A textile process optimization method based on artificial intelligence

Abstract

The present invention belongs to the technical field of textile manufacturing process optimization, and in particular to a textile process optimization method based on artificial intelligence. First, the data of the entire production process is collected, cleaned, and the correlation is analyzed based on the Pearson coefficient to screen out weakly correlated variables. The neural network parameters are optimized using an adaptive spiral Golden Eagle optimization algorithm. The algorithm initializes the threshold value as the individual position vector of the Golden Eagle and the historical best position, calculates the adaptive factor and the search radius, and iterates through spiral motion displacement calculation, attack and cruise vector and coefficient improvement to find the best. Finally, the optimized neural network prediction is used to guide the setting of production equipment operating parameters, accurately control product quality, improve the tensile strength, tear strength and wear resistance of textiles, reduce the defective rate, enhance production efficiency and corporate competitiveness, and promote the intelligent upgrading of the industry.

Description

Technical Field

[0001] The present invention belongs to the technical field of textile manufacturing process optimization, and in particular relates to a textile process optimization method based on artificial intelligence. Background Art

[0002] In today's textile manufacturing industry, the market's requirements for product quality continue to rise, covering multi-dimensional indicators such as tensile strength, tear strength and abrasion resistance. The textile production process is complex, which profoundly affects the quality of finished products and production efficiency. Traditional process optimization relies on long-term exploration and a large number of trial and error experiments by experienced technicians. This method is extremely time-consuming and costly. It is difficult to accurately capture the complex nonlinear relationship between various factors, resulting in poor product quality stability, difficulty in reducing the defective rate, obstacles to improving production efficiency, and poor control of energy consumption and raw material loss. With the development of science and technology, artificial intelligence has attracted attention in the transformation of multiple industries. Neural networks have great potential in complex system modeling and optimization, and can fit highly nonlinear functional relationships. However, the key to their performance lies in the accuracy and efficiency of parameter optimization. Although a variety of optimization algorithms have emerged, there are still significant shortcomings in the context of textile process optimization. For example, some algorithms are sensitive to initial values, easily fall into the local optimal trap, and miss the global optimal process parameter combination; some convergence speeds are too slow to meet the needs of real-time production optimization. The standard Golden Eagle algorithm tends to converge prematurely when dealing with the complex high-dimensional search space of textile processes, making it difficult to fully explore the global optimal solution. It is also unable to provide ideal parameter configurations for the neural network to cope with the challenges of collaborative optimization of complex processes such as fiber interweaving, yarn forming, and fabric weaving in textile production. Summary of the invention

[0003] The present invention solves the technical problems existing in the above-mentioned background technology and proposes a textile process optimization method based on artificial intelligence which is simple, theoretically strong and feasible.

[0004] In order to achieve the above object, the technical solution adopted by the present invention comprises the following steps:

[0005] S1. First, collect data from the production process of textiles;

[0006] S2. Clean the data, perform correlation analysis on the data, and remove weakly correlated variables;

[0007] S3, using the adaptive spiral golden eagle optimization algorithm ASGEO to optimize the neural network and obtain the neural network parameters with the best fitting effect;

[0008] The optimization implementation of the adaptive spiral golden eagle optimization algorithm ASGEO is as follows:

[0009] S31, first initialize the threshold, weight, and number of layers to n individual position vectors of golden eagles , initialize the best historical position of each Golden Eagle individual ;

[0010] S32, calculate the adaptive factor before iteration : ,in is the maximum number of iterations, t is the current number of iterations, and the search radius r is defined;

[0011] S33, then for the golden eagle individual i, select another golden eagle individual j and the best historical position according to the adaptive search strategy , calculate the spiral motion displacement: ,in Control parameters for the spiral shape;

[0012] S34, then calculate the attack vector according to the standard Golden Eagle algorithm Vector with cruise , improve attack coefficient and cruise coefficient ;

[0013] S35, calculate the new position: ,in are the position changes related to the spiral search strategy and the movement strategy in the i-th dimension respectively;

[0014] S36, calculate fitness based on new position , update the historical best position of individuals and populations, if but ;

[0015] S37, finally continue the iteration step until the maximum number of iterations is reached Finish;

[0016] S4. Finally, the optimized neural network is used for prediction, and the prediction results guide the setting of operating parameters of production equipment.

[0017] Preferably, the data of the textile production process in step S1 include fiber component ratio, fiber strength, twist, twist direction, yarn count, spinning speed, warp and weft density, and tensile strength, tear strength and abrasion resistance of finished product quality inspection data.

[0018] Preferably, the correlation analysis in step S2 is calculated using the Pearson correlation coefficient, and features whose absolute value of the correlation coefficient is greater than 0.45 are selected.

[0019] Preferably, the adaptive factor in step S32 It decreases from 1 as the number of iterations increases, dynamically controlling the degree of search range scaling.

[0020] Preferably, the search radius r in step S32 is an initial radius set according to the scale and characteristics of the process optimization problem. , , so that the search range shrinks adaptively with iterations.

[0021] Preferably, the attack coefficient in step S34 is Updated to: ,in is the initial attack coefficient, is the maximum attack coefficient, Update the cruise coefficient in the same way.

[0022] Preferably, the position change amount related to the mobile strategy in step S35 .

[0023] Compared with the prior art, the advantages and positive effects of the present invention are that it changes the traditional model of relying on experience and trial and error, collects multi-dimensional production data comprehensively and accurately, and through strict cleaning and scientific correlation analysis, digs out key factors, and uses optimized neural networks to accurately guide the setting of production parameters, significantly enhancing the tensile strength, tear strength and wear resistance of products, reducing the defective rate, and stabilizing quality. In terms of efficiency improvement, the adaptive spiral Golden Eagle optimization algorithm is introduced to optimize the parameters of the neural network. Its dynamic search characteristics accelerate convergence, avoid local optimality, greatly shorten the process optimization cycle, quickly respond to market demand, improve enterprise production capacity and competitiveness, and achieve efficient and high-quality production. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0025] Figure 1 Flowchart of the algorithm of the present invention. DETAILED DESCRIPTION

[0026] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described below in conjunction with the accompanying drawings and embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0027] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways than those described herein. Therefore, the present invention is not limited to the specific embodiments of the following disclosure.

[0028] Example: In the field of textile manufacturing, product quality stability has always been a core concern. Traditional process optimization relies on manual experience and a lot of trial and error. It is often powerless in the face of factors such as diverse fiber composition, complex yarn parameters, and variable production processes that affect the quality of finished products. For example, when adjusting the spinning speed and warp and weft density combination to improve tensile strength, it is easy to lose sight of one factor while focusing on another because it is impossible to accurately grasp the synergistic relationship between various factors, resulting in large fluctuations in product quality and a high defective rate, making it difficult to stably meet the stringent market standards.

[0029] The overall process of the present invention is as follows Figure 1 As shown. First, a comprehensive data collection and intelligent analysis solution is used to accurately collect all-round production information such as fiber composition ratio, strength, twist, twist direction, yarn count, spinning speed, warp and weft density, and finished product quality inspection data to build a large and detailed data set. Data cleaning is to draw a box plot based on the data distribution characteristics and determine the quartiles. , and interquartile range , will be less than or greater than The data are marked as suspected outliers, and an outlier index set is constructed. Then, for the suspected outliers, a data distribution statistical model is constructed, and a hypothesis test is used to verify whether it is significantly different from the overall distribution. Then a correlation analysis is performed. The correlation analysis is calculated using the Pearson correlation coefficient. Features with an absolute value of the correlation coefficient greater than 0.45 are selected, and redundant weakly correlated variables are removed. The dimension is reduced while retaining key information.

[0030] Then, considering that the long trial-and-error process of traditional process optimization greatly slows down the production rhythm, hindering the release of enterprise production capacity and the improvement of market response speed. In order to improve production efficiency and stabilize product quality, this paper introduces the adaptive spiral golden eagle optimization algorithm (ASGEO) and the neural network fusion innovation solution. First, the threshold, weight, and number of layers are initialized to n golden eagle individual position vectors. , initialize the best historical position of each Golden Eagle individual ; Calculate the adaptive factor before iteration : , adaptive factor As the number of iterations increases, it decreases from 1 to dynamically control the degree of search range scaling, where is the maximum number of iterations, t is the current number of iterations, and the search radius r is defined. The search radius r is the initial radius set according to the scale and characteristics of the process optimization problem. , , so that the search range shrinks with iteration adaptively; then for the golden eagle individual i, another golden eagle individual j and the best historical position are selected according to the adaptive search strategy , calculate the spiral motion displacement: ,in Control parameters for the spiral shape; then calculate the attack vector according to the standard Golden Eagle algorithm Vector with cruise , improve attack coefficient and cruise coefficient , attack coefficient Updated to: ,in is the initial attack coefficient, is the maximum attack coefficient, Similarly, update the cruise coefficient; then calculate the new position: ,in are the position changes related to the spiral search strategy and the mobile strategy on the i-th dimension, respectively, where the position change related to the mobile strategy ; Then calculate the fitness based on the new position , update the historical best position of individuals and populations, if but ; Finally, continue the iteration step until the maximum number of iterations is reached Finish.

[0031] Finally, the relevant data of the production batch to be optimized is input into the optimized neural network. Based on the complex process parameter and quality relationship model it has learned, the neural network accurately predicts key parameters such as the optimal fiber mixing ratio adjustment value, the optimal spinning speed range, and the ideal warp and weft density setting. The present invention deeply explores the relationship between process parameters and quality, effectively avoiding the limitations of traditional methods. By inputting the production batch data into the optimized neural network, key parameters such as fiber mixing ratio, spinning speed, warp and weft density can be accurately predicted. This brings product quality control to a new level, greatly improves tensile strength, tear strength and wear resistance, and has stable quality.

[0032] The above description is only a preferred embodiment of the present invention and does not limit the present invention in other forms. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A textile process optimization method based on artificial intelligence, characterized in that: The following steps are involved: S1. First, collect data from the production process of textiles; S2. Clean the data, perform correlation analysis on the data, and remove weakly correlated variables; S3, using the adaptive spiral golden eagle optimization algorithm ASGEO to optimize the neural network and obtain the neural network parameters with the best fitting effect; The optimization implementation of the adaptive spiral golden eagle optimization algorithm ASGEO is as follows: S31, first initialize the threshold, weight, and number of layers to n individual position vectors of golden eagles , initialize the best historical position of each Golden Eagle individual ; S32, calculate the adaptive factor before iteration : ,in is the maximum number of iterations, t is the current number of iterations, and the search radius r is defined; S33, then for the golden eagle individual i, select another golden eagle individual j and the best historical position according to the adaptive search strategy , calculate the spiral motion displacement: ,in Control parameters for the spiral shape; S34, then calculate the attack vector according to the standard Golden Eagle algorithm Vector with cruise , improve attack coefficient and cruise coefficient ; S35, calculate the new position: ,in are the position changes related to the spiral search strategy and the movement strategy in the i-th dimension respectively; S36, calculate fitness based on new position , update the historical best position of individuals and populations, if but ; S37, finally continue the iteration step until the maximum number of iterations is reached Finish; S4. Finally, the optimized neural network is used for prediction, and the prediction results guide the setting of operating parameters of production equipment.

2. The method for optimizing textile process based on artificial intelligence according to claim 1, characterized in that: The data of the textile production process in step S1 include fiber component ratio, fiber strength, twist, twist direction, yarn count, spinning speed, warp and weft density, and tensile strength, tear strength and abrasion resistance of finished product quality inspection data.

3. The method for optimizing textile process based on artificial intelligence according to claim 1, characterized in that: The correlation analysis in step S2 is calculated using the Pearson correlation coefficient, and features with an absolute value of the correlation coefficient greater than 0.45 are selected.

4. The method for optimizing textile process based on artificial intelligence according to claim 1, characterized in that: The adaptive factor in step S32 It decreases from 1 as the number of iterations increases, dynamically controlling the degree of search range scaling.

5. The method for optimizing textile process based on artificial intelligence according to claim 1, characterized in that: The search radius r in step S32 is an initial radius set according to the scale and characteristics of the process optimization problem. , , so that the search range shrinks adaptively with iterations.

6. The method for optimizing textile process based on artificial intelligence according to claim 1, characterized in that: The attack coefficient of step S34 Updated to: ,in is the initial attack coefficient, is the maximum attack coefficient, Update the cruise coefficient in the same way.

7. The method for optimizing textile process based on artificial intelligence according to claim 1, characterized in that: The position change amount related to the mobile strategy in step S35 .

Images