Intelligent control system and method for ultrafine fiber production line

Abstract

The present invention discloses an intelligent control system and method for a superfine fiber production line, which relates to the technical field of fiber production. The system comprises: extracting K key nodes from the superfine fiber production line for data association mining to form a key node control data set; extracting production evaluation indicators and screening out a set of quality evaluation indicators for K nodes. Performing feature dimension reduction to obtain K key node production control feature data sets. Performing production prediction training to generate K node production quality prediction models, and fusing to generate a fiber production quality adaptive prediction model. Monitoring the current production line working data flow, performing production quality prediction based on the adaptive prediction model, and obtaining target production quality prediction information. Optimizing the control strategy according to the prediction results, forming the target production line control strategy parameters, and realizing adaptive feedback control. Thereby achieving the technical effect of reducing control costs, improving control response and product quality.

Description

Technical Field

[0001] The invention relates to the technical field of fiber production, and in particular to an intelligent control system and method for a superfine fiber production line. Background Art

[0002] The production process of the microfiber production line is complex and involves multiple key nodes. The control parameters of each node have a significant impact on the quality of the final product. Existing production control methods often rely on the experience of professionals and simple feedback control, lack an in-depth understanding of production control parameters, and have technical problems such as slow control response, high control cost, and unstable product quality. Summary of the invention

[0003] The present invention provides an intelligent control system and method for a superfine fiber production line to solve the technical problems of slow control response, high control cost and unstable product quality in the prior art, and achieve the technical effect of reducing control cost and improving control response and product quality.

[0004] In a first aspect, the present invention provides an intelligent control system for a superfine fiber production line, wherein the system comprises:

[0005] A data mining module is used to extract key nodes from the ultrafine fiber production line to obtain K key production nodes, perform production data association mining based on the K key production nodes, and obtain K key production node control data sets.

[0006] The indicator selection module is used to extract production evaluation indicators and perform correlation evaluation screening on the K key production node control data sets in turn to obtain a set of K node production quality evaluation indicators.

[0007] A data simplification module is used to perform feature association dimensionality reduction on the K key production node control data sets according to the K node production quality assessment indicator sets to obtain K key node production control feature data sets.

[0008] A prediction training module is used to perform production prediction training optimization on the K key node production control feature data sets respectively, obtain K node production quality prediction models, and integrate the K node production quality prediction models to generate a fiber yarn production quality adaptive prediction model.

[0009] A quality prediction module is used to monitor and obtain the current production line working data flow of the ultrafine fiber production line, perform production quality prediction on the current production line working data flow based on the fiber production quality adaptive prediction model, and obtain target fiber production quality prediction information.

[0010] An optimization control module is used to perform control strategy optimization analysis based on the target fiber yarn production quality prediction information, obtain target production line control strategy parameters, and perform adaptive feedback control on the ultrafine fiber yarn production line through the target production line control strategy parameters.

[0011] In a second aspect, the present invention further provides an intelligent control method for a superfine fiber production line, wherein the method comprises:

[0012] Key nodes of the ultrafine fiber production line are extracted to obtain K key production nodes, and production data association mining is performed based on the K key production nodes to obtain K key production node control data sets.

[0013] The K key production node control data sets are sequentially subjected to production evaluation index extraction and correlation evaluation screening to obtain a set of K node production quality evaluation indicators.

[0014] According to the K node production quality assessment indicator sets, feature association dimensionality reduction is performed on the K key production node control data sets to obtain K key node production control feature data sets.

[0015] The K key node production control feature data sets are respectively subjected to production prediction training optimization to obtain K node production quality prediction models, and the K node production quality prediction models are subjected to influence fusion to generate a fiber yarn production quality adaptive prediction model.

[0016] The current production line working data flow of the ultrafine fiber production line is monitored and acquired, and production quality prediction is performed on the current production line working data flow based on the fiber production quality adaptive prediction model to obtain target fiber production quality prediction information.

[0017] Based on the target fiber yarn production quality prediction information, a control strategy optimization analysis is performed to obtain a target production line control strategy parameter, and the ultrafine fiber yarn production line is adaptively feedback controlled by the target production line control strategy parameter.

[0018] The intelligent control system and method for a superfine fiber production line disclosed in the present invention solve the technical problems of slow control response, high control cost, and unstable product quality, and achieve the technical effect of reducing control cost and improving control response and product quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a structural schematic diagram of the intelligent control system used in the ultrafine fiber production line of the present invention;

[0020] Figure 2The figure is a schematic flow chart of the intelligent control method for ultrafine fiber production line of the present invention.

[0021] Explanation of the accompanying drawings: data mining module 11, indicator selection module 12, data simplification module 13, prediction training module 14, quality prediction module 15, optimization control module 16. DETAILED DESCRIPTION

[0022] The technical solution provided in the embodiments of the present invention is to solve the technical problems of slow control response, high control cost and unstable product quality in the prior art. The overall idea adopted is as follows:

[0023] First, the key nodes of the ultrafine fiber production line are extracted to obtain multiple key production nodes; then, production data association mining is performed based on the multiple key production nodes to obtain multiple key production node control data sets; then, production evaluation index extraction and correlation evaluation screening are performed on the multiple key production node control data sets in turn to obtain multiple node production quality evaluation index sets; then, feature association dimension reduction is performed on the multiple key production node control data sets according to the evaluation index set to obtain multiple key node production control feature data sets; then, production prediction training and optimization are performed on these feature data sets respectively to obtain multiple node production quality prediction models, and these models are fused to generate a fiber production quality adaptive prediction model; finally, the current production line working data flow of the ultrafine fiber production line is monitored and obtained, and the production quality prediction of the current data flow is performed based on the adaptive prediction model to obtain the target fiber production quality prediction information; control strategy optimization analysis is performed based on the prediction information to obtain the target production line control strategy parameters, and the ultrafine fiber production line is adaptively feedback controlled by these parameters.

[0024] The above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods of the specification to better understand the above technical solution. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments used only to explain the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. In addition, it should be noted that, for the convenience of description, only the parts related to the present invention are shown in the drawings, rather than all of them.

[0025] Embodiment 1

[0026] Figure 1 The structure diagram of the intelligent control system for the ultrafine fiber production line of the present invention is shown in FIG. 1 , wherein the system comprises:

[0027] The data mining module 11 is used to extract key nodes of the ultrafine fiber production line to obtain K key production nodes, and to perform production data association mining based on the K key production nodes to obtain K key production node control data sets.

[0028] Specifically, first, analyze the target ultrafine fiber production line, and obtain multiple control nodes of the target ultrafine fiber production line according to the adopted production process and production line equipment, including raw material preparation, raw material input, spinning, fiber stretching, post-processing, fiber finishing, etc.; then, determine K key production nodes through an interactive expert system or according to the direct correlation between multiple control nodes and fiber forming. Exemplarily, the key production nodes include raw material preparation, dissolving pulping, spinning, stretching and shaping, post-processing, etc.

[0029] Optionally, use data analysis tools (such as flow charts, Pareto analysis, etc.) to evaluate the impact of each node on production efficiency and product quality to ensure that the extracted nodes are representative and important.

[0030] Specifically, according to the obtained K key production nodes, the corresponding associated production control data are extracted to obtain K key production node control data sets. Among them, the K key production node control data sets have the same specific time span. In other words, the control data at any time in any key production node control data set can find the corresponding control data in all other key production node control data sets.

[0031] The indicator selection module 12 is used to extract production evaluation indicators and perform correlation evaluation screening on the K key production node control data sets in turn to obtain a set of K node production quality evaluation indicators.

[0032] In some embodiments, obtaining a set of K node production quality assessment indicators, the execution steps include:

[0033] The production evaluation indicators of the K key production node control data sets are extracted in turn to obtain K node-related production evaluation indicator sets; performance indicator clustering analysis is performed on the K node-related production evaluation indicator sets to obtain K node production evaluation indicator cluster sets; correlation analysis is performed on each indicator information in the K node production evaluation indicator cluster sets to obtain K node indicator correlation coefficient sets; based on the K node indicator correlation coefficient sets, the K node-related production evaluation indicator sets are strongly correlated and screened to obtain the K node production quality evaluation indicator sets.

[0034] Specifically, first, the data set of each node is preprocessed to remove missing values, outliers, etc., and then, production-related evaluation indicators such as output, quality, efficiency, energy consumption, etc. are extracted as a set of K node-associated production evaluation indicators, which is a set of all evaluation indicators related to the production quality of each node.

[0035] Specifically, cluster analysis is performed on the standardized data to group similar evaluation indicators into one category to form indicator clusters. Each indicator cluster represents a large type of performance evaluation indicator. For example, the node evaluation indicator cluster is divided into: physical performance indicator cluster (such as fiber diameter, strength, elongation), chemical performance indicator cluster (such as antistatic, antibacterial), etc.

[0036] Specifically, a correlation analysis is performed on the production evaluation index cluster set of each node, and the correlation coefficients (such as the Pearson correlation coefficient) between the indicators are calculated to generate a correlation coefficient matrix, in which the elements in the matrix represent the correlation strength between different indicators. The above correlation analysis process is repeated to obtain the indicator correlation coefficient sets of K nodes, where each indicator correlation coefficient set contains the correlation coefficient matrix between the evaluation indicators in the production evaluation index cluster set of the node.

[0037] Furthermore, based on the correlation coefficient matrix, evaluation indicators with strong correlation are screened out. For example, the eigenvalues ​​and eigenvectors of the correlation coefficient matrix can be calculated, and the contribution of multiple evaluation indicators can be obtained, and then the evaluation indicators with strong contribution (strong correlation) can be screened out as the production quality evaluation indicator set of the node.

[0038] Specifically, the K node indicator correlation coefficient sets are traversed to obtain the node production quality evaluation indicator sets of the K nodes, each set containing several evaluation indicators with strong correlation of the node.

[0039] Through the above steps, the control data sets of K key production nodes can be systematically analyzed to ensure that the extracted production quality evaluation indicators are highly representative and relevant, which helps to more accurately evaluate and optimize the production process.

[0040] In some implementations, obtaining a set of K node indicator correlation coefficients includes:

[0041] Perform node production quality correlation analysis on each indicator cluster in the K node production evaluation indicator cluster set respectively to obtain a set of correlation coefficients of K node indicator clusters; determine a set of association strength factors of K node indicator clusters based on the set of correlation coefficients of K node indicator clusters; perform performance indicator correlation evaluation on the indicator information in each indicator cluster of the K node production evaluation indicator cluster set in turn to obtain a set of correlation coefficients of indicator cluster elements; and use the product of the set of association strength factors of K node indicator clusters and the set of correlation coefficients of indicator cluster elements as the K node indicator correlation coefficient sets of each indicator information in the K node production evaluation indicator cluster set.

[0042] Specifically, first, the correlation between each indicator cluster in the node production evaluation indicator cluster set and the production quality is analyzed (such as using methods such as the Pearson correlation coefficient and the Spearman correlation coefficient), and a set of node indicator cluster correlation coefficients is obtained. The node indicator cluster correlation coefficient set reflects the contribution of different types of evaluation indicators (such as physical and chemical indicators) to production quality.

[0043] Specifically, traverse to obtain a set of correlation coefficients of K node indicator clusters, and then normalize each obtained set of correlation coefficients of node indicator clusters to obtain a set of association strength factors of K node indicator clusters, thereby converting the correlation coefficients of different nodes to the same scale for subsequent calculations.

[0044] Specifically, the performance indicator correlation evaluation is performed on each indicator in each indicator cluster of each node and the production quality, and the correlation coefficient is calculated to generate a set of correlation coefficients of indicator cluster elements of each node. In other words, the correlation between different evaluation indicators of each indicator cluster of each node and the production quality is also different.

[0045] Furthermore, for each indicator in each indicator cluster of each node, its indicator cluster association strength factor is multiplied by the indicator cluster element correlation coefficient to obtain the final correlation coefficient of the information of each indicator, which is output as the node indicator correlation coefficient. The node indicator correlation coefficient integrates the correlation differences between indicator clusters and between the evaluation indicators within the indicator cluster, so as to more accurately understand the relationship between different evaluation indicators and production quality.

[0046] The data reduction module 13 is used to perform feature association dimensionality reduction on the K key production node control data sets according to the K node production quality assessment indicator sets to obtain K key node production control feature data sets.

[0047] Furthermore, based on the determined set of K node production quality evaluation indicators, feature correlation dimensionality reduction is performed to clean up unnecessary or low-relevance redundant data in the K key production node control data sets, thereby improving the quality of the data sets.

[0048] In some embodiments, obtaining K key node production control feature data sets, the execution steps include:

[0049] Acquire a production line process control database, perform control feature hierarchical association based on the production line process control database, and construct a production control feature graph structure network; perform control feature association on the K key production node control data sets according to the K node production quality assessment indicator sets, and obtain K node associated control feature sets; determine the K node indicator dimensionality reduction depth set according to the K node indicator correlation coefficient set; perform associated dimensionality reduction on the K node associated control feature sets based on the production control feature graph structure network and the K node indicator dimensionality reduction depth set, and obtain K node production control feature sets; map and integrate the K node production control feature sets with the K key production node control data sets, and obtain the K key node production control feature data sets.

[0050] Specifically, firstly, based on the production line process control database, combined with the graph neural network technology, the control feature hierarchical association is performed according to the subordinate relationship of the control features to obtain the production control feature graph structure network. Among them, the production line process control database contains all the control feature parameters related to the production line control, such as the temperature, pressure, speed, etc. of each node.

[0051] Specifically, according to the K node production quality evaluation index sets and the known feature association relationships in the expert system (such as extrusion speed affects the fiber surface molding quality), the above K key production node control data sets are divided into K node association control feature sets. Among them, the node association control feature set is a set of multiple association control feature groups, and the control features in different sets can appear repeatedly. For example, if the spinning pressure affects the elongation and strength index of the fiber at the same time, the spinning pressure (control feature) can appear in the association control feature group of the elongation and the association control feature group of the strength index at the same time.

[0052] Specifically, the dimensionality reduction degree of each feature is determined according to the size of the correlation coefficient. The larger the correlation coefficient, the larger the amount of feature data required, and the smaller the feature dimensionality reduction degree, that is, the smaller the dimensionality reduction depth. The dimensionality reduction depth determines the number of features that need to be discarded.

[0053] Optionally, based on the production control feature graph structure network and the dimensionality reduction depth set, the associated control feature set of K nodes is reduced in dimension to extract key production control features. Among them, the dimensionality reduction methods include principal component analysis (PCA), linear discriminant analysis (LDA), feature selection, etc. Through the dimensionality reduction process, the features with strong correlation with production quality indicators can be retained.

[0054] Specifically, according to the K node production control feature sets, the corresponding control data in the K key production node control data sets are called and stored as K key node production control feature data sets. Among them, the key node production control feature data set is a data set after data screening, including the control feature data related to the node production control feature set and the corresponding production quality data, which helps to reduce the computing power consumption of subsequent analysis and calculation.

[0055] In some implementations, constructing a production control feature graph structure network, the execution steps include:

[0056] Based on K key production nodes, control features are associated with the production line process control database to obtain a first-order node association control feature set; based on the first-order node association control feature set, neighbor feature nodes are collected in the production line process control database as a second-order node association control feature set; the second-order node association control feature set is traversed, and N-order node growth is performed in the production line process control database to obtain an N-order node association control feature set, where 2≤N≤5; based on the first-order node association control feature set, the second-order node association control feature set, and the N-order node association control feature set, feature levels are increased until the N-order node association control feature set, and the production control feature graph structure network is constructed.

[0057] Specifically, a production control feature graph structure network based on graph neural network is constructed. First, the control features directly related to each key node are extracted from the production line process control database to form a first-order node associated control feature set. Among them, the first-order node associated control feature set contains the most basic major control features. Then, based on the first-order node associated control feature set, the neighbor feature nodes directly related to these features are identified as the second-order node associated control feature set, and the associated adjacent features are collected in turn, that is, the next level of control features associated with them are cascaded layer by layer until the N-order node associated control feature set is generated. Optionally, N is determined based on the control accuracy and control computing power of the target scenario, 2≤N≤5.

[0058] Furthermore, a graph structure is initialized with K key production nodes as the starting point; then, starting from the first-order node association control feature set, nodes, lower-level nodes and edges are added one by one to form a multi-level graph structure, and it is ensured that the nodes and edges of each layer can be correctly associated with the nodes of the previous layer until the N-order nodes form a complete graph structure network. Through the above process, a production control feature graph structure network can be systematically constructed to better understand and optimize the key control parameters in the production process.

[0059] The prediction training module 14 is used to perform production prediction training optimization on the K key node production control feature data sets respectively, obtain K node production quality prediction models, and integrate the influence of the K node production quality prediction models to generate a fiber yarn production quality adaptive prediction model.

[0060] Specifically, taking K key node production control feature data sets as sample data, a K-node production quality prediction model is obtained through training. The K-node production quality prediction model is used to specifically construct a mapping relationship between the control feature combination of K nodes and the production quality, thereby acquiring the ability to predict the production quality.

[0061] Optionally, the production quality prediction model of each node is built based on a machine learning or deep learning model (such as linear regression, random forest, support vector machine, neural network, etc.), and different key production nodes use different model categories, which are determined based on the node characteristics of the key production nodes. For example, a linear regression model is used for key production nodes with simple response relationships.

[0062] In some embodiments, the fiber filament production quality adaptive prediction model is generated, and the execution steps include:

[0063] The K key production nodes are evaluated for their impact on production quality to obtain K key production node impact coefficients; the K key production nodes are analyzed for the impact of the next node in the order of production nodes to obtain K-1 node additional impact coefficients; K node fusion impact coefficients are determined based on the mapping and summation of the K key production node impact coefficients and the K-1 node additional impact coefficients; the K node production quality prediction models are weightedly fused based on the K node fusion impact coefficients to generate the fiber production quality adaptive prediction model.

[0064] Specifically, based on the statistical analysis method, the impact level of K key production nodes on production quality is evaluated, and the K key production node impact coefficients are obtained, which reflect the impact of a single key production node on production quality. Then, the K key production nodes are analyzed for the impact degree of the next node in the order of production nodes, and K-1 node additional impact coefficients are obtained. The node additional impact coefficients characterize the impact degree of each node on the next node, among which the last node has no additional impact coefficient.

[0065] Specifically, the K node fusion influence coefficients are determined based on the mapping and summation of the influence coefficients of the K key production nodes and the additional influence coefficients of the K-1 nodes, including: mapping the influence coefficient of each node with its additional influence coefficient to determine the fusion influence coefficient of each node, such as the influence coefficient of the n+1th key production node corresponds to the additional influence coefficient of the nth node, and the first key production node influence coefficient has no corresponding node additional influence coefficient. According to the mapping result, the product of the influence coefficients of the K key production nodes and the additional influence coefficients of the K-1 nodes is obtained as the fusion influence coefficient of the K nodes. The production quality prediction model of the K nodes is weightedly fused using the fusion influence coefficient to obtain a fiber production quality adaptive prediction model, which fully considers the influence relationship and influence transfer between key nodes through weighted fusion, and helps to improve the accuracy of quality prediction.

[0066] The quality prediction module 15 is used to monitor and obtain the current production line working data flow of the ultrafine fiber production line, and perform production quality prediction on the current production line working data flow based on the fiber production quality adaptive prediction model to obtain target fiber production quality prediction information.

[0067] Specifically, the management system or management platform of the interactive ultrafine fiber production line collects the current production line working data stream, and then pre-processes the acquired data stream, and inputs the pre-processed current production line working data stream into the prediction model for prediction, and generates the target fiber production quality prediction information. By real-time monitoring of the ultrafine fiber production line and making quality predictions based on the monitoring data, it is helpful to timely discover possible quality accident risks in the production process.

[0068] The optimization control module 16 is used to perform control strategy optimization analysis based on the target fiber yarn production quality prediction information, obtain target production line control strategy parameters, and perform adaptive feedback control on the ultrafine fiber yarn production line through the target production line control strategy parameters.

[0069] Specifically, if the target fiber production quality prediction information shows that there is a quality risk, or the probability of the quality risk does not meet the production quality control requirements of the target scenario, the control strategy is adjusted and optimized. Adaptive feedback control of the ultrafine fiber production line through the target production line control strategy parameters helps to achieve continuous optimization and stabilization of the production process while reducing optimization costs.

[0070] In some embodiments, obtaining the target production line control strategy parameters, the execution steps include:

[0071] The target fiber yarn production quality prediction information is subjected to non-standard quality factor identification and control strategy analysis to obtain a target fiber yarn control optimization strategy; the target fiber yarn production quality prediction information is subjected to control parameter analysis based on the target fiber yarn control optimization strategy to determine a control strategy parameter selection threshold; a production simulation quality assessment is performed based on the control strategy parameter selection threshold, and control parameter optimization is performed on the control strategy parameter selection threshold according to the production simulation quality assessment result to obtain the control strategy parameters of the target production line.

[0072] Specifically, first, the target fiber production quality prediction information is analyzed to identify quality factors that do not meet the standards, such as fiber diameter, strength, etc. Then, based on the above-constructed production control feature graph structure network, the causes of quality deviation are located, such as unstable temperature, pressure fluctuations, unreasonable digital settings, etc. Then, according to the identified non-standard factors, corresponding control strategies are formulated, such as adjusting parameters such as temperature, pressure, and speed.

[0073] Specifically, the formulated control strategy is analyzed to determine the specific parameters that need to be adjusted and their adjustment directions. Then, based on historical data and expert experience, the selection threshold of each control parameter is determined to ensure that the adjusted parameters are within a reasonable range. In other words, the control strategy parameter selection threshold is the optimization parameter space for subsequent optimization.

[0074] Furthermore, the production simulation quality evaluation is performed based on the threshold value of the control strategy parameter selection, and the control parameter selection threshold is optimized according to the evaluation result to obtain the control strategy parameters of the target production line, including: randomly producing new production line control strategy parameters according to the threshold value of the control strategy parameter selection; simulating the new production line control strategy parameters through the fiber production quality adaptive prediction model to evaluate its impact on production quality. According to the simulation results, the effect after adjustment is analyzed, and whether the production quality meets the standard is evaluated, and the control parameters are further optimized, and the adjustment and evaluation are repeated until the best control strategy parameters are obtained.

[0075] The above steps optimize the control parameters through production simulation and quality evaluation to obtain the best control strategy parameters, thereby optimizing and stabilizing the production process and improving the production quality and efficiency of the ultrafine fiber production line.

[0076] In summary, the intelligent control system for the ultrafine fiber production line provided by the present invention has the following technical effects:

[0077] The key nodes of the ultrafine fiber production line are extracted through the data mining module, K key production nodes are identified, and the production data are mined based on these nodes, so as to obtain the control data set of K key nodes; the indicator selection module extracts and screens the production evaluation indicators of these control data sets in turn, and finally obtains the production quality evaluation indicator set of K nodes; the data simplification module performs feature correlation dimension reduction on the control data set of K key nodes according to the indicator set, and obtains the production control feature data set of K key nodes; the prediction training module performs production prediction training optimization on these feature data sets to form a production quality prediction model of K nodes, and generates a fiber production quality adaptive prediction model through influence fusion; the quality prediction module monitors and obtains the working data flow of the current production line, uses the adaptive prediction model to predict its quality, and obtains the production quality prediction information of the target fiber; the optimization control module performs optimization analysis of the control strategy based on the prediction information, obtains the control strategy parameters of the target production line, and uses this to perform adaptive feedback control on the ultrafine fiber production line. Thus, the technical effect of reducing control costs and improving control response and product quality is achieved.

[0078] Embodiment 2

[0079] Figure 2 Schematic diagram of the process of the intelligent control method for ultrafine fiber production line of the present invention. Figure 1 The structure of the intelligent control system for the ultrafine fiber production line of the present invention can be used to achieve the following Figure 2 The process shown.

[0080] Based on the same concept as the intelligent control system for the ultrafine fiber production line in the embodiment, the present invention also provides an intelligent control method for the ultrafine fiber production line, including:

[0081] Key nodes of the ultrafine fiber production line are extracted to obtain K key production nodes, and production data association mining is performed based on the K key production nodes to obtain K key production node control data sets.

[0082] The K key production node control data sets are sequentially subjected to production evaluation index extraction and correlation evaluation screening to obtain a set of K node production quality evaluation indicators.

[0083] According to the K node production quality assessment indicator sets, feature association dimensionality reduction is performed on the K key production node control data sets to obtain K key node production control feature data sets.

[0084] The K key node production control feature data sets are respectively subjected to production prediction training optimization to obtain K node production quality prediction models, and the K node production quality prediction models are subjected to influence fusion to generate a fiber yarn production quality adaptive prediction model.

[0085] The current production line working data flow of the ultrafine fiber production line is monitored and acquired, and production quality prediction is performed on the current production line working data flow based on the fiber production quality adaptive prediction model to obtain target fiber production quality prediction information.

[0086] Based on the target fiber yarn production quality prediction information, a control strategy optimization analysis is performed to obtain a target production line control strategy parameter, and the ultrafine fiber yarn production line is adaptively feedback controlled by the target production line control strategy parameter.

[0087] In some embodiments, obtaining a set of K node production quality assessment indicators includes:

[0088] The production evaluation indicators of the K key production node control data sets are extracted in turn to obtain K node-related production evaluation indicator sets; performance indicator clustering analysis is performed on the K node-related production evaluation indicator sets to obtain K node production evaluation indicator cluster sets; correlation analysis is performed on each indicator information in the K node production evaluation indicator cluster sets to obtain K node indicator correlation coefficient sets; based on the K node indicator correlation coefficient sets, the K node-related production evaluation indicator sets are strongly correlated and screened to obtain the K node production quality evaluation indicator sets.

[0089] In some implementations, obtaining a set of K node index correlation coefficients includes:

[0090] Perform node production quality correlation analysis on each indicator cluster in the K node production evaluation indicator cluster set respectively to obtain a set of correlation coefficients of K node indicator clusters; determine a set of association strength factors of K node indicator clusters based on the set of correlation coefficients of K node indicator clusters; perform performance indicator correlation evaluation on the indicator information in each indicator cluster of the K node production evaluation indicator cluster set in turn to obtain a set of correlation coefficients of indicator cluster elements; and use the product of the set of association strength factors of K node indicator clusters and the set of correlation coefficients of indicator cluster elements as the K node indicator correlation coefficient sets of each indicator information in the K node production evaluation indicator cluster set.

[0091] In some embodiments, the step of obtaining K key node production control feature data sets includes:

[0092] Acquire a production line process control database, perform control feature hierarchical association based on the production line process control database, and construct a production control feature graph structure network; perform control feature association on the K key production node control data sets according to the K node production quality assessment indicator sets, and obtain K node associated control feature sets; determine the K node indicator dimensionality reduction depth set according to the K node indicator correlation coefficient set; perform associated dimensionality reduction on the K node associated control feature sets based on the production control feature graph structure network and the K node indicator dimensionality reduction depth set, and obtain K node production control feature sets; map and integrate the K node production control feature sets with the K key production node control data sets, and obtain the K key node production control feature data sets.

[0093] In some implementations, constructing a production control feature graph structure network includes:

[0094] Based on K key production nodes, control features are associated with the production line process control database to obtain a first-order node association control feature set; based on the first-order node association control feature set, neighbor feature nodes are collected in the production line process control database as a second-order node association control feature set; the second-order node association control feature set is traversed, and N-order node growth is performed in the production line process control database to obtain an N-order node association control feature set, where 2≤N≤5; based on the first-order node association control feature set, the second-order node association control feature set, and the N-order node association control feature set, feature levels are increased until the N-order node association control feature set, and the production control feature graph structure network is constructed.

[0095] In some embodiments, generating an adaptive prediction model for fiber filament production quality includes:

[0096] The K key production nodes are evaluated for their impact on production quality to obtain K key production node impact coefficients; the K key production nodes are analyzed for the impact of the next node in the order of production nodes to obtain K-1 node additional impact coefficients; K node fusion impact coefficients are determined based on the mapping and summation of the K key production node impact coefficients and the K-1 node additional impact coefficients; the K node production quality prediction models are weightedly fused based on the K node fusion impact coefficients to generate the fiber production quality adaptive prediction model.

[0097] In some embodiments, obtaining the target production line control strategy parameters includes:

[0098] The target fiber yarn production quality prediction information is subjected to non-standard quality factor identification and control strategy analysis to obtain a target fiber yarn control optimization strategy; the target fiber yarn production quality prediction information is subjected to control parameter analysis based on the target fiber yarn control optimization strategy to determine a control strategy parameter selection threshold; a production simulation quality assessment is performed based on the control strategy parameter selection threshold, and control parameter optimization is performed on the control strategy parameter selection threshold according to the production simulation quality assessment result to obtain the control strategy parameters of the target production line.

[0099] It should be understood that the embodiments mentioned in this specification focus on their differences from other embodiments. The specific embodiments in the aforementioned embodiment one are also applicable to the intelligent control method for the ultrafine fiber production line described in embodiment two. For the sake of brevity of the specification, they will not be further elaborated here.

[0100] It should be understood that the embodiments disclosed in the present invention and the above description can enable those skilled in the art to use the present invention to implement the present invention. At the same time, the present invention is not limited to the above-mentioned embodiments. It should be understood that those skilled in the art can still modify the technical solutions recorded in the above-mentioned embodiments, or replace some of the technical features therein by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. An intelligent control system for a microfiber production line, characterized in that: The system comprises: A data mining module, wherein the data mining module is used to extract key nodes of the ultrafine fiber production line to obtain K key production nodes, and to perform production data association mining based on the K key production nodes to obtain K key production node control data sets; An indicator selection module, the indicator selection module is used to extract production evaluation indicators and perform correlation evaluation screening on the K key production node control data sets in turn to obtain a set of K node production quality evaluation indicators; A data simplification module, the data simplification module is used to perform feature association dimensionality reduction on the K key production node control data sets according to the K node production quality assessment indicator sets, to obtain K key node production control feature data sets; A prediction training module, wherein the prediction training module is used to perform production prediction training optimization on the K key node production control feature data sets respectively, obtain K node production quality prediction models, and perform influence fusion on the K node production quality prediction models to generate a fiber yarn production quality adaptive prediction model; A quality prediction module, the quality prediction module is used to monitor and obtain the current production line work data flow of the ultrafine fiber production line, perform production quality prediction on the current production line work data flow based on the fiber production quality adaptive prediction model, and obtain target fiber production quality prediction information; An optimization control module, the optimization control module is used to perform control strategy optimization analysis based on the target fiber yarn production quality prediction information, obtain target production line control strategy parameters, and perform adaptive feedback control on the ultrafine fiber yarn production line through the target production line control strategy parameters; The obtaining of a set of K node production quality assessment indicators includes: Extracting production evaluation indicators from the K key production node control data sets in turn to obtain a set of K node-related production evaluation indicators; Performing performance indicator cluster analysis on the K node-related production evaluation indicator sets to obtain K node production evaluation indicator cluster sets; Performing correlation analysis on each indicator information in the K node production evaluation indicator cluster set respectively to obtain a set of K node indicator correlation coefficients; Based on the K node indicator correlation coefficient sets, the K node-related production evaluation indicator sets are screened for strong correlation to obtain the K node production quality evaluation indicator sets; The method of generating a fiber yarn production quality adaptive prediction model comprises: Evaluate the impact degree of the K key production nodes on production quality to obtain the impact coefficients of the K key production nodes; Analyze the influence degree of the K key production nodes on the next node in the order of production nodes to obtain K-1 additional influence coefficients of the nodes; Determine K node fusion influence coefficients according to the mapping sum of the K key production node influence coefficients and the K-1 node additional influence coefficients; The K node production quality prediction models are weightedly fused based on the K node fusion influence coefficients to generate the fiber yarn production quality adaptive prediction model.

2. The intelligent control system for a superfine fiber production line according to claim 1, characterized in that: The step of obtaining a set of correlation coefficients of K node indicators includes: Performing node production quality correlation analysis on each indicator cluster in the K node production evaluation indicator cluster sets respectively to obtain a set of correlation coefficients of the K node indicator clusters; Determine a set of K node indicator cluster association strength factors according to the K node indicator cluster correlation coefficient sets; Performing performance indicator correlation evaluation on the indicator information in each indicator cluster of the K node production evaluation indicator cluster set in turn to obtain a set of correlation coefficients of indicator cluster elements; The product of the K node indicator cluster association strength factor set and the indicator cluster element correlation coefficient set is respectively used as the K node indicator correlation coefficient set of each indicator information in the K node production evaluation indicator cluster set.

3. The intelligent control system for a superfine fiber production line according to claim 1, characterized in that: The step of obtaining K key node production control feature data sets includes: Acquire a production line process control database, perform control feature hierarchical association based on the production line process control database, and construct a production control feature graph structure network; According to the K node production quality assessment indicator sets, control feature association is performed on the K key production node control data sets to obtain K node association control feature sets; Determine a set of K node indicator dimension reduction depths according to the K node indicator correlation coefficient sets; Based on the production control feature graph structure network and the K node indicator dimension reduction depth sets, the K node association control feature sets are associated with dimension reduction to obtain K node production control feature sets; Based on mapping and integrating the K node production control feature sets with the K key production node control data sets, the K key node production control feature data sets are obtained.

4. The intelligent control system for a superfine fiber production line according to claim 3, characterized in that: The construction of the production control feature graph structure network includes: Based on K key production nodes and the production line process control database, control feature association is performed to obtain a first-order node association control feature set; Based on the first-order node association control feature set, neighbor feature nodes are collected in the production line process control database as a second-order node association control feature set; Traversing the second-order node association control feature set, performing N-order node growth in the production line process control database, and obtaining an N-order node association control feature set, where 2≤N≤5; The production control feature graph structure network is constructed by increasing the feature level based on the first-order node association control feature set, the second-order node association control feature set, and the N-order node association control feature set.

5. The intelligent control system for a superfine fiber production line according to claim 1, characterized in that: The obtaining of the target production line control strategy parameters includes: Identifying substandard quality factors and analyzing control strategies for the target fiber yarn production quality prediction information to obtain a target fiber yarn control optimization strategy; Based on the target fiber yarn control optimization strategy, the control parameter analysis of the target fiber yarn production quality prediction information is performed to determine the control strategy parameter selection threshold; A production simulation quality assessment is performed based on the control strategy parameter selection threshold, and control parameter optimization is performed on the control strategy parameter selection threshold according to the production simulation quality assessment result to obtain the target production line control strategy parameters.

6. An intelligent control method for a superfine fiber production line, characterized in that: The method is applied to the intelligent control system for the ultrafine fiber production line according to any one of claims 1 to 5, and the method comprises: Extract key nodes from the microfiber production line to obtain K key production nodes, perform production data association mining based on the K key production nodes, and obtain K key production node control data sets; Extracting production evaluation indicators and performing correlation evaluation screening on the K key production node control data sets in turn to obtain a set of K node production quality evaluation indicators; According to the K node production quality assessment indicator sets, feature correlation dimensionality reduction is performed on the K key production node control data sets to obtain K key node production control feature data sets; Performing production prediction training optimization on the K key node production control feature data sets respectively to obtain K node production quality prediction models, and integrating the K node production quality prediction models to generate a fiber yarn production quality adaptive prediction model; Monitor and obtain the current production line working data flow of the ultrafine fiber production line, perform production quality prediction on the current production line working data flow based on the fiber production quality adaptive prediction model, and obtain target fiber production quality prediction information; Based on the target fiber yarn production quality prediction information, a control strategy optimization analysis is performed to obtain a target production line control strategy parameter, and the ultrafine fiber yarn production line is adaptively feedback controlled by the target production line control strategy parameter.

Images