Knitted fabric weavability verification method and system

Abstract

The invention relates to the technical field of clothing fabrics, in particular to a knitted fabric weavability verification method and system. The method comprises the following steps: firstly, collecting historical and real-time knitted fabric yarn formulas, collecting weaving performance data corresponding to the historical knitted fabric yarn formulas, and carrying out data preprocessing; secondly, identifying correlation between a historical knitted fabric yarn formula and knitted weaving performance data by using mutual information, and extracting significant features; non-linear correlation features are generated based on the significant features. And then, splicing the historical knitted fabric yarn formula, the significant features and the nonlinear correlation features to obtain splicing features, obtaining polynomial features based on the splicing features, inputting a performance prediction model for modeling, and outputting a weaving performance prediction result. And finally, establishing a weavability verification system, and carrying out weighted fusion on weaving performance prediction results to generate a weavability score. Weavability is verified based on the weavability score. According to the method, the nonlinear relation between the features can be captured, and the weavability verification accuracy is improved.

Description

Technical Field

[0001] The invention relates to the technical field of clothing fabrics, and in particular to a method and system for verifying the weavability of knitted fabrics. Background Art

[0002] Yarn is the basic material that makes up knitted fabrics and determines many physical and chemical properties of the fabrics. The type of yarn (such as cotton, wool, polyester, etc.), fiber composition (such as the ratio of natural fiber to synthetic fiber), thickness (such as yarn count and diameter) and other characteristics directly affect the performance and characteristics of the final fabric. These characteristics will affect the strength, elasticity, breathability, comfort and durability of the fabric. Therefore, the selection and formulation of yarns are crucial in the production of knitted fabrics. The types and proportions of yarns must be reasonably allocated according to market demand and product positioning to achieve the best weaving effect and fabric performance.

[0003] In order to ensure the weavability of knitted fabrics, many technical methods have emerged, especially in the control of yarn formula. Traditional methods usually rely on experience and manual experiments, mainly including physical testing and visual evaluation to judge weavability. These methods usually take a lot of time and manpower, and the results often rely on the operator's experience and judgment, which is highly subjective and uncertain, and can easily lead to inaccurate selection of yarn formula.

[0004] Existing technologies mainly include weaving performance prediction based on data analysis and the application of machine learning models. These technologies use historical data to build models to predict new yarn formulas, improving the scientificity and accuracy of decision-making. However, many existing technologies are often not comprehensive enough in feature selection and extraction. In addition, existing technologies have difficulty in capturing the nonlinear and multi-dimensional relationship between different yarn formulas and weaving performance, which will cause inaccurate weavability verification.

[0005] Therefore, a knitted fabric weavability verification method and system are proposed. Summary of the invention

[0006] The purpose of the present invention is to provide a knitted fabric weavability verification method and system for improving the accuracy of knitted fabric weavability verification. First, collect historical and real-time knitted fabric yarn formulas, including yarn types, fiber components and yarn thickness, and collect weaving performance data corresponding to historical knitted fabric yarn formulas, including strength, elasticity, wrinkle resistance and air permeability; perform data preprocessing on historical knitted fabric yarn formulas and real-time knitted fabric yarn formulas. Secondly, use mutual information to identify the correlation between historical knitted fabric yarn formulas and knitting weaving performance data, extract significant features; use significant features to generate nonlinear correlation features based on nonlinear models. Then, splice the historical knitted fabric yarn formula, significant features and nonlinear correlation features to obtain splicing features, and calculate polynomial features based on the splicing features; input the polynomial features into the performance prediction model for model training, and output the weaving performance prediction results. Finally, the real-time knitted fabric yarn formula is subjected to the same data operations as the historical knitted fabric yarn formula to obtain polynomial features; the polynomial features are input into the trained performance prediction model to obtain the weaving performance prediction results; a weavability verification system is established to perform weighted fusion on the various performance results in the weaving performance prediction results to generate a weavability score; and the weavability is verified based on the weavability score.

[0007] To achieve the above object, the present invention provides the following technical solutions: A method for verifying the weavability of a knitted fabric, comprising: Collecting historical and real-time knitted fabric yarn formulas, including yarn types, fiber composition, and yarn thickness, and collecting weaving performance data corresponding to the historical knitted fabric yarn formulas, including strength, elasticity, wrinkle resistance, and air permeability; performing data preprocessing on the historical knitted fabric yarn formulas and the real-time knitted fabric yarn formulas; Using mutual information to identify the correlation between historical knitted fabric yarn formula and the knitted weaving performance data, and extracting significant features; using the significant features, generating nonlinear correlation features based on a nonlinear model, wherein the nonlinear correlation features include mutual influences between features; Splicing the historical knitted fabric yarn formula, the significant features and the nonlinear related features to obtain splicing features, and calculating polynomial features corresponding to the historical data based on the splicing features; inputting the polynomial features into a performance prediction model for model training; The real-time knitted fabric yarn formula is calculated to obtain polynomial features corresponding to the real-time data; the polynomial features are input into the trained performance prediction model to obtain weaving performance prediction results; a weavability verification system is established to perform weighted fusion on each performance result in the weaving performance prediction results to generate a weavability score; and weavability is verified based on the weavability score.

[0008] Furthermore, the process of acquiring the significant features is as follows: The historical knitted fabric yarn formula and the knitted weaving performance data are defined as a feature set and a target variable, respectively; the marginal probability distribution of the feature set and the target variable is calculated, and the joint probability distribution of the feature set and the target variable is calculated; Substituting the marginal probability distribution and the joint probability distribution into the mutual information formula, the mutual information between the feature set and the target variable is obtained; the features are sorted according to the mutual information values, and the first N features with the largest mutual information are selected as significant features.

[0009] Furthermore, the process of obtaining the polynomial feature is as follows: Each feature in the spliced ​​features is taken as a monomial; the square of each feature in the spliced ​​features and the quadratic combination are taken as quadratic terms; the cube of each feature in the spliced ​​features, the combination of the square of each feature in the spliced ​​features and another feature, and the cubic combination of all features in the spliced ​​features are taken as cubic terms; the monomial, quadratic terms and cubic terms are combined to generate the polynomial feature.

[0010] Furthermore, the performance prediction model includes an input layer, a first hidden layer, a second hidden layer, a third hidden layer and an output layer; the input layer receives the polynomial features; the first hidden layer includes 256 neurons and a ReLU activation function; the second hidden layer includes 128 neurons and a ReLU activation function; the third hidden layer includes 64 neurons and a ReLU activation function; the output layer includes 1 neuron, and the output is the weaving performance prediction result.

[0011] Furthermore, in the weavability verification system, the specific process is: Each performance result in the weaving performance prediction result is selected, and a weavability score is calculated by geometric mean; a weavability score threshold is set, and the weavability is verified based on the weavability score threshold, which is divided into a first weavability, a second weavability and a third weavability.

[0012] A knitted fabric weavability verification system, comprising: A data collection and processing unit collects historical and real-time knitted fabric yarn formulas, including yarn types, fiber composition, yarn thickness, etc., and collects weaving performance data corresponding to historical knitted fabric yarn formulas, including strength, elasticity, wrinkle resistance, and air permeability; and performs data preprocessing on the historical knitted fabric yarn formulas and the real-time knitted fabric yarn formulas; A feature extraction unit, using mutual information to identify the correlation between the historical knitted fabric yarn formula and the knitted weaving performance data, and extract significant features; using the significant features, generating nonlinear correlation features based on a nonlinear model, wherein the nonlinear correlation features include mutual influences between features; A performance prediction unit, splicing the historical knitted fabric yarn formula, the significant features and the nonlinear related features to obtain splicing features, and calculating polynomial features corresponding to the historical data based on the splicing features; inputting the polynomial features into a performance prediction model for model training; A weavability verification unit calculates the real-time knitted fabric yarn formula to obtain polynomial features corresponding to the real-time data; inputs the polynomial features into the trained performance prediction model to obtain a weaving performance prediction result; establishes a weavability verification system, performs weighted fusion on each performance result in the weaving performance prediction result to generate a weavability score; and verifies weavability based on the weavability score.

[0013] Furthermore, the process of acquiring the significant features is as follows: The historical knitted fabric yarn formula and the knitted weaving performance data are defined as a feature set and a target variable, respectively; the marginal probability distribution of the feature set and the target variable is calculated, and the joint probability distribution of the feature set and the target variable is calculated; Substituting the marginal probability distribution and the joint probability distribution into the mutual information formula, the mutual information between the feature set and the target variable is obtained; the features are sorted according to the mutual information values, and the first N features with the largest mutual information are selected as significant features.

[0014] Furthermore, the process of obtaining the polynomial feature is as follows: Each feature in the spliced ​​features is taken as a monomial; the square of each feature in the spliced ​​features and the quadratic combination are taken as quadratic terms; the cube of each feature in the spliced ​​features, the combination of the square of each feature in the spliced ​​features and another feature, and the cubic combination of all features in the spliced ​​features are taken as cubic terms; the monomial, quadratic terms and cubic terms are combined to generate the polynomial feature.

[0015] Furthermore, the performance prediction model includes an input layer, a first hidden layer, a second hidden layer, a third hidden layer and an output layer; the input layer receives the polynomial features; the first hidden layer includes 256 neurons and a ReLU activation function; the second hidden layer includes 128 neurons and a ReLU activation function; the third hidden layer includes 64 neurons and a ReLU activation function; the output layer includes 1 neuron, and the output is the weaving performance prediction result.

[0016] Furthermore, in the weavability verification system, the specific process is: Each performance result in the weaving performance prediction result is selected, and a weavability score is calculated by geometric mean; a weavability score threshold is set, and the weavability is verified based on the weavability score threshold, which is divided into a first weavability, a second weavability and a third weavability.

[0017] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention uses mutual information to identify the correlation between historical knitted fabric yarn formula and weaving performance data, and can effectively extract features that are significantly related to weaving performance. This step can screen out the factors that have the greatest impact on performance by quantifying the interdependence between features. This helps to reduce the dimensionality of the feature space, improve the training efficiency of the model, and improve the accuracy of the prediction results.

[0018] 2. The present invention combines historical knitted fabric yarn formulas, significant features and nonlinear related features and calculates polynomial features, which can effectively expand the feature space and enhance the expressive power of the model. Through nonlinear transformation, more complex relationships between variables are captured, and the influence of each feature on weaving performance is more comprehensively reflected. The introduction of polynomial features enables the model to adapt to various nonlinear situations more flexibly and improve the accuracy of prediction.

[0019] 3. The present invention converts the prediction results into weavability scores, so that the weavability verification system can timely identify the advantages and disadvantages of different yarn formulas and assist in decision-making. In addition, this verification system can integrate various weaving performance indicators, comprehensively consider multiple performance factors, improve the comprehensiveness and accuracy of the evaluation, optimize the production process, and improve weaving efficiency and product quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A method flow chart of a knitted fabric weavability verification method of the present invention; Figure 2 A model structure diagram of the performance prediction model of the present invention; Figure 3 The system structure diagram of a knitted fabric weavability verification system of the present invention. DETAILED DESCRIPTION

[0021] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] In order to improve the accuracy of knitting fabric weavability verification, the present invention provides a knitting fabric weavability verification method and system. In order to illustrate the function of the present invention, the effectiveness of the present invention will be described from the following examples.

[0023] Embodiment 1 A Knitting Co., Ltd. is dedicated to the production and manufacturing of knitted and crocheted fabrics and their products. Accurate weavability verification is crucial to the production of high-quality knitted fabrics. In order to quickly evaluate the impact of different yarn formulas on weaving performance, the company used a knitted fabric weavability verification method. Figure 1 As shown, a method flow chart of a knitted fabric weavability verification method is presented.

[0024] Reference Figure 1 In step S01, historical and real-time knitted fabric yarn formulas are collected, and weaving performance data corresponding to the historical knitted fabric yarn formulas are collected; data preprocessing is performed on the historical knitted fabric yarn formulas and the real-time knitted fabric yarn formulas. As shown in Table 1, the formulas and performance data of some yarn types are displayed.

[0025] Table 1 Yarn type formulation and performance data

[0026] Specifically, the preprocessing steps include data cleaning: removing duplicate records, missing values ​​and outliers in historical and real-time knitted fabric yarn recipes, as well as weaving performance data; standardization: standardizing different features to make them on the same scale.

[0027] By collecting historical and real-time yarn recipe and weaving performance data, a comprehensive data foundation is provided for the model, ensuring the diversity and consistency of information, thereby improving the accuracy and reliability of subsequent analysis.

[0028] Further, refer to Figure 1 In step S02, mutual information is used to identify the correlation between historical knitted fabric yarn formula and knitting weaving performance data, and significant features are extracted; and nonlinear correlation features are generated based on a nonlinear model using the significant features.

[0029] Specifically, the process of obtaining salient features is as follows: The historical knitted fabric yarn formula and the knitted weaving performance data are respectively defined as feature sets and the target variable ; Calculate the joint probability distribution of the feature set and the target variable: Calculate the feature set and the target variable For each pair of variables The frequency of simultaneous occurrence is divided by the total number of samples to obtain the joint probability, which is given by: ,in, Representation variables and The number of times it appears at the same time, represents the total number of samples; By summing the joint probability distribution, the feature set is calculated and the target variable Marginal probability distribution of variables in: For the feature set Medium Variable The marginal distribution probability can be obtained by the variables of the joint probability distribution Summing it up, we get Similarly, for the variable The marginal probability distribution of can be calculated as .

[0030] Substitute the marginal probability distribution and the joint probability distribution into the mutual information formula to obtain the feature set and the target variable The mutual information between: ; in, Representation feature set and the target variable The mutual information between represents the joint probability distribution; and represents the marginal probability distribution; The features are sorted according to the mutual information value, and the first N features with the largest mutual information are selected as significant features.

[0031] By using mutual information to identify the correlation between historical knitted fabric yarn formula and weaving performance data, the features that have a significant impact on weaving performance can be effectively extracted. This process not only improves the scientific nature of feature selection, but also reduces redundant information and improves the training efficiency and prediction accuracy of subsequent models.

[0032] Furthermore, the nonlinear correlation features are generated based on the nonlinear model using the significant features. In this embodiment, the nonlinear model can be implemented using nonlinear formulas such as logarithmic function, square function, sine function, etc., or a combination of these nonlinear formulas.

[0033] Further, refer to Figure 1In step S03, historical knitted fabric yarn formulas, significant features and nonlinear related features are spliced ​​to obtain splicing features, and polynomial features are calculated based on the splicing features; the polynomial features are input into the performance prediction model for model training, and the weaving performance prediction results are output.

[0034] Specifically, the historical knitted fabric yarn formula, significant features and nonlinear related features are spliced ​​to obtain the splicing features , splicing features is a feature vector. In this embodiment, the splicing feature is selected Two of the features and As a feature example, the process of generating polynomial features is described. The actual splicing feature All features in the and steps to generate polynomial features.

[0035] Specifically, the characteristics and As a monomial and ; The feature and The square and quadratic combination of as quadratic terms include , and ; The cube of each feature, the combination of the square of each feature and another feature, and the cubic combination of all the features are taken as cubic terms, including , , and ; Combine the monomial, quadratic term and cubic term to generate the polynomial feature, which is .

[0036] By generating a combination of monomials, quadratic terms, and cubic terms, polynomial features enhance the expressiveness of the data, enabling the model to capture more complex nonlinear relationships, improve prediction accuracy, and provide richer information for subsequent analysis.

[0037] Furthermore, the polynomial features are input into the performance prediction model for model training, and the weaving performance prediction results are output. Figure 2As shown, the model structure diagram of the performance prediction model is shown. The performance prediction model includes an input layer, a first hidden layer, a second hidden layer, a third hidden layer and an output layer; the input layer receives the polynomial features; the first hidden layer includes 256 neurons and a ReLU activation function; the second hidden layer includes 128 neurons and a ReLU activation function; the third hidden layer includes 64 neurons and a ReLU activation function; the output layer includes 1 neuron, and the output is the weaving performance prediction result. As shown in Table 2, some weaving performance prediction results are shown, where the range of each indicator is 0-1, and the closer to 1, the better the performance of the indicator.

[0038] Table 2 Prediction results of some weaving properties

[0039] By setting up a multi-layer neural network, the model is able to extract features layer by layer and learn complex patterns. The ReLU activation function promotes nonlinear transformation, enhances the model's fitting ability, and improves the accuracy and robustness of the prediction of weaving performance.

[0040] Further, refer to Figure 1 In step S04, the real-time knitted fabric yarn formula is calculated to obtain polynomial features; the polynomial features are input into the trained performance prediction model to obtain the weaving performance prediction results; a weavability verification system is established to weightedly fuse the various performance results in the weaving performance prediction results to generate a weavability score; and the weavability is verified based on the weavability score.

[0041] Specifically, in the weavability verification system, the specific process is: Select each performance result in the weaving performance prediction result and calculate the weavability score by geometric mean : ;in, , , and The four performance results of strength, elasticity, wrinkle resistance and air permeability are shown respectively. In this embodiment, according to Table 2, the weavability scores of the five formulations can be calculated, as shown in Table 3.

[0042] Table 3 Some weavability scoring results

[0043] A weavability scoring threshold is set, and weavability is verified based on the weavability scoring threshold, and is divided into first weavability, second weavability, and third weavability. As shown in Table 4, the weavability scoring division is shown.

[0044] Table 4 Weavability score classification

[0045] As can be seen from Table 4, based on the weavability score, weavability verification can be divided into three intervals, and the first weavability, the second weavability, and the third weavability represent high weavability, medium weavability, and low weavability, respectively. In practical applications, a certain A Knitting Co., Ltd. selects a knitting formula with high weavability to produce high-performance fabrics.

[0046] By performing the same data operations on the real-time knitted fabric yarn formula as the historical formula, the consistency of data processing is ensured. The model can generate accurate weaving performance predictions based on the latest formula data, and the weavability evaluation system uses these prediction results for weighted fusion to form a weavability score, thereby quickly identifying and verifying the actual performance of the new formula, which helps to improve production efficiency and product quality.

[0047] A knitted fabric weavability verification method used in this embodiment combines historical and real-time knitted fabric yarn formulas and weaving performance data, uses mutual information to identify the correlation between features, and generates new features through nonlinear models, thereby improving the accuracy of weaving performance prediction. Through the combination of polynomial features and performance prediction models, the weaving performance of different yarn formulas can be predicted, and then a weavability evaluation system can be established to quantitatively verify the weavability of the fabric. This method can effectively improve the design efficiency and quality control of knitted fabrics.

[0048] Embodiment 2 In order to share fabric performance data with suppliers and ensure the consistency of raw material selection and processing technology, a fabric processing company B selected a knitted fabric weavability verification system. Figure 3 As shown, a system structure diagram of a knitted fabric weavability verification system is shown.

[0049] Reference Figure 3 The system includes a data acquisition and processing unit, which collects historical and real-time knitted fabric yarn formulas, including yarn types, fiber components, yarn thickness, etc., and collects weaving performance data corresponding to historical knitted fabric yarn formulas, including strength, elasticity, wrinkle resistance and air permeability; and performs data preprocessing on the historical knitted fabric yarn formulas and the real-time knitted fabric yarn formulas.

[0050] Specifically, the preprocessing steps include data cleaning: removing duplicate records, missing values ​​and outliers in historical and real-time knitted fabric yarn recipes, as well as weaving performance data; standardization: standardizing different features to make them on the same scale.

[0051] By collecting historical and real-time yarn recipe and weaving performance data, a comprehensive data foundation is provided for the model, ensuring the diversity and consistency of information, thereby improving the accuracy and reliability of subsequent analysis.

[0052] The system also includes a feature extraction unit, which uses mutual information to identify the correlation between historical knitted fabric yarn formulas and the knitted weaving performance data to extract significant features; and uses the significant features to generate nonlinear correlation features based on a nonlinear model.

[0053] Specifically, the process of obtaining salient features is as follows: The historical knitted fabric yarn formula and the knitted weaving performance data are respectively defined as feature sets and the target variable ; Calculate the joint probability distribution of the feature set and the target variable: Calculate the feature set and the target variable For each pair of variables The frequency of simultaneous occurrence is divided by the total number of samples to obtain the joint probability, which is given by: ,in, Representation variables and The number of times it appears at the same time, represents the total number of samples; By summing the joint probability distribution, the feature set is calculated and the target variable Marginal probability distribution of variables in: For the feature set Medium Variable The marginal distribution probability can be obtained by the variables of the joint probability distribution Summing it up, we get Similarly, for the variable The marginal probability distribution of can be calculated as .

[0054] Substitute the marginal probability distribution and the joint probability distribution into the mutual information formula to obtain the feature set and the target variable The mutual information between: ; in, Representation feature set and the target variable The mutual information between represents the joint probability distribution; and represents the marginal probability distribution; The features are sorted according to the mutual information value, and the first N features with the largest mutual information are selected as significant features.

[0055] Furthermore, the nonlinear correlation features are generated based on the nonlinear model using the significant features. In this embodiment, the nonlinear model can be implemented using nonlinear formulas such as logarithmic function, square function, sine function, etc., or a combination of these nonlinear formulas.

[0056] The system also includes a performance prediction unit, which splices the historical knitted fabric yarn formula, the significant features and the nonlinear related features to obtain splicing features, and calculates polynomial features corresponding to historical data based on the splicing features; and inputs the polynomial features into a performance prediction model for model training.

[0057] Specifically, the historical knitted fabric yarn formula, significant features and nonlinear related features are spliced ​​to obtain the splicing features , splicing features is a feature vector. In this embodiment, the splicing feature is selected Two of the features and As a feature example, the process of generating polynomial features is described. The actual splicing feature All features in the and steps to generate polynomial features.

[0058] Specifically, the characteristics and As a monomial and ; The feature and The square and quadratic combination of as quadratic terms include , and ; The cube of each feature, the combination of the square of each feature and another feature, and the cubic combination of all the features are taken as cubic terms, including , , and ; Combine the monomial, quadratic term and cubic term to generate the polynomial feature, which is .

[0059] Furthermore, the polynomial features are input into the performance prediction model for model training, and the weaving performance prediction results are output. The performance prediction model includes an input layer, a first hidden layer, a second hidden layer, a third hidden layer and an output layer; the input layer receives the polynomial features; the first hidden layer includes 256 neurons and a ReLU activation function; the second hidden layer includes 128 neurons and a ReLU activation function; the third hidden layer includes 64 neurons and a ReLU activation function; the output layer includes 1 neuron, and the output is the weaving performance prediction result.

[0060] The system also includes a weavability verification unit, which calculates the real-time knitted fabric yarn formula to obtain polynomial features corresponding to the real-time data; inputs the polynomial features into the trained performance prediction model to obtain weaving performance prediction results; establishes a weavability verification system, performs weighted fusion on each performance result in the weaving performance prediction results to generate a weavability score; and verifies weavability based on the weavability score.

[0061] Specifically, in the weavability verification system, the specific process is: Select each performance result in the weaving performance prediction result and calculate the weavability score by geometric mean : ; in, , , and They represent the four performance results of strength, elasticity, wrinkle resistance and breathability respectively.

[0062] A weavability scoring threshold is set, and weavability is verified based on the weavability scoring threshold, which is divided into first weavability, second weavability and third weavability.

[0063] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for verifying the weavability of a knitted fabric, characterized in that: include: Collect historical and real-time knitted fabric yarn formulas, including yarn type, fiber composition and yarn thickness, and collect corresponding weaving performance data of historical knitted fabric yarn formulas, including strength, elasticity, wrinkle resistance and breathability; Performing data preprocessing on the historical knitted fabric yarn formula and the real-time knitted fabric yarn formula; Using mutual information to identify the correlation between historical knitted fabric yarn formula and the knitted weaving performance data, and extracting significant features; Using the significant features, generating nonlinear correlation features based on a nonlinear model, wherein the nonlinear correlation features include mutual influences between features; Splicing the historical knitted fabric yarn formula, the significant features and the nonlinear related features to obtain splicing features, and calculating polynomial features corresponding to the historical data based on the splicing features; inputting the polynomial features into a performance prediction model for model training; The real-time knitted fabric yarn formula is calculated to obtain polynomial features corresponding to the real-time data; the polynomial features are input into the trained performance prediction model to obtain weaving performance prediction results; a weavability verification system is established to perform weighted fusion on each performance result in the weaving performance prediction results to generate a weavability score; and weavability is verified based on the weavability score.

2. A knitted fabric weavability verification method according to claim 1, characterized in that: The process of obtaining the salient features is as follows: The historical knitted fabric yarn formula and the knitted weaving performance data are defined as a feature set and a target variable, respectively; the marginal probability distribution of the feature set and the target variable is calculated, and the joint probability distribution of the feature set and the target variable is calculated; Substituting the marginal probability distribution and the joint probability distribution into the mutual information formula, the mutual information between the feature set and the target variable is obtained; the features are sorted according to the mutual information values, and the first N features with the largest mutual information are selected as significant features.

3. A knitted fabric weavability verification method according to claim 1, characterized in that: The process of obtaining the polynomial features is as follows: Each feature in the spliced ​​features is taken as a monomial; the square of each feature in the spliced ​​features and the quadratic combination are taken as quadratic terms; the cube of each feature in the spliced ​​features, the combination of the square of each feature in the spliced ​​features and another feature, and the cubic combination of all features in the spliced ​​features are taken as cubic terms; the monomial, quadratic terms and cubic terms are combined to generate the polynomial feature.

4. A knitted fabric weavability verification method according to claim 1, characterized in that: The performance prediction model includes an input layer, a first hidden layer, a second hidden layer, a third hidden layer and an output layer; the input layer receives the polynomial features; the first hidden layer includes 256 neurons and a ReLU activation function; the second hidden layer includes 128 neurons and a ReLU activation function; the third hidden layer includes 64 neurons and a ReLU activation function; the output layer includes 1 neuron, and the output is the weaving performance prediction result.

5. A knitted fabric weavability verification method according to claim 1, characterized in that: In the weavability verification system, the specific process is: Each performance result in the weaving performance prediction result is selected, and a weavability score is calculated by geometric mean; a weavability score threshold is set, and the weavability is verified based on the weavability score threshold, which is divided into a first weavability, a second weavability and a third weavability.

6. A knitted fabric weavability verification system, characterized in that: include: Data collection and processing unit, which collects historical and real-time knitted fabric yarn formulas, including yarn type, fiber composition and yarn thickness, and collects corresponding weaving performance data of historical knitted fabric yarn formulas, including strength, elasticity, wrinkle resistance and breathability; Performing data preprocessing on the historical knitted fabric yarn formula and the real-time knitted fabric yarn formula; A feature extraction unit, using mutual information to identify the correlation between the historical knitted fabric yarn formula and the knitted weaving performance data, and extract significant features; using the significant features, generating nonlinear correlation features based on a nonlinear model, wherein the nonlinear correlation features include mutual influences between features; A performance prediction unit, splicing the historical knitted fabric yarn formula, the significant features and the nonlinear related features to obtain splicing features, and calculating polynomial features corresponding to the historical data based on the splicing features; inputting the polynomial features into a performance prediction model for model training; A weavability verification unit calculates the real-time knitted fabric yarn formula to obtain polynomial features corresponding to the real-time data; inputs the polynomial features into the trained performance prediction model to obtain a weaving performance prediction result; establishes a weavability verification system, performs weighted fusion on each performance result in the weaving performance prediction result to generate a weavability score; and verifies weavability based on the weavability score.

7. A knitted fabric weavability verification system according to claim 6, characterized in that: The process of obtaining the salient features is as follows: The historical knitted fabric yarn formula and the knitted weaving performance data are defined as a feature set and a target variable, respectively; the marginal probability distribution of the feature set and the target variable is calculated, and the joint probability distribution of the feature set and the target variable is calculated; Substituting the marginal probability distribution and the joint probability distribution into the mutual information formula, the mutual information between the feature set and the target variable is obtained; the features are sorted according to the mutual information values, and the first N features with the largest mutual information are selected as significant features.

8. A knitted fabric weavability verification system according to claim 6, characterized in that: The process of obtaining the polynomial features is as follows: Each feature in the spliced ​​features is taken as a monomial; the square of each feature in the spliced ​​features and the quadratic combination are taken as quadratic terms; the cube of each feature in the spliced ​​features, the combination of the square of each feature in the spliced ​​features and another feature, and the cubic combination of all features in the spliced ​​features are taken as cubic terms; the monomial, quadratic terms and cubic terms are combined to generate the polynomial feature.

9. A knitted fabric weavability verification system according to claim 6, characterized in that: The performance prediction model includes an input layer, a first hidden layer, a second hidden layer, a third hidden layer and an output layer; the input layer receives the polynomial features; the first hidden layer includes 256 neurons and a ReLU activation function; the second hidden layer includes 128 neurons and a ReLU activation function; the third hidden layer includes 64 neurons and a ReLU activation function; the output layer includes 1 neuron, and the output is the weaving performance prediction result.

10. A knitted fabric weavability verification system according to claim 6, characterized in that: In the weavability verification system, the specific process is: Each performance result in the weaving performance prediction result is selected, and a weavability score is calculated by geometric mean; a weavability score threshold is set, and the weavability is verified based on the weavability score threshold, which is divided into a first weavability, a second weavability and a third weavability.

Images