related to the impurity fluctuation amplitude of manually picked cotton. Specifically: The impurity fluctuation amplitude of manually picked cotton reflects the uncertainty of the production environment, and the perturbation amplitude needs to be linearly or non-linearly positively correlated with it. Thus, when the impurity fluctuation is large (such as ), the parameters are allowed to be perturbed within a larger range (for example, ±10%) to ensure coverage of the strong coupling relationship between "impurities - parameters" (for example, higher impurity requires a higher beater speed to remove); when the impurity fluctuation is small (for example ), the perturbation range is reduced (for example, ±4%) to avoid the simulation results being distorted due to the parameters deviating from the reasonable range.

[0044] Meanwhile, the sampling density of the random perturbation set by the present invention is also related to the impurity fluctuation amplitude of manually picked cotton. Specifically: The greater the impurity fluctuation, the more dispersed the effective solution space of the parameter combination, and it is necessary to increase the sampling points to improve the coverage probability of the solution, that is, more samples are needed to capture the non-linear relationship between the parameters. For example, when the impurity fluctuation amplitude changes from to , the number of effective parameter combinations may increase by more than 50%, and the sampling density needs to increase from 800 times to 1500 times to ensure statistical significance.

[0045] In addition, the first ginning process data simultaneously includes highly sensitive parameters that affect the formation of cotton knots, such as the frequency of the pre-cleaner, air flow intensity, beater speed, etc. For these highly sensitive parameters, an additional perturbation amplitude, such as 3%, 5%, is added. By expanding the perturbation range of these highly sensitive parameters, the performance of the corresponding first ginning process data under extreme conditions can be verified to ensure that the optimization result is robust to the fluctuations of the highly sensitive parameters. Among them, the highly sensitive parameters can be determined through correlation analysis of historical data, which will not be elaborated herein.

[0046] Finally, after obtaining the above perturbation amplitude, target perturbation amplitude, and sampling density, the Monte Carlo simulation is started to randomly perturb each first ginning process data for a preset number of times, thereby generating a sample of the parameter space.

[0047] It can be understood that the parameter space samples refer to a dataset containing parameter value combinations and their corresponding output results generated after performing random perturbation experiments on the first ginning process data. As shown in Table 1 below:

[0048]

[0049] It should be noted that the above processing method is for the first ginning process data corresponding to manual picking, while for the first ginning process data corresponding to machine picking, a fixed perturbation is directly used for the random perturbation experiment, which will not be elaborated here.

[0050] As an example, the clustering analysis of each of the parameter space samples and extracting the parameter combination with the highest frequency of occurrence as the initial solution includes: performing PCA dimensionality reduction on the parameter space samples, and extracting the principal components with a cumulative contribution rate reaching or exceeding a preset value as new feature vectors; using the DPC algorithm to automatically identify the clustering centers by calculating the local density of sample points and the distance decision graph; using the FCM algorithm to perform secondary optimization on the clustering results of DPC to obtain each clustering cluster; for each clustering cluster, calculating the mean, standard deviation of the process parameters and the confidence interval of the performance index, and removing the abnormal clusters with a standard deviation exceeding the threshold; taking the central parameter combination of the remaining clustering clusters as the preliminary initial solution, and obtaining the initial solution by weighted averaging of each preliminary initial solution.

[0051] The above-mentioned parameter space samples generated by Monte Carlo simulation have high dimensionality, redundancy, and are susceptible to random perturbations and extreme values. Traditional clustering algorithms such as K-means need to preset the number of clusters and are sensitive to the initial centers, making it difficult to handle irregular clusters and the problem of ambiguous membership of boundary samples, resulting in distorted clustering results and low reliability of the initial solution.

[0052] To address the above problems, principal component analysis (PCA) is used to reduce the dimensionality of the parameter space samples, and the principal components with a cumulative contribution rate reaching or exceeding a preset value (≥95%) are extracted as new feature vectors, which can eliminate the correlation between parameters and reduce the computational complexity; then, the density peak clustering algorithm (DPC) is used to automatically identify the clustering centers by calculating the local density of sample points and the distance decision graph, avoiding the limitation of presetting the number of clusters and effectively handling clusters with irregular shapes. At the same time, the fuzzy C-means clustering (FCM) algorithm is introduced to perform secondary optimization on the clustering results of DPC, calculating the membership degree of sample points to different clusters, and solving the problem of ambiguous membership of boundary samples, thereby obtaining multiple clustering clusters.

[0053] For each cluster, calculate the standard deviation of the process parameters. If it exceeds 20% of the mean, it is determined as an unstable cluster (e.g., caused by noise or extreme disturbances). Calculate the confidence interval (e.g., 95% CI) of the performance indicators (nep content, fiber damage rate), and eliminate the clusters with excessive fluctuations. Combine the central parameters of the remaining clusters as the preliminary initial solutions, and obtain the initial solution by weighted averaging of the preliminary initial solutions, making the initial solution closer to the optimal region of the highly sensitive parameters, significantly improving the quality and reliability of the initial solution, and accelerating the convergence rate of the subsequent optimization algorithm.

[0054] The expression of the initial solution is: ; where is the cluster center, is the weight, represents the number of effective clusters remaining after screening.

[0055] As Figure 3 shown, an embodiment of the present invention further provides an intelligent textile production control system 100 based on deep learning. The system includes a processing module 200 and a storage module 300. Computer program codes are stored in the storage module and are called and executed by the processing module to implement the following steps: obtaining the opening cotton process data of the spinning mill that sends the ginning task, and matching the nep content according to the opening cotton process data; performing similarity calculations in the ginning process database according to the nep content and the type of lint picking to obtain several first ginning process data; where the picking types include machine picking and manual picking; using a joint simulation analysis model to perform joint simulation analysis on the nep content and each of the first ginning process data to obtain the optimal second ginning process data.

[0056] An embodiment of the present invention further provides an electronic device, which includes: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor. When the computer program is executed by the processor, it implements the method described in any one of the foregoing.

[0057] An embodiment of the present invention further provides a computer storage medium, which stores a computer program that can be executed by a processor to implement the method described in any one of the foregoing.

[0058] An embodiment of the present invention further provides a computer program product, which includes a computer program that can be executed by a processor to implement the method described in any one of the foregoing.

[0059] Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. To clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described according to functions in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0060] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of various equivalent modifications or substitutions, and these modifications or substitutions should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. An intelligent control method for textile production based on deep learning, characterized in that: The method steps include: obtaining the opening process data of the spinning mill that sends the ginning task, and matching the nep content according to the opening process data; performing similarity calculation in the ginning process database according to the nep content and the type of lint picking, and obtaining a number of first ginning process data; wherein, the picking types include machine picking and manual picking; using a joint simulation analysis model to perform joint simulation analysis on the nep content and each of the first ginning process data to obtain the optimal second ginning process data; Using a joint simulation analysis model to perform joint simulation analysis on the nep content and each of the first ginning process data to obtain the optimal second ginning process data, including: generating a random noise signal and adding it to the first ginning process data corresponding to the manual picking type; the standard deviation of the random noise signal is dynamically adjusted according to the impurity fluctuation range of the manually picked cotton in this ginning cycle; performing random perturbation tests on each of the first ginning process data more than a preset number of times through Monte Carlo simulation to generate a parameter space sample, performing cluster analysis on each of the parameter space samples, and extracting the parameter combination with the highest occurrence frequency as the initial solution; using an optimization algorithm to perform iterative optimization on the initial solution and outputting the optimal second ginning process data that meets the end condition; for the first ginning process data corresponding to machine picking, directly performing a random perturbation test with a fixed perturbation.

2. The intelligent control method for textile production based on deep learning according to claim 1, characterized in that: Matching the nep content according to the opening process data includes: retrieving a mapping model, which is trained by a deep learning algorithm based on the historical production data of the spinning mill; wherein, the historical production data includes the number of pre-opening channels, the number of fine-opening channels, the speed of the opening machine, the interval between the beater and the dust bar, the air flow pressure parameters and their corresponding nep detection data under different opening process configurations; inputting the opening process data into the mapping model to map and obtain the predicted value of the corresponding nep content; wherein, when the opening process data includes two pre-opening channels and two fine-opening channels, the predicted value of the nep content output by the mapping model is higher than that when it includes one pre-opening channel and one fine-opening channel, and the predicted value does not exceed the nep removal threshold of the carding process of the spinning mill.

3. The intelligent control method for textile production based on deep learning according to claim 1, characterized in that: Performing random perturbation tests on each of the first ginning process data more than a preset number of times through Monte Carlo simulation to generate a parameter space sample, including: determining the perturbation amplitude and sampling density of the random perturbation according to the impurity fluctuation range of the manually picked cotton, and configuring an additional amplitude for the highly sensitive parameters in the first ginning process data on the basis of the perturbation amplitude to obtain the target perturbation amplitude; wherein, the perturbation amplitude is positively correlated with the impurity fluctuation range of the manually picked cotton, and the additional amplitude is a preset value; performing random perturbation on each of the first ginning process data according to the perturbation amplitude, the target perturbation amplitude and the sampling density for a preset number of times to generate a parameter space sample.

4. The intelligent control method for textile production based on deep learning according to claim 1, characterized in that: Performing clustering analysis on each of the parameter space samples and extracting the parameter combination with the highest occurrence frequency as the initial solution, including: performing PCA dimensionality reduction on the parameter space samples, and extracting the principal components with a cumulative contribution rate reaching or exceeding a preset value as new feature vectors; using the DPC algorithm to automatically identify the clustering centers by calculating the local density of sample points and the distance decision graph; using the FCM algorithm to perform secondary optimization on the clustering results of DPC to obtain each clustering cluster; for each clustering cluster, calculating the mean, standard deviation of process parameters, and the confidence interval of performance indicators, and removing the abnormal clusters with a standard deviation exceeding the threshold; taking the central parameter combination of the remaining clustering clusters as the preliminary initial solution, and obtaining the initial solution by weighted averaging of each preliminary initial solution.

5. An intelligent control system for textile production based on deep learning, characterized in that, The system includes a processing module and a storage module. The storage module stores computer program code, and the computer program code is called and executed by the processing module to implement the following steps: obtaining the opening process data of the spinning mill that sends the ginning task, and matching the neps content according to the opening process data; performing similarity calculation in the ginning process database according to the neps content and the type of cotton picking, and obtaining several first ginning process data; wherein, the picking types include machine picking and manual picking; using the joint simulation analysis model to perform joint simulation analysis on the neps content and each of the first ginning process data to obtain the optimal second ginning process data. Using the joint simulation analysis model to perform joint simulation analysis on the neps content and each of the first ginning process data to obtain the optimal second ginning process data, including: generating a random noise signal and adding it to the first ginning process data corresponding to the manual picking type; the standard deviation of the random noise signal is dynamically adjusted according to the impurity fluctuation range of the manually picked cotton in this ginning cycle; performing random perturbation tests on each of the first ginning process data more than a preset number of times through Monte Carlo simulation to generate parameter space samples, performing clustering analysis on each of the parameter space samples, and extracting the parameter combination with the highest occurrence frequency as the initial solution; using an optimization algorithm to perform iterative optimization on the initial solution and outputting the optimal second ginning process data that meets the end condition; for the first ginning process data corresponding to machine picking, directly performing random perturbation tests with fixed perturbations.

6. The intelligent control system for textile production based on deep learning according to claim 5, characterized in that: Matching the neps content according to the opening process data, including: retrieving the mapping model, which is trained by a deep learning algorithm based on the historical production data of the spinning mill; wherein, the historical production data includes the number of pre-opening passages, the number of fine-opening passages, the rotational speed of the opening machine, the interval distance between the beater and the dust rod, the air flow pressure parameters, and the corresponding neps detection data under different opening process configurations; inputting the opening process data into the mapping model to map and obtain the predicted value of the corresponding neps content; wherein, when the opening process data includes two pre-opening passages and two fine-opening passages, the predicted value of the neps content output by the mapping model is higher than that when it includes one pre-opening passage and one fine-opening passage, and the predicted value does not exceed the neps removal threshold of the carding process of the spinning mill.

7. An electronic device, characterized in that: The electronic device includes: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, where the computer program, when executed by the processor, implements the method according to any one of claims 1-4.

8. A computer storage medium, characterized in that: The computer storage medium stores a computer program executable by a processor to implement the method according to any one of claims 1-4.

9. A computer program product, characterized in that: The computer program product includes a computer program executable by a processor to implement the method according to any one of claims 1-4.

Images