High-elasticity wear-resistant knitted jean fabric and preparation method thereof

Abstract

The invention provides a high-elasticity wear-resistant knitted jean fabric. The high-elasticity wear-resistant knitted jean fabric comprises polyurethane elastic fibers, polyester fibers, cotton fibers and regenerated cellulose fibers. Meanwhile, the invention further provides an intelligent and automatic method for preparing the fabric, and dynamic optimization of the production process is achieved by monitoring the yarn performance in real time, predicting the change trend and combining fabric structure density analysis. The high-elasticity wear-resistant environment-friendly knitted denim fabric has the advantages that the knitted denim fabric with high elasticity, wear resistance and environment-friendly characteristics is successfully developed through a scientific and reasonable formula design and an advanced production technology, and the knitted denim fabric can meet diversified market requirements and lead the industry development trend.

Description

Technical Field

[0001] The invention belongs to the technical field of fabrics and industrialized automated production, and specifically relates to a high-elasticity, wear-resistant knitted denim fabric and a preparation method thereof. Background Art

[0002] In the textile industry, knitted denim fabrics are widely popular for their unique style and comfort. However, existing knitted denim fabric technology has some significant shortcomings, primarily in the following areas: 1. Insufficient elasticity: Traditional knitted denim fabrics typically use cotton fiber as their primary raw material. While this material offers excellent softness and breathability, it suffers from poor elasticity and ductility. This makes it prone to the fabric failing to return to its original shape after being stretched, impacting the garment's fit and comfort. 2. Poor abrasion resistance: Due to the knitted structure and traditional yarn selection, knitted denim fabrics are prone to wear and tear, pilling, and other issues under high-intensity conditions (such as frequent friction or prolonged wear). These issues not only reduce the fabric's lifespan but also affect the product's appearance and quality. 3. Difficulty ensuring product consistency: Traditional textile processes rely on manual operation and empirical judgment, making it prone to fluctuations in product quality due to insufficient equipment precision or improper operation during the production process. Furthermore, color and thickness variations between batches are common, posing a challenge for both brands and consumers.

[0003] In the fabric field, researchers have also been conducting in-depth research to address the aforementioned technical issues. For example, reference patent CN216001752U discloses a heat-storage knitted denim fabric that enhances thermal insulation by adding a multi-layer structure. However, the fabric's elasticity and wear resistance are not optimized, resulting in limitations in comfort and durability. Separately, reference patent CN210796802U proposes a knitted, four-way stretch denim fabric that improves elasticity by optimizing the structure of elastic cords. However, its design does not specifically focus on wear resistance, and the improvement in elasticity may sacrifice other fabric properties, such as strength and comfort.

[0004] In response to the above problems, this application is committed to developing a highly elastic and wear-resistant knitted denim fabric, aiming to solve the problems of poor elasticity and wear resistance of knitted denim fabric in the existing technology. By changing the formula composition of the fabric and introducing intelligent preparation methods, it is hoped to produce a denim fabric with excellent elasticity and wear resistance, while ensuring product consistency and production efficiency. Summary of the Invention

[0005] The purpose of the present invention is to solve the above-mentioned technical problems in the prior art.

[0006] In order to achieve the above-mentioned purpose, the technical solution provided by the present invention is a highly elastic and wear-resistant knitted denim fabric, comprising polyurethane elastic fiber, polyester fiber, cotton fiber, and regenerated cellulose fiber. The mass percentages of each component in the fabric are 10-20% polyurethane elastic fiber, 10-20% polyester fiber, 40-50% cotton fiber, and 5-10% regenerated cellulose fiber. The polyurethane elastic fiber used in the present invention provides excellent elasticity and resilience. High-strength polyester fiber (High Tenacity Polyester) enhances the wear resistance and tear resistance of the fabric. Cotton fiber ensures a soft touch and natural breathability. Regenerated cellulose fiber increases skin-friendliness.

[0007] Furthermore, the regenerated cellulose fiber is lyocell fiber or modal fiber.

[0008] Furthermore, it also includes nano-silica, antimicrobial agents, and UV protectants. Nano-silica coating: used for surface treatment to improve wear resistance and stain resistance. Antimicrobial agents impart a long-lasting antimicrobial effect to fabrics. UV protectants: by adding organic UV absorbers, they protect the skin from UV damage.

[0009] The present invention also provides a method for preparing the fabric intelligent automation, comprising the following steps:

[0010] (1) The yarn thickness, twist and tension data are collected through sensors, and the yarn performance is judged to be within the target range based on the preset threshold value to obtain the real-time yarn status parameters;

[0011] (2) Based on the real-time yarn state parameters, a long short-term memory network model is used to predict the yarn performance change trend. The input layer contains three features: diameter, twist, and tension. The time step is set to 10, the number of hidden layer neurons is 32, the learning rate is 0.001, and the training data is the sampling data in the past hour. The yarn performance changes in the next 5 minutes are predicted;

[0012] (3) Obtain elasticity and functionality data from the fabric structure density obtained from the textile machine, determine whether the target elasticity requirements are met, and obtain fabric performance parameters;

[0013] (4) Based on the fabric performance parameters, the quality consistency of the finished product is analyzed through the online quality inspection system, and the resource waste rate is predicted in combination with historical data to determine the production process optimization plan, thereby obtaining a highly elastic and wear-resistant knitted denim fabric.

[0014] Furthermore, the specific operation steps of step (1) are as follows: (11) collecting yarn thickness, twist, and tension data through a sensor to generate a raw data set;

[0015] (12) Using signal processing technology to denoise the original data set to obtain a smoothed data set;

[0016] (13) If the yarn thickness, twist, and tension values ​​in the smoothed data set are compared with the preset thresholds, a logical judgment is performed to determine whether each parameter is within the target range, and the performance compliance state is obtained;

[0017] (14) According to the performance compliance status, the decision tree algorithm is used to classify the yarn performance and obtain the classification results;

[0018] (15) By comparing the classification results with historical data, the fluctuation trend of the performance parameters is calculated and the fluctuation characteristics are obtained;

[0019] (16) If the fluctuation characteristics exceed the preset fluctuation range, an anomaly detection algorithm is used to identify potential defects and obtain a defect mark;

[0020] (17) Based on the defect marks, real-time status parameters are generated and the monitoring results of yarn performance are output.

[0021] Furthermore, the specific operation steps of step (3) are:

[0022] (31) obtaining fabric structure density data through textile machine sensors and storing the data as an initial data set;

[0023] (32) Data preprocessing techniques are used to denoise and standardize the initial dataset to obtain a structured density dataset;

[0024] (33) Extracting elastic parameters and functional parameters from the structured density data set to generate a fabric characteristic parameter set;

[0025] (34) If the difference between the elastic parameter in the fabric characteristic parameter set and the preset target elastic requirement is less than a preset threshold, it is determined that the requirement is met and an elastic compliance mark is generated;

[0026] (35) The fabric characteristic parameter set is classified by random forest algorithm to determine the fabric performance index category;

[0027] (36) Generate comprehensive fabric performance evaluation results based on elastic compliance marks and performance index categories;

[0028] (37) If the comprehensive evaluation results of fabric performance show that the performance index category is consistent with the target elasticity requirement, the final fabric performance index is output.

[0029] Furthermore, the specific operation steps of step (4) are as follows: (41) collecting performance parameters of high-elasticity and wear-resistant knitted denim fabric through an online quality detection system, and using data cleaning technology to denoise and standardize the collected performance parameters to obtain a consistent data set;

[0030] (42) Extracting quality characteristic parameters from the consistency data set, combining them with pre-stored historical data, and applying the support vector machine algorithm to analyze the quality characteristic parameters and historical data to generate a resource waste rate indicator;

[0031] (43) If the resource waste rate index is greater than the preset threshold, the consistent data set is classified by cluster analysis technology to determine the optimization plan of the production process and obtain the optimized adjustment parameters;

[0032] (44) updating the equipment configuration parameters of the production process according to the optimized adjustment parameters, and generating an updated equipment operation data set;

[0033] (45) Extracting operation status characteristics from the updated equipment operation data set, using statistical analysis technology to evaluate the operation status characteristics, judging the stability of the production process, and obtaining stability evaluation results;

[0034] (46) If the stability assessment results show that the production process does not meet the preset stability standards, the optimization plan is adjusted according to the operating status characteristics to generate new adjustment parameters;

[0035] (47) The production process is optimized again through new adjustment parameters to generate the final production optimization data set.

[0036] Utilize advanced intelligent textile equipment and technical means to optimize production processes and ensure product consistency: by real-time monitoring of yarn performance and prediction of changing trends, combined with fabric structure density analysis, dynamic optimization of the production process is achieved. Specifically, the present invention collects yarn thickness, twist and tension data, uses a long short-term memory network model to predict changes in yarn performance, and simultaneously obtains fabric elasticity and functionality data to determine whether the target requirements are met. Based on these parameters, the present invention performs online quality detection and resource waste rate prediction to determine the optimal production plan. This method can effectively improve the production efficiency and quality stability of high-performance knitted denim fabrics, reduce resource waste, and provide a new technical path for intelligent textile production.

[0037] In summary, the present invention has successfully developed a knitted denim fabric with high elasticity, wear resistance and environmental protection characteristics through scientific and reasonable formula design and advanced production technology, which can meet diverse market demands and lead the development trend of the industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 This is a production process roadmap for Example 1 of the present invention. DETAILED DESCRIPTION

[0039] Example 1: The present invention provides a high elasticity wear-resistant knitted denim fabric, which is composed of the following components

[0040] composition:

[0041] Element content Polyurethane elastic fiber 10％ polyester fiber 10％ cotton fiber 50％ Lyocell fiber 5％ Nanosilica 1％ Silver ions margin Benzophenone-4 1％

[0042] The process for preparing the denim fabric comprises: Figure 1 As shown: It includes the following steps:

[0043] (1) The yarn thickness, twist and tension data are collected through sensors, and the yarn performance is judged to be within the target range based on the preset threshold value to obtain the real-time yarn status parameters;

[0044] (2) Based on the real-time yarn state parameters, a long short-term memory network model is used to predict the yarn performance change trend. The input layer contains three features: diameter, twist, and tension. The time step is set to 10, the number of hidden layer neurons is 32, the learning rate is 0.001, and the training data is the sampling data in the past hour. The yarn performance changes in the next 5 minutes are predicted;

[0045] (3) Obtain elasticity and functionality data from the fabric structure density obtained from the textile machine, determine whether the target elasticity requirements are met, and obtain fabric performance parameters;

[0046] (4) Based on the fabric performance parameters, the quality consistency of the finished product is analyzed through the online quality inspection system, and the resource waste rate is predicted in combination with historical data to determine the production process optimization plan, thereby obtaining a highly elastic and wear-resistant knitted denim fabric.

[0047] The specific operation steps of step (1) are as follows: (11) collecting yarn thickness, twist, and tension data through sensors to generate a raw data set;

[0048] (12) Using signal processing technology to denoise the original data set to obtain a smoothed data set;

[0049] (13) If the yarn thickness, twist, and tension values ​​in the smoothed data set are compared with the preset thresholds, a logical judgment is performed to determine whether each parameter is within the target range, and the performance compliance state is obtained;

[0050] (14) According to the performance compliance status, the decision tree algorithm is used to classify the yarn performance and obtain the classification results;

[0051] (15) By comparing the classification results with historical data, the fluctuation trend of the performance parameters is calculated and the fluctuation characteristics are obtained;

[0052] (16) If the fluctuation characteristics exceed the preset fluctuation range, an anomaly detection algorithm is used to identify potential defects and obtain a defect mark;

[0053] (17) Based on the defect marks, real-time status parameters are generated and the monitoring results of yarn performance are output.

[0054] The specific operation steps of step (3) are:

[0055] (31) obtaining fabric structure density data through textile machine sensors and storing the data as an initial data set;

[0056] (32) Data preprocessing techniques are used to denoise and standardize the initial dataset to obtain a structured density dataset;

[0057] (33) Extracting elastic parameters and functional parameters from the structured density data set to generate a fabric characteristic parameter set;

[0058] (34) If the difference between the elastic parameter in the fabric characteristic parameter set and the preset target elastic requirement is less than a preset threshold, it is determined that the requirement is met and an elastic compliance mark is generated;

[0059] (35) The fabric characteristic parameter set is classified by random forest algorithm to determine the fabric performance index category;

[0060] (36) Generate comprehensive fabric performance evaluation results based on elastic compliance marks and performance index categories;

[0061] (37) If the comprehensive evaluation results of fabric performance show that the performance index category is consistent with the target elasticity requirement, the final fabric performance index is output.

[0062] The specific operation steps of step (4) are as follows: (41) collecting performance parameters of high-elasticity and wear-resistant knitted denim fabric through an online quality detection system, and using data cleaning technology to denoise and standardize the collected performance parameters to obtain a consistent data set;

[0063] (42) Extracting quality characteristic parameters from the consistency data set, combining them with pre-stored historical data, and applying the support vector machine algorithm to analyze the quality characteristic parameters and historical data to generate a resource waste rate indicator;

[0064] (43) If the resource waste rate index is greater than the preset threshold, the consistent data set is classified by cluster analysis technology to determine the optimization plan of the production process and obtain the optimized adjustment parameters;

[0065] (44) updating the equipment configuration parameters of the production process according to the optimized adjustment parameters, and generating an updated equipment operation data set;

[0066] (45) Extracting operation status characteristics from the updated equipment operation data set, using statistical analysis technology to evaluate the operation status characteristics, judging the stability of the production process, and obtaining stability evaluation results;

[0067] (46) If the stability assessment results show that the production process does not meet the preset stability standards, the optimization plan is adjusted according to the operating status characteristics to generate new adjustment parameters;

[0068] (47) The production process is optimized again through new adjustment parameters to generate the final production optimization data set.

[0069] Sensors collecting data on yarn thickness, twist, and tension are fundamental to yarn performance monitoring. These sensors are typically deployed on spinning equipment to record yarn parameters in real time. For example, laser diameter gauges can measure yarn thickness with an accuracy of 0.01 mm; photoelectric sensors detect twist and can record 1000 twists per minute; and tension sensors measure yarn tension in a range of 0.1 to 10 Newtons. Raw data sets contain timestamps and parameter values, such as the yarn thickness of 0.2 mm, twist of 800 twists / meter, and tension of 2 Newtons at a given moment. These data may fluctuate abnormally due to equipment vibration or ambient noise. In one possible implementation, signal processing techniques are used for denoising. Preferably, wavelet transforms are used to decompose the signal, remove high-frequency noise, and preserve the trend characteristics of the yarn parameters. For example, a segment of yarn thickness data contains random noise. After wavelet denoising, the data curve becomes smoother, with the standard deviation reduced from 0.05 mm to 0.02 mm. Smoothed data sets more accurately reflect the true state of the yarn and facilitate subsequent analysis. Specifically, smoothed data is compared against preset thresholds to determine whether parameters meet the target. For example, the thresholds for yarn thickness are 0.18 to 0.22 mm, twist 750 to 850 twists / m, and tension 1.5 to 2.5 Newtons. If a data point shows thickness of 0.19 mm, twist 820 twists / m, and tension 2.2 Newtons, a logical analysis confirms that all parameters are within the target range and the data point is marked as qualified. It should be noted that the thresholds are set based on historical experience and product requirements to ensure quality consistency. In one embodiment, a decision tree algorithm is used to classify yarn properties. The decision tree divides nodes based on thickness, twist, and tension values ​​to generate classification rules. For example, a thickness less than 0.18 mm may indicate insufficient strength and be marked as unqualified; excessive twist may affect softness and be marked as suboptimal. After classification of a batch of yarn data using the decision tree, 80% was marked as high-quality, 15% as suboptimal, and 5% as unqualified. This classification method is intuitive and efficient, making it easy to trace the root cause of any problem. For example, comparing the classification results with historical data can identify fluctuation trends. Historical data shows that the thickness fluctuation range of high-quality yarn is ±0.02 mm. If the current batch fluctuates by ±0.03 mm, it indicates unstable performance. Fluctuation characteristics are obtained through statistical analysis, such as standard deviation or range, which reflect the reliability of the production process. It is understandable that the anomaly detection algorithm is used to identify potential defects. Preferably, the isolation forest algorithm is used to detect abnormal points based on fluctuation characteristics. For example, at a certain moment, the tension suddenly increases to 3 Newtons, which exceeds the normal fluctuation range. The algorithm marks it as a potential defect, which may be caused by equipment failure. Defect marking improves the efficiency of problem location and reduces the defective rate. In one embodiment, real-time status parameters are generated based on defect marking, including thickness, twist, tension and defect type, and output to the monitoring system. For example, a batch of yarn shows abnormal tension, and the monitoring results prompt to check the tension wheel. Real-time monitoring ensures rapid response and improves production efficiency and product quality.

[0070] A technical solution that uses a long short-term memory (LSTM) model to predict yarn performance trends based on real-time collected yarn state parameters has significant application value in the yarn production field. LSTM is a neural network suitable for processing time series data, capable of capturing the dynamic characteristics of yarn parameters over time. For example, LSTM uses memory cells and a gating mechanism to preserve long-term dependencies in historical data while filtering out irrelevant noise, ensuring prediction accuracy. The input layer contains three features: diameter, twist, and tension. The time step is 10, meaning the model uses data from the previous 10 time points to predict future trends. The number of hidden layer neurons is 32, and the learning rate is 0.001, ensuring a balance between model complexity and training stability. Training data is sampled from the past hour, and performance changes predicted for the next 5 minutes provide real-time basis for production adjustments. Specifically, diameter data can be collected using a high-precision optical caliper with a sampling frequency of 500 Hz and a measurement range of 0.05-1.5 mm. In one embodiment, the sensor collects 500 diameter data points per second. The LSTM model organizes this data into a time series and analyzes the pattern of diameter changes over time. For example, the model may discover periodic fluctuations in diameter over certain time periods, indicating mechanical vibration in the spinning machine and providing a reference for equipment maintenance. It should be noted that preprocessing of the diameter data requires removing outliers, such as using a median filter to eliminate sudden changes, to ensure stable feature data input to the LSTM. In one possible implementation, twist data is collected using a high-resolution encoder that measures the number of twists per 10 cm of yarn, ranging from 20 to 80 twists per meter. The LSTM analyzes the twist sequence to predict twist trends over the next five minutes. For example, the model might detect that twist is gradually deviating from the target value, prompting the operator to adjust the drafting parameters of the spinning machine. Preferably, the twist data input feature can incorporate ambient humidity, as humidity can affect the physical properties of yarn fibers and, thus, indirectly influence twist stability. For example, tension data can be collected using a high-sensitivity strain sensor with a range of 0-400 cN. The LSTM model uses the tension sequence to predict whether the tension will exceed a set range, such as 70-90 cN, within the next five minutes. In one embodiment, if the model predicts that the tension continues to rise, it may indicate an imminent risk of yarn breakage, and the system will issue an early warning and recommend reducing the spinning speed. It should be noted that the fluctuation of tension data may be related to the wear of the traction wheel of the spinning machine. The prediction results of LSTM can provide data support for equipment diagnosis. It is understandable that the choice of a time step of 10 balances the model's reliance on historical data and computational efficiency. In one embodiment, the time step can be dynamically adjusted according to the production speed, such as shortening it to 8 steps during high-speed spinning to capture finer-grained changes. The training data uses sampling data from the past hour, covering a sufficient production cycle to ensure that the model learns the regular characteristics of yarn performance.Setting a learning rate of 0.001 prevents the model from falling into a local optimum during training, and is suitable for scenarios where data distribution is relatively stable in yarn production. In one possible implementation, the output of the LSTM model can be linked to the production control system. For example, if the prediction results show that the diameter may be too small in the next 5 minutes, the system can automatically adjust the amount of fiber fed to optimize the yarn quality. Preferably, the model's prediction results will also be stored in a database for subsequent quality analysis, such as tracing the causes of performance defects in a batch of yarn. For example, the prediction capability of LSTM can also be extended to multi-feature joint analysis, such as comprehensively analyzing changes in diameter, twist, and tension to determine whether the overall performance of the yarn meets the requirements of high-end textiles. Through the above solution, the LSTM model provides an intelligent prediction tool for yarn production, capturing performance change trends from multiple dimensions such as diameter, twist, and tension, and providing data support for process optimization and quality control.

[0071] Acquiring fabric structural density data through textile machine sensors is fundamental to building a fabric performance evaluation system. Sensors are typically installed in key locations on textile machines, such as needles or yarn guides, to collect real-time warp and weft density information. For example, the sensors can use laser ranging or infrared scanning technology to collect data 1,000 times per second, recording the number of warp and weft yarns per square centimeter to form an initial dataset. For example, the initial dataset for a fabric sample shows a warp density of 120 yarns / cm and a weft density of 100 yarns / cm. In one possible implementation, data preprocessing techniques perform denoising and normalization on the initial dataset to eliminate noise interference and data bias. Denoising can employ a mean filter to remove outliers caused by sensor jitter. For example, density values ​​fluctuating by more than ±5% are replaced with the nearest mean. Normalization normalizes the data to the range 0-1 to ensure comparability across batches. This processing results in a structured density dataset, for example, normalizing the warp density to 0.6 and the weft density to 0.5. Specifically, extracting elastic parameters and functional parameters from the structured density data set is the key to generating a fabric characteristic parameter set. Elastic parameters include tensile modulus and recovery rate, which can be derived through correlation analysis between density and yarn material. For example, high-density fabrics usually have lower tensile modulus. Functional parameters such as air permeability and wear resistance are related to the uniformity of density distribution. In one embodiment, a fabric has a tensile modulus of 50N / cm, a recovery rate of 85%, and an air permeability of 200cm. 3 / s / cm 2, generating a characteristic parameter set. It should be noted that the generation of the elastic compliance flag relies on comparing the elastic parameters with the target elasticity requirements. Assuming the target tensile modulus is 48-52 N / cm and the threshold is ±2 N / cm, if the actual value is 50 N / cm, the difference meets the requirement, and a compliance flag is generated. If it does not meet the requirement, the flag is marked as non-compliant, which helps to quickly screen high-quality fabrics. In one embodiment, a random forest algorithm is used to classify the fabric characteristic parameter set and determine the performance indicator category. Random forest integrates multiple decision trees, analyzing the nonlinear relationships between characteristic parameters and classifying fabrics into three categories: high, medium, and low performance. For example, given tensile modulus, recovery rate, and breathability, the algorithm predicts a fabric as being in the high-performance category with an accuracy of 90%, improving classification reliability. Preferably, the comprehensive fabric performance evaluation results are generated by combining the elastic compliance flag and performance indicator category. If the flag is compliant and the category is high-performance, it is considered consistent with the target elasticity requirement, and the final performance indicator is output. For example, if a fabric has a tensile modulus of 50 N / cm and is classified as high-performance, the output indicator is "Excellent Elastic Fabric," which is used to guide production optimization. The advantage of this approach lies in its data-driven nature throughout the entire process. Sensor acquisition ensures real-time data, preprocessing improves data quality, and the random forest algorithm enhances classification accuracy. The final evaluation results provide clear guidance for production. For example, the screening of high-performance fabrics can reduce the defective rate by 10%, improving market competitiveness.

[0072] In the production of highly elastic and wear-resistant knitted denim fabric, an online quality inspection system uses sensors to collect performance parameters such as tensile strength, number of abrasions, and fabric thickness in real time. For example, the system uses a high-precision tension sensor to measure the stress of the fabric when stretched to 10%, recording it as 500N. Simultaneously, an abrasion tester records the weight loss of the fabric after 1000 abrasions, which is 0.5%. These parameters may contain noise due to equipment vibration or environmental interference. In one possible implementation, data cleaning technology uses median filtering to remove outliers, for example, removing records exceeding three standard deviations in tensile strength data, and normalizing thickness data to a mean of 0 and a standard deviation of 1 using Z-score standardization to generate a consistent data set. It should be noted that the cleaned data maintains the relative relationships between parameters, facilitating subsequent analysis. When extracting quality characteristic parameters from the consistent data set, tensile strength, abrasion loss rate, and thickness uniformity are preferably selected as core features. For example, a tensile strength of 500N corresponds to high elasticity requirements, a abrasion loss rate of 0.5% reflects wear resistance, and a thickness uniformity deviation of 2% indicates production consistency. Combining historical data, such as the characteristic parameters of fabrics from the past 100 batches, a support vector machine algorithm is applied to analyze the match between current and historical data to generate a resource waste rate indicator. For example, the algorithm identifies batches with high thickness uniformity deviations, potentially resulting in 10% fabric cutting waste, exceeding a threshold of 8%, triggering optimization. In one embodiment, cluster analysis technology uses a K-means algorithm to classify the consistent data set into two categories: high waste and low waste. Specifically, the high waste category shows thickness deviations concentrated between 2% and 3%, indicating unstable knitting machine tension control. The optimization solution recommends adjusting the knitting machine tension to 15N and reducing the yarn feed speed by 10%, generating optimized adjustment parameters. After these parameters are updated to the equipment, a new equipment operating data set is generated, recording, for example, that the adjusted tension remains stable at 15±0.5N. It is understood that the operating status feature extraction focuses on tension fluctuations, yarn feed speed consistency, and equipment vibration frequency. Statistical analysis technology assesses production stability by calculating the standard deviation of tension fluctuations, 0.5N. If the stability standard requires a standard deviation below 0.3N, the current status does not meet the standard. In one embodiment, based on tension fluctuation characteristics, the optimization solution was adjusted to increase the tension sensor feedback frequency to 10 times per second, generating new adjustment parameters. Ultimately, through further optimization, the resulting production optimization dataset showed that tension fluctuations were reduced to 0.2N and thickness uniformity deviation was reduced to 1.5%, significantly improving production stability. For example, this optimization solution uses a data-driven approach to precisely identify production issues, reduce resource waste, and improve fabric quality consistency. Preferably, the system records the effectiveness data of each optimization step to provide a reference for subsequent production and enhance the ability to continuously improve the process.

[0073] Example 2: The present invention provides a high elastic wear-resistant knitted denim fabric, which is composed of the following components

[0074] composition:

[0075] Element content Polyurethane elastic fiber 20％ polyester fiber 20％ cotton fiber 40％ Modal 5％ Nanosilica 5％ Dodecyldimethylbenzyl ammonium chloride margin Benzophenone-4 5％

[0076] The process for preparing the denim fabric comprises:

[0077] (1) The yarn thickness, twist and tension data are collected through sensors, and the yarn performance is judged to be within the target range based on the preset threshold value to obtain the real-time yarn status parameters;

[0078] (2) Based on the real-time yarn state parameters, a long short-term memory network model is used to predict the yarn performance change trend. The input layer contains three features: diameter, twist, and tension. The time step is set to 10, the number of hidden layer neurons is 32, the learning rate is 0.001, and the training data is the sampling data in the past hour. The yarn performance changes in the next 5 minutes are predicted;

[0079] (3) Obtain elasticity and functionality data from the fabric structure density obtained from the textile machine, determine whether the target elasticity requirements are met, and obtain fabric performance parameters;

[0080] (4) Based on the fabric performance parameters, the quality consistency of the finished product is analyzed through the online quality inspection system, and the resource waste rate is predicted in combination with historical data to determine the production process optimization plan, thereby obtaining a highly elastic and wear-resistant knitted denim fabric.

[0081] The specific operation steps of step (1) are as follows: (11) collecting yarn thickness, twist, and tension data through sensors to generate a raw data set;

[0082] (12) Using signal processing technology to denoise the original data set to obtain a smoothed data set;

[0083] (13) If the yarn thickness, twist, and tension values ​​in the smoothed data set are compared with the preset thresholds, a logical judgment is performed to determine whether each parameter is within the target range, and the performance compliance state is obtained;

[0084] (14) According to the performance compliance status, the decision tree algorithm is used to classify the yarn performance and obtain the classification results;

[0085] (15) By comparing the classification results with historical data, the fluctuation trend of the performance parameters is calculated and the fluctuation characteristics are obtained;

[0086] (16) If the fluctuation characteristics exceed the preset fluctuation range, an anomaly detection algorithm is used to identify potential defects and obtain a defect mark;

[0087] (17) Based on the defect marks, real-time status parameters are generated and the monitoring results of yarn performance are output.

[0088] The specific operation steps of step (3) are:

[0089] (31) obtaining fabric structure density data through textile machine sensors and storing the data as an initial data set;

[0090] (32) Data preprocessing techniques are used to denoise and standardize the initial dataset to obtain a structured density dataset;

[0091] (33) Extracting elastic parameters and functional parameters from the structured density data set to generate a fabric characteristic parameter set;

[0092] (34) If the difference between the elastic parameter in the fabric characteristic parameter set and the preset target elastic requirement is less than a preset threshold, it is determined that the requirement is met and an elastic compliance mark is generated;

[0093] (35) The fabric characteristic parameter set is classified by random forest algorithm to determine the fabric performance index category;

[0094] (36) Generate comprehensive fabric performance evaluation results based on elastic compliance marks and performance index categories;

[0095] (37) If the comprehensive evaluation results of fabric performance show that the performance index category is consistent with the target elasticity requirement, the final fabric performance index is output.

[0096] The specific operation steps of step (4) are as follows: (41) collecting performance parameters of high-elasticity and wear-resistant knitted denim fabric through an online quality detection system, and using data cleaning technology to denoise and standardize the collected performance parameters to obtain a consistent data set;

[0097] (42) Extracting quality characteristic parameters from the consistency data set, combining them with pre-stored historical data, and applying the support vector machine algorithm to analyze the quality characteristic parameters and historical data to generate a resource waste rate indicator;

[0098] (43) If the resource waste rate index is greater than the preset threshold, the consistent data set is classified through cluster analysis technology to determine the optimization plan of the production process and obtain the optimized adjustment parameters;

[0099] (44) updating the equipment configuration parameters of the production process according to the optimized adjustment parameters, and generating an updated equipment operation data set;

[0100] (45) Extracting operation status characteristics from the updated equipment operation data set, using statistical analysis technology to evaluate the operation status characteristics, judging the stability of the production process, and obtaining stability evaluation results;

[0101] (46) If the stability assessment results show that the production process does not meet the preset stability standards, the optimization plan is adjusted according to the operating status characteristics to generate new adjustment parameters;

[0102] (47) The production process is optimized again through new adjustment parameters to generate the final production optimization data set.

[0103] Sensors collecting data on yarn thickness, twist, and tension are fundamental to yarn performance monitoring. These sensors are typically deployed on spinning equipment to record yarn parameters in real time. For example, laser diameter gauges can measure yarn thickness with an accuracy of 0.01 mm; photoelectric sensors detect twist and can record 1000 twists per minute; and tension sensors measure yarn tension in a range of 0.1 to 10 Newtons. Raw data sets contain timestamps and parameter values, such as the yarn thickness of 0.2 mm, twist of 800 twists / meter, and tension of 2 Newtons at a given moment. These data may fluctuate abnormally due to equipment vibration or ambient noise. In one possible implementation, signal processing techniques are used for denoising. Preferably, wavelet transforms are used to decompose the signal, remove high-frequency noise, and preserve the trend characteristics of the yarn parameters. For example, a segment of yarn thickness data contains random noise. After wavelet denoising, the data curve becomes smoother, with the standard deviation reduced from 0.05 mm to 0.02 mm. Smoothed data sets more accurately reflect the true state of the yarn and facilitate subsequent analysis. Specifically, smoothed data is compared against preset thresholds to determine whether parameters meet the target. For example, the thresholds for yarn thickness are 0.18 to 0.22 mm, twist 750 to 850 twists / m, and tension 1.5 to 2.5 Newtons. If a data point shows thickness of 0.19 mm, twist 820 twists / m, and tension 2.2 Newtons, a logical analysis confirms that all parameters are within the target range and the data point is marked as qualified. It should be noted that the thresholds are set based on historical experience and product requirements to ensure quality consistency. In one embodiment, a decision tree algorithm is used to classify yarn properties. The decision tree divides nodes based on thickness, twist, and tension values ​​to generate classification rules. For example, a thickness less than 0.18 mm may indicate insufficient strength and be marked as unqualified; excessive twist may affect softness and be marked as suboptimal. After classification of a batch of yarn data using the decision tree, 80% was marked as high-quality, 15% as suboptimal, and 5% as unqualified. This classification method is intuitive and efficient, making it easy to trace the root cause of any problem. For example, comparing the classification results with historical data can identify fluctuation trends. Historical data shows that the thickness fluctuation range of high-quality yarn is ±0.02 mm. If the current batch fluctuates by ±0.03 mm, it indicates unstable performance. Fluctuation characteristics are obtained through statistical analysis, such as standard deviation or range, which reflect the reliability of the production process. It is understandable that the anomaly detection algorithm is used to identify potential defects. Preferably, the isolation forest algorithm is used to detect abnormal points based on fluctuation characteristics. For example, at a certain moment, the tension suddenly increases to 3 Newtons, which exceeds the normal fluctuation range. The algorithm marks it as a potential defect, which may be caused by equipment failure. Defect marking improves the efficiency of problem location and reduces the defective rate. In one embodiment, real-time status parameters are generated based on defect marking, including thickness, twist, tension and defect type, and output to the monitoring system. For example, a batch of yarn shows abnormal tension, and the monitoring results prompt to check the tension wheel. Real-time monitoring ensures rapid response and improves production efficiency and product quality.

[0104] A technical solution that uses a long short-term memory (LSTM) model to predict yarn performance trends based on real-time collected yarn state parameters has significant application value in the yarn production field. LSTM is a neural network suitable for processing time series data, capable of capturing the dynamic characteristics of yarn parameters over time. For example, LSTM uses memory cells and a gating mechanism to preserve long-term dependencies in historical data while filtering out irrelevant noise, ensuring prediction accuracy. The input layer contains three features: diameter, twist, and tension. The time step is 10, meaning the model uses data from the previous 10 time points to predict future trends. The number of hidden layer neurons is 32, and the learning rate is 0.001, ensuring a balance between model complexity and training stability. Training data is sampled from the past hour, and performance changes predicted for the next 5 minutes provide real-time basis for production adjustments. Specifically, diameter data can be collected using a high-precision optical caliper with a sampling frequency of 500 Hz and a measurement range of 0.05-1.5 mm. In one embodiment, the sensor collects 500 diameter data points per second. The LSTM model organizes this data into a time series and analyzes the pattern of diameter changes over time. For example, the model may discover periodic fluctuations in diameter over certain time periods, indicating mechanical vibration in the spinning machine and providing a reference for equipment maintenance. It should be noted that preprocessing of the diameter data requires removing outliers, such as using a median filter to eliminate sudden changes, to ensure stable feature data input to the LSTM. In one possible implementation, twist data is collected using a high-resolution encoder that measures the number of twists per 10 cm of yarn, ranging from 20 to 80 twists per meter. The LSTM analyzes the twist sequence to predict twist trends over the next five minutes. For example, the model might detect that twist is gradually deviating from the target value, prompting the operator to adjust the drafting parameters of the spinning machine. Preferably, the twist data input feature can incorporate ambient humidity, as humidity can affect the physical properties of yarn fibers and, thus, indirectly influence twist stability. For example, tension data can be collected using a high-sensitivity strain sensor with a range of 0-400 cN. The LSTM model uses the tension sequence to predict whether the tension will exceed a set range, such as 70-90 cN, within the next five minutes. In one embodiment, if the model predicts that the tension continues to rise, it may indicate an imminent risk of yarn breakage, and the system will issue an early warning and recommend reducing the spinning speed. It should be noted that the fluctuation of tension data may be related to the wear of the traction wheel of the spinning machine. The prediction results of LSTM can provide data support for equipment diagnosis. It is understandable that the choice of a time step of 10 balances the model's reliance on historical data and computational efficiency. In one embodiment, the time step can be dynamically adjusted according to the production speed, such as shortening it to 8 steps during high-speed spinning to capture finer-grained changes. The training data uses sampling data from the past hour, covering a sufficient production cycle to ensure that the model learns the regular characteristics of yarn performance.Setting a learning rate of 0.001 prevents the model from falling into a local optimum during training, and is suitable for scenarios where data distribution is relatively stable in yarn production. In one possible implementation, the output of the LSTM model can be linked to the production control system. For example, if the prediction results show that the diameter may be too small in the next 5 minutes, the system can automatically adjust the amount of fiber fed to optimize the yarn quality. Preferably, the model's prediction results will also be stored in a database for subsequent quality analysis, such as tracing the causes of performance defects in a batch of yarn. For example, the prediction capability of LSTM can also be extended to multi-feature joint analysis, such as comprehensively analyzing changes in diameter, twist, and tension to determine whether the overall performance of the yarn meets the requirements of high-end textiles. Through the above solution, the LSTM model provides an intelligent prediction tool for yarn production, capturing performance change trends from multiple dimensions such as diameter, twist, and tension, and providing data support for process optimization and quality control.

[0105] Acquiring fabric structural density data through textile machine sensors is fundamental to building a fabric performance evaluation system. Sensors are typically installed in key locations on textile machines, such as needles or yarn guides, to collect real-time warp and weft density information. For example, the sensors can use laser ranging or infrared scanning technology to collect data 1,000 times per second, recording the number of warp and weft yarns per square centimeter to form an initial dataset. For example, the initial dataset for a fabric sample shows a warp density of 120 yarns / cm and a weft density of 100 yarns / cm. In one possible implementation, data preprocessing techniques perform denoising and normalization on the initial dataset to eliminate noise interference and data bias. Denoising can employ a mean filter to remove outliers caused by sensor jitter. For example, density values ​​fluctuating by more than ±5% are replaced with the nearest mean. Normalization normalizes the data to the range 0-1 to ensure comparability across batches. This processing results in a structured density dataset, for example, normalizing the warp density to 0.6 and the weft density to 0.5. Specifically, extracting elastic parameters and functional parameters from the structured density data set is the key to generating a fabric characteristic parameter set. Elastic parameters include tensile modulus and recovery rate, which can be derived through correlation analysis between density and yarn material. For example, high-density fabrics usually have lower tensile modulus. Functional parameters such as air permeability and wear resistance are related to the uniformity of density distribution. In one embodiment, a fabric has a tensile modulus of 50N / cm, a recovery rate of 85%, and an air permeability of 200cm. 3 / s / cm 2, generating a characteristic parameter set. It should be noted that the generation of the elastic compliance flag relies on comparing the elastic parameters with the target elasticity requirements. Assuming the target tensile modulus is 48-52 N / cm and the threshold is ±2 N / cm, if the actual value is 50 N / cm, the difference meets the requirement, and a compliance flag is generated. If it does not meet the requirement, the flag is marked as non-compliant, which helps to quickly screen high-quality fabrics. In one embodiment, a random forest algorithm is used to classify the fabric characteristic parameter set and determine the performance indicator category. Random forest integrates multiple decision trees, analyzing the nonlinear relationships between characteristic parameters and classifying fabrics into three categories: high, medium, and low performance. For example, given tensile modulus, recovery rate, and breathability, the algorithm predicts a fabric as being in the high-performance category with an accuracy of 90%, improving classification reliability. Preferably, the comprehensive fabric performance evaluation results are generated by combining the elastic compliance flag and performance indicator category. If the flag is compliant and the category is high-performance, it is considered consistent with the target elasticity requirement, and the final performance indicator is output. For example, if a fabric has a tensile modulus of 50 N / cm and is classified as high-performance, the output indicator is "Excellent Elastic Fabric," which is used to guide production optimization. The advantage of this approach lies in its data-driven nature throughout the entire process. Sensor acquisition ensures real-time data, preprocessing improves data quality, and the random forest algorithm enhances classification accuracy. The final evaluation results provide clear guidance for production. For example, the screening of high-performance fabrics can reduce the defective rate by 10%, improving market competitiveness.

[0106] In the production of highly elastic and wear-resistant knitted denim fabric, an online quality inspection system uses sensors to collect performance parameters such as tensile strength, number of abrasions, and fabric thickness in real time. For example, the system uses a high-precision tension sensor to measure the stress of the fabric when stretched to 10%, recording it as 500N. Simultaneously, an abrasion tester records the weight loss of the fabric after 1000 abrasions, which is 0.5%. These parameters may contain noise due to equipment vibration or environmental interference. In one possible implementation, data cleaning technology uses median filtering to remove outliers, for example, removing records exceeding three standard deviations in tensile strength data, and normalizing thickness data to a mean of 0 and a standard deviation of 1 using Z-score standardization to generate a consistent data set. It should be noted that the cleaned data maintains the relative relationships between parameters, facilitating subsequent analysis. When extracting quality characteristic parameters from the consistent data set, tensile strength, abrasion loss rate, and thickness uniformity are preferably selected as core features. For example, a tensile strength of 500N corresponds to high elasticity requirements, a abrasion loss rate of 0.5% reflects wear resistance, and a thickness uniformity deviation of 2% indicates production consistency. Combining historical data, such as the characteristic parameters of fabrics from the past 100 batches, a support vector machine algorithm is applied to analyze the match between current and historical data to generate a resource waste rate indicator. For example, the algorithm identifies batches with high thickness uniformity deviations, potentially resulting in 10% fabric cutting waste, exceeding a threshold of 8%, triggering optimization. In one embodiment, cluster analysis technology uses a K-means algorithm to classify the consistent data set into two categories: high waste and low waste. Specifically, the high waste category shows thickness deviations concentrated between 2% and 3%, indicating unstable knitting machine tension control. The optimization solution recommends adjusting the knitting machine tension to 15N and reducing the yarn feed speed by 10%, generating optimized adjustment parameters. After these parameters are updated to the equipment, a new equipment operating data set is generated, recording, for example, that the adjusted tension remains stable at 15±0.5N. It is understood that the operating status feature extraction focuses on tension fluctuations, yarn feed speed consistency, and equipment vibration frequency. Statistical analysis technology assesses production stability by calculating the standard deviation of tension fluctuations, 0.5N. If the stability standard requires a standard deviation below 0.3N, the current status does not meet the standard. In one embodiment, based on tension fluctuation characteristics, the optimization solution was adjusted to increase the tension sensor feedback frequency to 10 times per second, generating new adjustment parameters. Ultimately, through further optimization, the resulting production optimization dataset showed that tension fluctuations were reduced to 0.2N and thickness uniformity deviation was reduced to 1.5%, significantly improving production stability. For example, this optimization solution uses a data-driven approach to precisely identify production issues, reduce resource waste, and improve fabric quality consistency. Preferably, the system records the effectiveness data of each optimization step to provide a reference for subsequent production and enhance the ability to continuously improve the process.

[0107] Example 3: The present invention provides a highly elastic and wear-resistant knitted denim fabric, comprising the following components:

[0108] Element content Polyurethane elastic fiber 15％ polyester fiber 15％ cotton fiber 45％ Modal 6％ Nanosilica 3％ Dodecyldimethylbenzyl ammonium chloride margin Benzophenone-4 3％

[0109] The process for preparing the denim fabric comprises:

[0110] (1) The yarn thickness, twist and tension data are collected through sensors, and the yarn performance is judged to be within the target range based on the preset threshold value to obtain the real-time yarn status parameters;

[0111] (2) Based on the real-time yarn state parameters, a long short-term memory network model is used to predict the yarn performance change trend. The input layer contains three features: diameter, twist, and tension. The time step is set to 10, the number of hidden layer neurons is 32, the learning rate is 0.001, and the training data is the sampling data in the past hour. The yarn performance changes in the next 5 minutes are predicted;

[0112] (3) Obtain elasticity and functionality data from the fabric structure density obtained from the textile machine, determine whether the target elasticity requirements are met, and obtain fabric performance parameters;

[0113] (4) Based on the fabric performance parameters, the quality consistency of the finished product is analyzed through the online quality inspection system, and the resource waste rate is predicted in combination with historical data to determine the production process optimization plan, thereby obtaining a highly elastic and wear-resistant knitted denim fabric.

[0114] The specific operation steps of step (1) are as follows: (11) collecting yarn thickness, twist, and tension data through sensors to generate a raw data set;

[0115] (12) Using signal processing technology to denoise the original data set to obtain a smoothed data set;

[0116] (13) If the yarn thickness, twist, and tension values ​​in the smoothed data set are compared with the preset thresholds, a logical judgment is performed to determine whether each parameter is within the target range, and the performance compliance state is obtained;

[0117] (14) According to the performance compliance status, the decision tree algorithm is used to classify the yarn performance and obtain the classification results;

[0118] (15) By comparing the classification results with historical data, the fluctuation trend of the performance parameters is calculated and the fluctuation characteristics are obtained;

[0119] (16) If the fluctuation characteristics exceed the preset fluctuation range, an anomaly detection algorithm is used to identify potential defects and obtain a defect mark;

[0120] (17) Based on the defect marks, real-time status parameters are generated and the monitoring results of yarn performance are output.

[0121] The specific operation steps of step (3) are:

[0122] (31) obtaining fabric structure density data through textile machine sensors and storing the data as an initial data set;

[0123] (32) Data preprocessing techniques are used to denoise and standardize the initial dataset to obtain a structured density dataset;

[0124] (33) Extracting elastic parameters and functional parameters from the structured density data set to generate a fabric characteristic parameter set;

[0125] (34) If the difference between the elastic parameter in the fabric characteristic parameter set and the preset target elastic requirement is less than a preset threshold, it is determined that the requirement is met and an elastic compliance mark is generated;

[0126] (35) The fabric characteristic parameter set is classified by random forest algorithm to determine the fabric performance index category;

[0127] (36) Generate comprehensive fabric performance evaluation results based on elastic compliance marks and performance index categories;

[0128] (37) If the comprehensive evaluation results of fabric performance show that the performance index category is consistent with the target elasticity requirement, the final fabric performance index is output.

[0129] The specific operation steps of step (4) are as follows: (41) collecting performance parameters of high-elasticity and wear-resistant knitted denim fabric through an online quality detection system, and using data cleaning technology to denoise and standardize the collected performance parameters to obtain a consistent data set;

[0130] (42) Extracting quality characteristic parameters from the consistency data set, combining them with pre-stored historical data, and applying the support vector machine algorithm to analyze the quality characteristic parameters and historical data to generate a resource waste rate indicator;

[0131] (43) If the resource waste rate index is greater than the preset threshold, the consistent data set is classified through cluster analysis technology to determine the optimization plan of the production process and obtain the optimized adjustment parameters;

[0132] (44) updating the equipment configuration parameters of the production process according to the optimized adjustment parameters, and generating an updated equipment operation data set;

[0133] (45) Extracting operation status characteristics from the updated equipment operation data set, using statistical analysis technology to evaluate the operation status characteristics, judging the stability of the production process, and obtaining stability evaluation results;

[0134] (46) If the stability assessment results show that the production process does not meet the preset stability standards, the optimization plan is adjusted according to the operating status characteristics to generate new adjustment parameters;

[0135] (47) The production process is optimized again through new adjustment parameters to generate the final production optimization data set.

[0136] Sensors collecting data on yarn thickness, twist, and tension are fundamental to yarn performance monitoring. These sensors are typically deployed on spinning equipment to record yarn parameters in real time. For example, laser diameter gauges can measure yarn thickness with an accuracy of 0.01 mm; photoelectric sensors detect twist and can record 1000 twists per minute; and tension sensors measure yarn tension in a range of 0.1 to 10 Newtons. Raw data sets contain timestamps and parameter values, such as the yarn thickness of 0.2 mm, twist of 800 twists / meter, and tension of 2 Newtons at a given moment. These data may fluctuate abnormally due to equipment vibration or ambient noise. In one possible implementation, signal processing techniques are used for denoising. Preferably, wavelet transforms are used to decompose the signal, remove high-frequency noise, and preserve the trend characteristics of the yarn parameters. For example, a segment of yarn thickness data contains random noise. After wavelet denoising, the data curve becomes smoother, with the standard deviation reduced from 0.05 mm to 0.02 mm. Smoothed data sets more accurately reflect the true state of the yarn and facilitate subsequent analysis. Specifically, smoothed data is compared against preset thresholds to determine whether parameters meet the target. For example, the thresholds for yarn thickness are 0.18 to 0.22 mm, twist 750 to 850 twists / m, and tension 1.5 to 2.5 Newtons. If a data point shows thickness of 0.19 mm, twist 820 twists / m, and tension 2.2 Newtons, a logical analysis confirms that all parameters are within the target range and the data point is marked as qualified. It should be noted that the thresholds are set based on historical experience and product requirements to ensure quality consistency. In one embodiment, a decision tree algorithm is used to classify yarn properties. The decision tree divides nodes based on thickness, twist, and tension values ​​to generate classification rules. For example, a thickness less than 0.18 mm may indicate insufficient strength and be marked as unqualified; excessive twist may affect softness and be marked as suboptimal. After classification of a batch of yarn data using the decision tree, 80% was marked as high-quality, 15% as suboptimal, and 5% as unqualified. This classification method is intuitive and efficient, making it easy to trace the root cause of any problem. For example, comparing the classification results with historical data can identify fluctuation trends. Historical data shows that the thickness fluctuation range of high-quality yarn is ±0.02 mm. If the current batch fluctuates by ±0.03 mm, it indicates unstable performance. Fluctuation characteristics are obtained through statistical analysis, such as standard deviation or range, which reflect the reliability of the production process. It is understandable that the anomaly detection algorithm is used to identify potential defects. Preferably, the isolation forest algorithm is used to detect abnormal points based on fluctuation characteristics. For example, at a certain moment, the tension suddenly increases to 3 Newtons, which exceeds the normal fluctuation range. The algorithm marks it as a potential defect, which may be caused by equipment failure. Defect marking improves the efficiency of problem location and reduces the defective rate. In one embodiment, real-time status parameters are generated based on defect marking, including thickness, twist, tension and defect type, and output to the monitoring system. For example, a batch of yarn shows abnormal tension, and the monitoring results prompt to check the tension wheel. Real-time monitoring ensures rapid response and improves production efficiency and product quality.

[0137] A technical solution that uses a long short-term memory (LSTM) model to predict yarn performance trends based on real-time collected yarn state parameters has significant application value in the yarn production field. LSTM is a neural network suitable for processing time series data, capable of capturing the dynamic characteristics of yarn parameters over time. For example, LSTM uses memory cells and a gating mechanism to preserve long-term dependencies in historical data while filtering out irrelevant noise, ensuring prediction accuracy. The input layer contains three features: diameter, twist, and tension. The time step is 10, meaning the model uses data from the previous 10 time points to predict future trends. The number of hidden layer neurons is 32, and the learning rate is 0.001, ensuring a balance between model complexity and training stability. Training data is sampled from the past hour, and performance changes predicted for the next 5 minutes provide real-time basis for production adjustments. Specifically, diameter data can be collected using a high-precision optical caliper with a sampling frequency of 500 Hz and a measurement range of 0.05-1.5 mm. In one embodiment, the sensor collects 500 diameter data points per second. The LSTM model organizes this data into a time series and analyzes the pattern of diameter changes over time. For example, the model may discover periodic fluctuations in diameter over certain time periods, indicating mechanical vibration in the spinning machine and providing a reference for equipment maintenance. It should be noted that preprocessing of the diameter data requires removing outliers, such as using a median filter to eliminate sudden changes, to ensure stable feature data input to the LSTM. In one possible implementation, twist data is collected using a high-resolution encoder that measures the number of twists per 10 cm of yarn, ranging from 20 to 80 twists per meter. The LSTM analyzes the twist sequence to predict twist trends over the next five minutes. For example, the model might detect that twist is gradually deviating from the target value, prompting the operator to adjust the drafting parameters of the spinning machine. Preferably, the twist data input feature can incorporate ambient humidity, as humidity can affect the physical properties of yarn fibers and, thus, indirectly influence twist stability. For example, tension data can be collected using a high-sensitivity strain sensor with a range of 0-400 cN. The LSTM model uses the tension sequence to predict whether the tension will exceed a set range, such as 70-90 cN, within the next five minutes. In one embodiment, if the model predicts that the tension continues to rise, it may indicate an imminent risk of yarn breakage, and the system will issue an early warning and recommend reducing the spinning speed. It should be noted that the fluctuation of tension data may be related to the wear of the traction wheel of the spinning machine. The prediction results of LSTM can provide data support for equipment diagnosis. It is understandable that the choice of a time step of 10 balances the model's reliance on historical data and computational efficiency. In one embodiment, the time step can be dynamically adjusted according to the production speed, such as shortening it to 8 steps during high-speed spinning to capture finer-grained changes. The training data uses sampling data from the past hour, covering a sufficient production cycle to ensure that the model learns the regular characteristics of yarn performance.Setting a learning rate of 0.001 prevents the model from falling into a local optimum during training, and is suitable for scenarios where data distribution is relatively stable in yarn production. In one possible implementation, the output of the LSTM model can be linked to the production control system. For example, if the prediction results show that the diameter may be too small in the next 5 minutes, the system can automatically adjust the amount of fiber fed to optimize the yarn quality. Preferably, the model's prediction results will also be stored in a database for subsequent quality analysis, such as tracing the causes of performance defects in a batch of yarn. For example, the prediction capability of LSTM can also be extended to multi-feature joint analysis, such as comprehensively analyzing changes in diameter, twist, and tension to determine whether the overall performance of the yarn meets the requirements of high-end textiles. Through the above solution, the LSTM model provides an intelligent prediction tool for yarn production, capturing performance change trends from multiple dimensions such as diameter, twist, and tension, and providing data support for process optimization and quality control.

[0138] Acquiring fabric structural density data through textile machine sensors is fundamental to building a fabric performance evaluation system. Sensors are typically installed in key locations on textile machines, such as needles or yarn guides, to collect real-time warp and weft density information. For example, the sensors can use laser ranging or infrared scanning technology to collect data 1,000 times per second, recording the number of warp and weft yarns per square centimeter to form an initial dataset. For example, the initial dataset for a fabric sample shows a warp density of 120 yarns / cm and a weft density of 100 yarns / cm. In one possible implementation, data preprocessing techniques perform denoising and normalization on the initial dataset to eliminate noise interference and data bias. Denoising can employ a mean filter to remove outliers caused by sensor jitter. For example, density values ​​fluctuating by more than ±5% are replaced with the nearest mean. Normalization normalizes the data to the range 0-1 to ensure comparability across batches. This processing results in a structured density dataset, for example, normalizing the warp density to 0.6 and the weft density to 0.5. Specifically, extracting elastic parameters and functional parameters from the structured density data set is the key to generating a fabric characteristic parameter set. Elastic parameters include tensile modulus and recovery rate, which can be derived through correlation analysis between density and yarn material. For example, high-density fabrics usually have lower tensile modulus. Functional parameters such as air permeability and wear resistance are related to the uniformity of density distribution. In one embodiment, a fabric has a tensile modulus of 50N / cm, a recovery rate of 85%, and an air permeability of 200cm. 3 / s / cm 2, generating a characteristic parameter set. It should be noted that the generation of the elastic compliance flag relies on comparing the elastic parameters with the target elasticity requirements. Assuming the target tensile modulus is 48-52 N / cm and the threshold is ±2 N / cm, if the actual value is 50 N / cm, the difference meets the requirement, and a compliance flag is generated. If it does not meet the requirement, the flag is marked as non-compliant, which helps to quickly screen high-quality fabrics. In one embodiment, a random forest algorithm is used to classify the fabric characteristic parameter set and determine the performance indicator category. Random forest integrates multiple decision trees, analyzing the nonlinear relationships between characteristic parameters and classifying fabrics into three categories: high, medium, and low performance. For example, given tensile modulus, recovery rate, and breathability, the algorithm predicts a fabric as being in the high-performance category with an accuracy of 90%, improving classification reliability. Preferably, the comprehensive fabric performance evaluation results are generated by combining the elastic compliance flag and performance indicator category. If the flag is compliant and the category is high-performance, it is considered consistent with the target elasticity requirement, and the final performance indicator is output. For example, if a fabric has a tensile modulus of 50 N / cm and is classified as high-performance, the output indicator is "Excellent Elastic Fabric," which is used to guide production optimization. The advantage of this approach lies in its data-driven nature throughout the entire process. Sensor acquisition ensures real-time data, preprocessing improves data quality, and the random forest algorithm enhances classification accuracy. The final evaluation results provide clear guidance for production. For example, the screening of high-performance fabrics can reduce the defective rate by 10%, improving market competitiveness.

[0139] In the production of highly elastic and wear-resistant knitted denim fabric, an online quality inspection system uses sensors to collect performance parameters such as tensile strength, number of abrasions, and fabric thickness in real time. For example, the system uses a high-precision tension sensor to measure the stress of the fabric when stretched to 10%, recording it as 500N. Simultaneously, an abrasion tester records the weight loss of the fabric after 1000 abrasions, which is 0.5%. These parameters may contain noise due to equipment vibration or environmental interference. In one possible implementation, data cleaning technology uses median filtering to remove outliers, for example, removing records exceeding three standard deviations in tensile strength data, and normalizing thickness data to a mean of 0 and a standard deviation of 1 using Z-score standardization to generate a consistent data set. It should be noted that the cleaned data maintains the relative relationships between parameters, facilitating subsequent analysis. When extracting quality characteristic parameters from the consistent data set, tensile strength, abrasion loss rate, and thickness uniformity are preferably selected as core features. For example, a tensile strength of 500N corresponds to high elasticity requirements, a abrasion loss rate of 0.5% reflects wear resistance, and a thickness uniformity deviation of 2% indicates production consistency. Combining historical data, such as the characteristic parameters of fabrics from the past 100 batches, a support vector machine algorithm is applied to analyze the match between current and historical data to generate a resource waste rate indicator. For example, the algorithm identifies batches with high thickness uniformity deviations, potentially resulting in 10% fabric cutting waste, exceeding a threshold of 8%, triggering optimization. In one embodiment, cluster analysis technology uses a K-means algorithm to classify the consistent data set into two categories: high waste and low waste. Specifically, the high waste category shows thickness deviations concentrated between 2% and 3%, indicating unstable knitting machine tension control. The optimization solution recommends adjusting the knitting machine tension to 15N and reducing the yarn feed speed by 10%, generating optimized adjustment parameters. After these parameters are updated to the equipment, a new equipment operating data set is generated, recording, for example, that the adjusted tension remains stable at 15±0.5N. It is understood that the operating status feature extraction focuses on tension fluctuations, yarn feed speed consistency, and equipment vibration frequency. Statistical analysis technology assesses production stability by calculating the standard deviation of tension fluctuations, 0.5N. If the stability standard requires a standard deviation below 0.3N, the current status does not meet the standard. In one embodiment, based on tension fluctuation characteristics, the optimization solution was adjusted to increase the tension sensor feedback frequency to 10 times per second, generating new adjustment parameters. Ultimately, through further optimization, the resulting production optimization dataset showed that tension fluctuations were reduced to 0.2N and thickness uniformity deviation was reduced to 1.5%, significantly improving production stability. For example, this optimization solution uses a data-driven approach to precisely identify production issues, reduce resource waste, and improve fabric quality consistency. Preferably, the system records the effectiveness data of each optimization step to provide a reference for subsequent production and enhance the ability to continuously improve the process.

[0140] The following experiments illustrate the elastic stability performance of the denim fabric of the present invention, to illustrate the beneficial effects obtained by the present invention: high elasticity, wear resistance, and product consistency.

[0141] 1. Experimental Method: A comparative example with Examples 1 to 3 is set up, and the comparative example is as follows:

[0142] The composition of the high elasticity and wear-resistant knitted denim fabric of Comparative Example 1 is as follows:

[0143] Element content cotton 90％ spandex 5％ graphene fiber 3％ Phenyl salicylate 2％

[0144] The composition of the high elasticity and wear-resistant knitted denim fabric of comparative example 2 is as follows:

[0145]

[0146]

[0147] The preparation method is prepared by the method of the embodiment of the present invention.

[0148] 2. Common testing methods for elasticity and abrasion resistance of denim fabrics:

[0149] (1) Denim fabric elasticity test: test the horizontal pulling depth and vertical pulling depth.

[0150] (2) Abrasion resistance test of denim fabric: Tested in accordance with FZ / T 81006-2017 "Denim Clothing" and GB / T21196.2-2007 "Textiles - Determination of the Abrasion Resistance of Fabrics by the Martindale Method - Part 2: Determination of Specimen Damage". Evaluation standard: Mass per unit area ≤ 339 g / m 2 For fabrics, if it is a superior product, the wear-resistant revolution number must be ≥15000r, and for first-class or qualified products, the wear-resistant revolution number must be ≥10000r.

[0151] 3. Experimental results:

[0152]

[0153] It should be noted that the above experimental data is the average value of more than 30 parallel experiments. This shows that only under the specific composition conditions of the present invention can a fabric with excellent elasticity and good wear resistance be obtained. It can also be seen that the product consistency is good. When changing the composition of the fabric product, the applicant actually conducted countless such experiments when obtaining the technical solution. Some ingredients may not be able to be combined together, resulting in problems such as the inability to obtain fabric. Since the experimental content is repetitive and too redundant, only a brief description is provided here.

[0154] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A highly elastic and wear-resistant knitted denim fabric, characterized in that: The fabric comprises polyurethane elastic fiber, polyester fiber, cotton fiber and regenerated cellulose fiber, and the mass percentages of the various components in the fabric are respectively 10-20% of the polyurethane elastic fiber, 10-20% of the polyester fiber, 40-50% of the cotton fiber and 5-10% of the regenerated cellulose fiber.

2. The fabric according to claim 1, characterized in that The regenerated cellulose fiber is lyocell fiber or modal fiber.

3. The fabric according to claim 1, characterized in that Also includes nano-silica, antimicrobial agents, and UV protection agents.

4. The fabric according to claim 3, characterized in that The mass percentage of the nano silicon dioxide in the fabric is 1-5%, the mass percentage of the ultraviolet protection agent in the fabric is 1-5%, and the balance is the antibacterial agent.

5. The fabric according to claim 3, characterized in that: The antibacterial agent is silver ion or dodecyldimethylbenzyl ammonium chloride, and the UV protection agent is benzophenone-4.

6. A method for preparing the fabric according to any one of claims 1 to 5, characterized in that: The steps include: (1) The yarn thickness, twist and tension data are collected through sensors, and the yarn performance is judged to be within the target range based on the preset threshold value to obtain the real-time yarn status parameters; (2) Based on the real-time yarn state parameters, a long short-term memory network model is used to predict the yarn performance change trend. The input layer contains three features: diameter, twist, and tension. The time step is set to 10, the number of hidden layer neurons is 32, the learning rate is 0.001, and the training data is the sampling data in the past hour. The yarn performance changes in the next 5 minutes are predicted; (3) Obtain elasticity and functionality data from the fabric structure density obtained from the textile machine, determine whether the target elasticity requirements are met, and obtain fabric performance parameters; (4) Based on the fabric performance parameters, the quality consistency of the finished product is analyzed through the online quality inspection system, and the resource waste rate is predicted in combination with historical data to determine the production process optimization plan, thereby obtaining a highly elastic and wear-resistant knitted denim fabric.

7. The method according to claim 6, characterized in that The specific operation steps of step (1) are as follows: (11) collecting yarn thickness, twist, and tension data through sensors to generate a raw data set; (12) Using signal processing technology to denoise the original data set to obtain a smoothed data set; (13) If the yarn thickness, twist, and tension values ​​in the smoothed data set are compared with the preset thresholds, a logical judgment is performed to determine whether each parameter is within the target range, and the performance compliance state is obtained; (14) According to the performance compliance status, the decision tree algorithm is used to classify the yarn performance and obtain the classification results; (15) By comparing the classification results with historical data, the fluctuation trend of the performance parameters is calculated and the fluctuation characteristics are obtained; (16) If the fluctuation characteristics exceed the preset fluctuation range, an anomaly detection algorithm is used to identify potential defects and obtain a defect mark; (17) Based on the defect marks, real-time status parameters are generated and the monitoring results of yarn performance are output.

8. The method according to claim 6, characterized in that The specific operation steps of step (3) are: (31) obtaining fabric structure density data through textile machine sensors and storing the data as an initial data set; (32) Data preprocessing techniques are used to denoise and standardize the initial dataset to obtain a structured density dataset; (33) Extracting elastic parameters and functional parameters from the structured density data set to generate a fabric characteristic parameter set; (34) If the difference between the elastic parameter in the fabric characteristic parameter set and the preset target elastic requirement is less than a preset threshold, it is determined that the requirement is met and an elastic compliance mark is generated; (35) The fabric characteristic parameter set is classified by random forest algorithm to determine the fabric performance index category; (36) Generate comprehensive fabric performance evaluation results based on elastic compliance marks and performance index categories; (37) If the comprehensive evaluation results of fabric performance show that the performance index category is consistent with the target elasticity requirement, the final fabric performance index is output.

9. The method according to claim 6, characterized in that The specific operation steps of step (4) are as follows: (41) collecting performance parameters of high-elasticity and wear-resistant knitted denim fabric through an online quality detection system, and using data cleaning technology to denoise and standardize the collected performance parameters to obtain a consistent data set; (42) Extracting quality characteristic parameters from the consistency data set, combining them with pre-stored historical data, and applying the support vector machine algorithm to analyze the quality characteristic parameters and historical data to generate a resource waste rate indicator; (43) If the resource waste rate index is greater than the preset threshold, the consistent data set is classified through cluster analysis technology to determine the optimization plan of the production process and obtain the optimized adjustment parameters; (44) updating the equipment configuration parameters of the production process according to the optimized adjustment parameters, and generating an updated equipment operation data set; (45) Extracting operation status characteristics from the updated equipment operation data set, using statistical analysis technology to evaluate the operation status characteristics, judging the stability of the production process, and obtaining stability evaluation results; (46) If the stability assessment results show that the production process does not meet the preset stability standards, the optimization plan is adjusted according to the operating status characteristics to generate new adjustment parameters; (47) The production process is optimized again through new adjustment parameters to generate the final production optimization data set.

Images