Method for simultaneous detection of multiple performance indicators of cotton textile yarns

Abstract

The application relates to the technical field of cotton yarn detection, and discloses a cotton textile yarn multi-performance index synchronous detection method. The method acquires current surface image parameters of a yarn body appearance layer, current internal structure parameters of a fiber structure layer and current molecular motion parameters of a molecular chain layer corresponding to a current production batch of yarn to be detected; then layered performance analysis is respectively performed on the parameters to obtain a first current performance state evaluation value of the yarn body appearance layer, a second current performance state evaluation value of the fiber structure layer and a third current performance state evaluation value of the molecular chain layer; finally, the evaluation values are matched with preset performance state evaluation benchmark ranges for judgment, and corresponding detection calibration operations are triggered according to the results. The method can synchronously acquire multi-layer parameters, realize comprehensive performance evaluation, solve problems such as single detection dimension, low efficiency, poor result correlation and lag, improve detection reliability and consistency, and help production quality control and process optimization.

Description

Technical Field

[0001] This invention relates to the field of cotton yarn testing technology, specifically a method for simultaneous testing of multiple performance indicators of cotton textile yarns. Background Technology

[0002] In the textile industry, the performance of cotton yarns directly affects the quality, comfort, and durability of subsequent textiles. Therefore, testing the performance indicators of cotton yarns is an indispensable part of the production process. Currently, there are many methods for testing the performance indicators of cotton yarns, but they generally suffer from problems such as limited testing dimensions, low testing efficiency, and poor correlation of test results.

[0003] The inspection of yarn surface parameters often relies on manual visual inspection or a single image acquisition device. Manual visual inspection is not only easily affected by the subjective factors of the inspectors, such as fatigue and experience differences, resulting in unstable accuracy of the inspection results, but also slow in speed, making it difficult to meet the inspection needs of large-scale production lines. A single image acquisition device can only obtain partial appearance information of the yarn surface, such as yarn diameter and the number of surface defects, and cannot fully reflect the overall state of the yarn surface. For example, key parameters such as the uniformity of yarn surface hair distribution and color consistency are often ignored.

[0004] The commonly used method for detecting the internal structural parameters of the fiber structure layer is to take samples and observe them under a microscope. This method is destructive and will cause damage to the yarn. In addition, the sampling process is random and cannot represent the fiber structure state of the entire production batch of yarn. At the same time, microscopic observation can only obtain static fiber arrangement information and cannot capture the dynamic structural changes of fibers under stress or environmental changes. The dynamic stability of the fiber structure has an important impact on the tensile strength, elastic recovery and other properties of the yarn. The limitations of existing detection methods make it impossible to accurately assess the contribution of the fiber structure layer to the overall performance of the yarn.

[0005] In the detection of molecular motion parameters at the molecular chain layer, existing technologies mostly employ offline testing methods. This requires sending yarn samples to specialized laboratories and using complex instruments such as nuclear magnetic resonance spectrometers and infrared spectrometers for analysis. The testing cycle is long, often taking several days from sample collection and transportation to completion. Test results cannot be promptly fed back to the production site, meaning that if abnormal molecular chain motion occurs during production, production parameters cannot be adjusted in time, easily leading to a large number of defective products. Furthermore, existing testing methods typically separate the performance testing of the yarn appearance layer, fiber structure layer, and molecular chain layer. Each testing stage is independent, lacking effective integration and correlation analysis of the test data. This makes it impossible to judge the overall performance status of the yarn as a whole, and it is difficult to accurately identify the interrelationships between different performance levels. For example, does abnormal molecular chain motion lead to a loose fiber structure, thus affecting the yarn appearance and overall strength? This decentralized testing model greatly limits the comprehensive and accurate evaluation of cotton textile yarn performance and is also detrimental to real-time optimization and quality control of the production process. Summary of the Invention

[0006] The purpose of this invention is to provide a method for simultaneous detection of multiple performance indicators of cotton textile yarns, so as to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides a method for simultaneous detection of multiple performance indicators of cotton textile yarns, the method comprising:

[0008] Acquire the current surface image parameters of the yarn body appearance layer, the current internal structure parameters of the fiber structure layer, and the current molecular motion parameters of the molecular chain layer corresponding to the current production batch of the yarn to be tested;

[0009] Layered performance analysis is performed on the current surface image parameters of the yarn surface layer, the current internal structure parameters of the fiber structure layer, and the current molecular motion parameters of the molecular chain layer to obtain the first current performance status evaluation value of the yarn surface layer, the second current performance status evaluation value of the fiber structure layer, and the third current performance status evaluation value of the molecular chain layer.

[0010] The first current performance status evaluation value, the second current performance status evaluation value, and the third current performance status evaluation value are matched and judged with the preset performance status evaluation benchmark range, and the corresponding detection and calibration operation is triggered according to the matching judgment result.

[0011] Preferably, the current surface image parameters of the yarn surface layer specifically include the yarn surface hair density value, the yarn surface texture clarity value, and the yarn surface color difference distribution value;

[0012] The current internal structural parameters of the fiber structure layer specifically include fiber arrangement density, fiber torsion angle, and fiber cross-sectional uniformity.

[0013] The current molecular motion parameters of the molecular chain layer specifically include the molecular chain vibration frequency value, the molecular chain orientation value, and the molecular chain crystallinity value.

[0014] Preferably, the specific steps for obtaining the first current performance status evaluation value of the yarn surface layer are as follows:

[0015] The yarn surface hair density value, the yarn surface texture clarity value, and the yarn surface color difference distribution value are normalized and converted.

[0016] Based on the normalized transformation of the yarn surface hair density value, yarn surface texture clarity value, and yarn surface color difference distribution value, and combined with the preset yarn appearance layer weight allocation strategy, a comprehensive calculation is performed to generate the surface state evaluation value of the yarn appearance layer.

[0017] Preferably, the specific steps for obtaining the second current performance status evaluation value of the fiber structure layer are as follows:

[0018] The fiber packing density value, the fiber torsion angle value, and the fiber cross-sectional uniformity value are standardized and converted.

[0019] Based on the standardized fiber arrangement density, fiber torsion angle, and fiber cross-sectional uniformity values, a weighted fusion calculation is performed using a preset dynamic influence factor for the fiber structure layer to generate an evaluation value of the internal state of the fiber structure layer.

[0020] Preferably, the specific steps for obtaining the third current performance state evaluation value of the molecular chain layer are as follows:

[0021] The molecular chain vibration frequency value, the molecular chain orientation value, and the molecular chain crystallinity value are subjected to feature quantification processing.

[0022] Based on the molecular chain vibration frequency value, molecular chain orientation value, and molecular chain crystallinity value after feature quantization, multi-dimensional aggregation calculation is performed in combination with the preset molecular chain layer coupling coefficient to generate the molecular chain layer motion state evaluation value.

[0023] Preferably, the specific steps for performing the matching and judgment operation are as follows:

[0024] A benchmark detection strategy set is constructed based on historical detection datasets, and a compensation detection strategy set is constructed based on real-time detection feedback data.

[0025] Based on the surface state evaluation value of the yarn appearance layer, the internal state evaluation value of the fiber structure layer, and the motion state evaluation value of the molecular chain layer, target detection strategy combinations are selected from the benchmark detection strategy set and the compensation detection strategy set, respectively.

[0026] Preferably, the specific steps for triggering the corresponding detection and calibration operation based on the matching judgment are as follows:

[0027] If the surface condition evaluation value of the yarn surface layer exceeds the preset performance condition evaluation benchmark range of the yarn surface layer, the yarn surface layer calibration strategy will be activated.

[0028] If the internal state evaluation value of the fiber structure layer exceeds the preset performance state evaluation benchmark range of the fiber structure layer, the fiber structure layer calibration strategy is activated.

[0029] If the molecular chain layer motion state evaluation value exceeds the preset molecular chain layer performance state evaluation benchmark range, then the molecular chain layer calibration strategy is activated.

[0030] Preferably, the method further includes a calibration feedback step:

[0031] Record the execution result data of the calibration strategy for the yarn appearance layer, the calibration strategy for the fiber structure layer, and the calibration strategy for the molecular chain layer;

[0032] By comparing the execution result data with the expected result data of the target detection strategy combination, the calibration difference factors are extracted.

[0033] Preferably, the method further includes a policy update step:

[0034] The weight allocation rules of the benchmark detection strategy set are revised based on the calibration difference factors;

[0035] The strategy confidence parameter of the compensation detection strategy set is adjusted based on the calibration difference factors.

[0036] Preferably, the method further includes a data loop closure step:

[0037] Update the revised weight allocation rules and adjusted strategy confidence parameters to the historical detection dataset;

[0038] The next detection cycle will perform simultaneous detection of multiple performance metrics based on the updated historical detection dataset.

[0039] Compared with the prior art, the beneficial effects of the present invention are:

[0040] This method for simultaneous testing of multiple performance indicators in cotton yarns can acquire test parameters simultaneously from three different levels: the yarn appearance layer, the fiber structure layer, and the molecular chain layer, breaking the limitation of independent testing at each level in existing methods. By simultaneously acquiring the current surface image parameters of the yarn appearance layer, the current internal structure parameters of the fiber structure layer, and the current molecular motion parameters of the molecular chain layer, multi-dimensional coverage of yarn performance indicators can be achieved. This avoids the problem of incomplete performance evaluation caused by single-dimensional testing, allowing testers to understand the overall performance status of the yarn and clearly grasp the correlation between performance at different levels. For example, it allows for direct observation of whether changes in molecular chain motion parameters affect the fiber structure and are reflected in changes in the yarn appearance, providing a comprehensive data foundation for in-depth analysis of yarn performance.

[0041] In the layered performance analysis stage, targeted analysis of parameters at different levels fully uncovers the performance information contained within each level. For surface image parameters of the yarn appearance layer, professional image analysis algorithms can accurately extract detailed information such as defect types, hair length distribution, and color uniformity on the yarn surface. Compared to traditional manual visual inspection, this effectively avoids the interference of subjective factors and improves the accuracy of appearance performance evaluation. For internal structural parameters of the fiber structure layer, parameters such as fiber arrangement density and inter-fiber bonding force obtained using advanced non-destructive testing technology can be accurately determined after layered analysis, avoiding yarn loss caused by destructive testing. At the same time, it can also capture the dynamic changes in fiber structure, which is more in line with the performance of yarn in actual use. For molecular motion parameters of the molecular chain layer, real-time acquisition and analysis can promptly grasp changes in molecular motion state, avoiding the lag of offline testing and providing timely basis for parameter adjustments during the production process.

[0042] By matching the current performance status assessment values ​​at each level with preset performance status assessment benchmark ranges and triggering corresponding testing and calibration operations based on the matching results, the reliability and consistency of test results can be effectively improved. When the assessment value of a certain level exceeds the benchmark range, timely calibration operations can correct deviations in the testing process. For example, if the assessment value of the yarn appearance layer is abnormal, the parameters of the image acquisition equipment can be calibrated to ensure the accuracy of subsequent test data; if the assessment value of the fiber structure layer or molecular chain layer is abnormal, the operating status of the testing equipment or the testing environment conditions can be checked to eliminate the influence of external factors on the test results. This dynamic calibration mechanism enables the testing system to always maintain optimal testing conditions, reducing inaccurate test results caused by equipment errors or environmental changes, thereby providing more reliable support for the control of yarn production quality.

[0043] This method enables simultaneous detection and analysis of multiple performance indicators, significantly improving detection efficiency. It meets the rapid performance testing needs of large-scale production lines, avoiding the time-consuming nature of traditional decentralized testing, thus helping to shorten production cycles and reduce costs. Simultaneously, comprehensive and accurate performance test results provide direction for optimizing yarn production processes. Production personnel can adjust production parameters such as raw material ratios, spinning speed, temperature, and humidity based on performance evaluations at various levels, thereby improving overall yarn quality, reducing defective products, and enhancing the company's competitiveness in the market. Attached Figure Description

[0044] Figure 1 This is a schematic diagram illustrating the working principle of the simultaneous detection method for multiple performance indicators of cotton textile yarns described in this invention.

[0045] Figure 2 A flowchart for obtaining the first current performance status evaluation value of the yarn body appearance layer;

[0046] Figure 3 A flowchart for obtaining the current performance status evaluation value of the third molecular chain layer;

[0047] Figure 4 This is a flowchart for the testing, calibration, and feedback of cotton textile yarns. Detailed Implementation

[0048] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0049] Please see Figure 1 This invention provides a method for simultaneous detection of multiple performance indicators of cotton textile yarns, the method comprising:

[0050] The process involves acquiring the current surface image parameters of the yarn's outer layer, the current internal structural parameters of the fiber structure layer, and the current molecular motion parameters of the molecular chain layer for the current production batch of the yarn under test. Yarn surface images are acquired using image acquisition equipment such as a high-resolution industrial camera, and surface feature parameters are extracted using image processing algorithms. Internal fiber structure images are acquired using microscopic imaging techniques such as scanning electron microscopy, and internal feature parameters are extracted using structural analysis algorithms. Molecular motion data are acquired using spectral analysis equipment such as an infrared spectrometer or nuclear magnetic resonance spectrometer, and molecular motion feature parameters are extracted using signal processing algorithms. Layered performance analysis is then performed on the current surface image parameters of the yarn's outer layer, the current internal structural parameters of the fiber structure layer, and the current molecular motion parameters of the molecular chain layer. Normalization, standardization, and feature quantization methods are used to process the parameters of each layer. A comprehensive calculation is then performed based on a preset weighting strategy, dynamic influence factors, and coupling coefficients to generate a first current performance status evaluation value for the yarn's outer layer, a second current performance status evaluation value for the fiber structure layer, and a third current performance status evaluation value for the molecular chain layer. The first, second, and third current performance status assessment values ​​are matched against a preset performance status assessment benchmark range. This benchmark range is based on historical test data and covers the status value range of qualified yarn products. If any assessment value exceeds the benchmark range, a corresponding test calibration operation is triggered, such as adjusting production equipment parameters or feeding back to the quality control module.

[0051] Example 1: See Figure 2 The current surface image parameters of the yarn body include yarn surface hairiness density, yarn surface texture sharpness, and yarn surface color difference distribution. Yarn surface hairiness density is calculated using digital image processing algorithms after acquiring yarn surface images with a high-resolution industrial camera. Specifically, it involves identifying and counting the number of protruding fibers per unit length of yarn surface, expressed as the number of fibers per millimeter. This parameter directly reflects the smoothness and potential pilling tendency of the yarn surface. Yarn surface texture sharpness is obtained through edge detection and texture analysis of the acquired yarn images. Texture contour features are extracted using the Sobel operator or Canny algorithm, and a quantitative index of contour sharpness is calculated, expressed as a percentage. This parameter characterizes the clarity and regularity of the yarn surface structure. Yarn surface color difference distribution is measured using a spectrophotometer or high-precision color sensor. By acquiring color data from different areas of the yarn surface, the standard color difference ΔE value is calculated based on the CIELAB color space model. This parameter objectively reflects the color uniformity and dyeing quality of the yarn surface.

[0052] The current internal structural parameters of the fiber structure layer include fiber packing density, fiber torsion angle, and fiber cross-sectional uniformity. Fiber packing density is measured using scanning electron microscopy (SEM) to obtain cross-sectional images of the fibers. Image analysis software is used to calculate the ratio of the area of ​​voids between fibers to the total area, which is then converted into a density coefficient. This parameter reflects the filling density and packing state of the fibers within the yarn. Fiber torsion angle is measured using microscopic imaging combined with digital image correlation methods. By tracing the axis of a single fiber, the change in torsion angle per unit length is calculated, using angles as the unit of measurement. This parameter characterizes the influence of the twisting process on the fiber structure. Fiber cross-sectional uniformity is assessed by analyzing the cross-sectional shape and dimensions of multiple fibers and calculating their coefficient of variation. Specifically, the ratio of the standard deviation to the mean is calculated after measuring the fiber cross-sectional diameter or area. This parameter reflects the uniformity of fiber thickness and structural consistency.

[0053] The current molecular motion parameters of the molecular chain layer include the molecular chain vibrational frequency, molecular chain orientation, and molecular chain crystallinity. The molecular chain vibrational frequency is determined by analyzing the frequency positions of characteristic absorption peaks in the infrared absorption spectrum of the sample using a Fourier transform infrared spectrometer. The values ​​are expressed in wavenumber units. This parameter reflects the chemical bond strength and thermal motion state of the molecular chains. The molecular chain orientation value is measured using a wide-angle X-ray diffractometer. By analyzing the azimuth intensity distribution in the diffraction pattern, the Hermann orientation factor is calculated as a quantitative indicator. This parameter characterizes the degree of orderliness of the macromolecular chains along the fiber axis. The molecular chain crystallinity value is calculated using differential scanning calorimetry by measuring the enthalpy of fusion of the sample and comparing it with the theoretical enthalpy of fusion of a fully crystalline sample. The result is expressed as a percentage of the crystalline region. This parameter reflects the material's crystallinity and fundamental mechanical properties.

[0054] Obtaining the first current performance status evaluation value of the yarn surface layer requires a systematic data processing and analysis process. The yarn surface hairiness density, texture clarity, and color difference distribution values ​​are normalized using a minimum-maximum normalization method to linearly transform the original data of each parameter to the [0,1] interval. During normalization, the minimum and maximum values ​​of each parameter are statistically analyzed based on historical datasets as a conversion benchmark, allowing for direct comparison and comprehensive calculation of parameters of different dimensions and magnitudes. Based on the normalized parameter values, a comprehensive evaluation value is generated using a pre-defined weight allocation strategy for the yarn surface layer. The weight allocation strategy is set based on the importance of each parameter to the yarn appearance quality, and the final weight coefficients are determined through expert evaluation and historical data regression analysis. Mathematical formulas are used for comprehensive evaluation calculations during implementation.

[0055]

[0056] in: This represents the surface condition assessment value of the yarn's outer layer. The weighting coefficient representing the yarn surface hairiness density value. This represents the yarn surface hairiness density value after normalization. The weighting coefficients representing the clarity of the yarn surface texture. This represents the yarn surface texture clarity value after normalization. The weighting coefficients representing the color difference distribution values ​​on the yarn surface. This represents the normalized color difference distribution value of the yarn surface. Weighting coefficient. , and The numerical values ​​are based on the analysis results of a large amount of historical data, and are typically set to 0.4, 0.3, and 0.3, respectively, while satisfying the constraint that the sum of the weights is 1. This calculation formula can comprehensively reflect the contribution of each parameter to the yarn appearance quality, generating an evaluation value with clear physical meaning. During the normalization transformation process, the yarn surface hairiness density value is processed as follows: the minimum value of this parameter is obtained from the historical database. and maximum value The original feather density value obtained from the current detection The values ​​are converted to normalized values ​​within the range of [0,1] using a linear transformation. A similar processing method is applied to both yarn surface texture clarity and yarn surface color difference distribution values, ensuring that all parameters are within the same numerical range for subsequent calculations. The weighting strategy needs to consider production process requirements and product quality standards. The specific values ​​of the weighting coefficients are determined by analyzing the correlation strength between each parameter and the final product quality. These weighting coefficients need to be periodically verified and adjusted based on new test data.

[0057] Once the evaluation value calculation is complete, the system compares and analyzes the generated surface condition evaluation value of the yarn appearance layer with the preset performance condition evaluation benchmark range. The benchmark range is set based on statistical process control principles. By collecting a large amount of evaluation value data under normal production conditions, its probability distribution characteristics are calculated to determine appropriate control limits. Typically, the mean plus or minus three standard deviations is used as the upper and lower limits of the benchmark range, covering normal production data. When the evaluation value exceeds the benchmark range, the system automatically triggers the corresponding calibration operation command, prompting production personnel to adjust process parameters or perform equipment maintenance checks. The entire implementation process forms a complete closed loop of data acquisition, processing, analysis, and feedback, achieving continuous monitoring and optimization of yarn appearance quality.

[0058] Example 2: See Figure 3The current internal structural parameters of the fiber structure layer include fiber packing density, fiber torsion angle, and fiber cross-sectional uniformity. Fiber packing density is calculated by acquiring cross-sectional images of the yarn using a scanning electron microscope and employing image processing algorithms to calculate the fiber filling coefficient. Specifically, this involves measuring the ratio of the actual cross-sectional area of ​​the fiber to the theoretical cross-sectional area of ​​the yarn. This parameter reflects the packing density and arrangement of the fibers within the yarn. The fiber torsion angle is measured by tracking the helical trajectory of a single fiber using digital image correlation technology. The measurement result is obtained by calculating the angle between the fiber axis and the yarn axis. This parameter characterizes the degree of influence of the twisting process on fiber orientation. The fiber cross-sectional uniformity is calculated by analyzing the cross-sectional morphology variations of multiple fibers and using statistical methods to calculate the coefficient of variation of the cross-sectional dimensions. This parameter reflects the consistency of fiber geometry and structural stability.

[0059] The current molecular motion parameters of the molecular chain layer include molecular chain vibrational frequency, molecular chain orientation degree, and molecular chain crystallinity. The molecular chain vibrational frequency is determined using Fourier transform infrared spectroscopy to detect the vibrational characteristics of chemical bonds. The molecular motion state is determined by analyzing the wavenumber positions of characteristic absorption peaks; this parameter reflects the dynamic characteristics of the molecular chain. The molecular chain orientation degree is measured using X-ray diffraction to determine the crystal orientation distribution. The quantification result is obtained by calculating the Hermann orientation factor; this parameter characterizes the degree of order in the arrangement of macromolecular chains. The molecular chain crystallinity is determined using differential scanning calorimetry to determine the enthalpy of fusion. The crystallinity ratio is calculated by comparing it with the theoretical value for fully crystalline materials; this parameter reflects the degree of crystallinity perfection of the material.

[0060] Obtaining the second current performance status assessment value of the fiber structure layer requires standardization of various parameters. The standardization process uses the Z-score method to convert the raw data into a distribution with a mean of zero and a standard deviation of 1. The arithmetic mean and standard deviation of historical data are used as baseline parameters during the transformation. Based on the standardized parameter values, a weighted fusion calculation is performed using a preset dynamic influence factor for the fiber structure layer. This dynamic influence factor is adjusted according to real-time environmental conditions and operating status to reflect the dynamic changes in production process parameters. A comprehensive evaluation calculation is performed using mathematical formulas during implementation.

[0061]

[0062] in: This represents the assessment value of the internal state of the fiber structure layer. The weighting coefficients representing the density of fiber arrangement. This represents the standardized fiber packing density value. The weighting coefficients representing the fiber torsion angle values. This represents the standardized fiber torsion angle value. The weighting coefficient representing the fiber cross-sectional uniformity value. This represents the standardized fiber cross-sectional uniformity value. Represents dynamic influencing factors. Weighting coefficients. , and The values ​​were determined using principal component analysis, and the dynamic influencing factors were... The results are calculated using empirical formulas based on real-time monitored parameters such as temperature, humidity, and equipment operating status.

[0063] Obtaining the third current performance state evaluation value of the molecular chain layer requires feature quantization. Feature quantization uses principal component analysis (PCA) to convert the original parameters into dimensionless eigenvalues, eliminating dimensionality through linear transformation and extracting the main feature components. At this stage, PCA plays a crucial role in dimensionality reduction and dimensional unification of the original parameters. First, the molecular chain vibrational frequency, molecular chain orientation, and molecular chain crystallinity values ​​are constructed into a three-dimensional original parameter matrix. By calculating the eigenvalues ​​and eigenvectors of the covariance matrix, principal component dimensions with a cumulative contribution rate ≥ 85% are selected. The three-dimensional original parameters are then mapped to a low-dimensional principal component space, generating dimensionless eigenvalues. , , This process preserves the core variation information of the original parameters while eliminating dimensional differences between different parameters, laying a standardized data foundation for subsequent multi-dimensional aggregation calculations. This stage does not involve weight allocation or interaction analysis of the evaluation values; it only completes data preprocessing. Based on the parameter values ​​after feature quantification, multi-dimensional aggregation calculations are performed using preset molecular chain layer coupling coefficients. These coupling coefficients reflect the interaction relationships between parameters, and their values ​​are determined through correlation and regression analysis. Mathematical formulas are used for comprehensive evaluation calculations during implementation.

[0064]

[0065] in: This represents the evaluation value of the molecular chain layer motion state. Weighting coefficients representing the vibrational frequency values ​​of molecular chains. This represents the vibrational frequency value of the molecular chain after feature quantization. The weighting coefficients representing the molecular chain orientation values This represents the molecular chain orientation value after feature quantization. The weighting coefficients representing the crystallinity of the molecular chain This represents the crystallinity value of the molecular chain after feature quantification. This represents the coupling coefficient. The weighting coefficient and coupling coefficient are determined based on the results of multiple regression analysis using a large amount of experimental data, reflecting the contribution of each parameter to the molecular motion state and the strength of their interactions. Here, the multiple regression analysis is only used for the single-parameter weighting coefficient. , , Independent calibration was performed, with the measured evaluation value of the molecular chain layer motion state as the dependent variable, and the quantified value obtained by principal component analysis. , , Construct a linear multiple regression model with variables as independent variables. The independent contribution weights of each individual parameter to the evaluation value are obtained by fitting using the least squares method. This process only characterizes the linear influence of a single parameter and does not consider the interaction effects between parameters. Therefore, a coupling coefficient needs to be introduced subsequently to improve the characterization of the synergistic effect of multiple parameters. During the standardization transformation process, the processing of fiber packing density values ​​requires first calculating the average value of historical data. and standard deviation , will the current detection value Convert to standardized value Similar processing methods were applied to fiber torsion angle values ​​and fiber cross-sectional uniformity values, ensuring that all parameters were converted to standardized values ​​of the same scale. Dynamic Influence Factor The calculation takes into account the ambient temperature. relative humidity and equipment speed And other factors, adopt The functional relationship is obtained by fitting experimental data. The above three methods form a phased and functional collaborative application system. Principal component analysis is the preprocessing step, solving the problem of redundancy in the dimensions and units of the original parameters; multiple regression analysis is the single-factor weight calibration step, clarifying the independent contribution of a single parameter to the evaluation value; response surface methodology is the interaction effect quantification step, supplementing the influence of the synergistic effect of multiple parameters. The three are sequentially connected and each performs its own function, jointly completing the accurate calculation of the molecular chain layer motion state evaluation value, without any problem of methodological application conflict.

[0066] Feature quantization employs principal component analysis (PCA). First, a parameter matrix containing molecular chain vibrational frequencies, orientation degrees, and crystallinity is constructed. Then, the eigenvalues ​​and eigenvectors of the covariance matrix are calculated, and the principal component directions are selected for data projection. The quantified eigenvalues ​​retain the main variation information of the original data while eliminating dimensional differences. Since multiple regression analysis cannot characterize the interactions between parameters, response surface methodology is introduced to specifically calibrate the coupling coefficients. Principal component quantization , , Using the measured evaluation value of the molecular chain layer motion state as the influencing factor and the actual evaluation value as the response value, a three-factor, three-level response surface experiment was designed, and a quadratic multinomial regression model was fitted. in For constant terms, This is a single-factor coefficient. The coefficient of the interaction term. The coefficients are quadratic terms. The combined effect value of the three-parameter interaction terms in the model is extracted, normalized, and then determined as the coupling coefficients. This characterizes the influence of the synergistic effect among three parameters—molecular chain vibration, orientation, and crystallization—on the evaluation value of the motion state. Coupling coefficient Determining the optimal parameters requires experimental research into the interaction effects between them, employing response surface methodology to establish a nonlinear relationship model. The entire calculation process is automated by a computer system, which collects and processes data in real time and outputs evaluation results, providing a basis for adjusting production processes. The evaluation results are compared with preset thresholds, and when outliers occur, corresponding calibration procedures are initiated, achieving continuous optimization of the production process through a feedback mechanism.

[0067] Example 3: Implementing the matching judgment operation requires constructing a benchmark detection strategy set based on a historical detection dataset. This dataset originates from long-term accumulated production quality records, containing the parameter detection values ​​of each layer of yarn from different batches and their corresponding quality rating results. Through cluster analysis and pattern recognition of this historical data, a standardized set of detection strategies is formed, where each strategy is associated with a parameter range and processing suggestions. Real-time detection feedback data comes from the current production line's online monitoring system, including environmental data such as temperature, humidity, and equipment operating parameters collected in real time by sensors, as well as the performance indicators of each layer of yarn obtained through immediate detection. This real-time data, after preprocessing, is used to construct a compensation detection strategy set, which mainly provides dynamic adjustment schemes for instantaneous fluctuations and abnormal situations during the production process.

[0068] Based on the surface state evaluation values ​​of the yarn exterior layer, the internal state evaluation values ​​of the fiber structure layer, and the motion state evaluation values ​​of the molecular chain layer, the system simultaneously queries both the baseline detection strategy set and the compensation detection strategy set. The query process employs a rule-based matching algorithm. First, it compares the current evaluation value with the threshold range in the baseline strategy set to identify baseline strategies that meet the standard range. Simultaneously, the system analyzes abnormal fluctuation patterns in real-time data and selects adjustment strategies suitable for the current production situation from the compensation strategy set. Finally, through a strategy fusion algorithm, the baseline strategy and the compensation strategy are combined into a target detection strategy combination suitable for the current situation.

[0069] When the surface condition assessment value of the yarn surface layer exceeds the preset performance condition assessment benchmark range, the system will automatically activate the yarn surface layer calibration strategy. For example, in a certain production batch, the yarn surface hairiness density value may be abnormally high, and the surface texture clarity value may be significantly reduced, causing the surface layer assessment value to exceed the normal range. In this case, the calibration strategy may include adjusting the tension settings of the winding machine, optimizing the operating parameters of the yarn guide, or activating the cleaning device to remove dust from the spinning components. The system will record the specific parameter adjustment values ​​and execution time of these calibration operations, forming a complete operation log.

[0070] If the internal condition assessment value of the fiber structure layer exceeds the preset performance condition assessment benchmark range, the corresponding fiber structure layer calibration strategy will be triggered. For example, when an abnormally low fiber packing density value or a fiber twist angle value deviating from the standard range is detected, the system may suggest adjusting the draft ratio of the drawing frame, modifying the twist setting of the roving frame, or optimizing the process parameters of the combing machine. These calibration measures aim to improve the fiber packing and twisting characteristics, restoring the internal structure of the yarn to a normal level.

[0071] When the molecular chain layer motion state assessment value exceeds the preset range, the molecular chain layer calibration strategy will be activated immediately. For example, when the molecular chain orientation value is abnormal or the crystallinity value does not meet the standard requirements, the system may recommend adjusting the temperature and humidity environmental parameters of the workshop, modifying the temperature and time settings of the heat setting process, or changing the yarn storage conditions. These calibration operations focus on regulating the motion state and arrangement structure of the molecular chains, thereby improving the intrinsic quality characteristics of the yarn. The entire matching judgment and calibration triggering process is implemented through a specially designed decision support system, which integrates a rule engine and a real-time data processing module. The rule engine is responsible for executing the strategy matching logic, while the real-time data processing module continuously monitors the dynamic changes in the production environment. All strategy execution results are fed back into the system to optimize the decision-making process. This implementation method makes yarn quality inspection no longer an isolated single-point inspection, but forms a closed-loop quality control system that runs through the entire production process.

[0072] In practical applications, the system establishes a strategy execution effectiveness evaluation mechanism. After each calibration strategy execution, the system re-examines the performance indicators of each layer, compares the data changes before and after calibration, and evaluates the effectiveness of the calibration strategy. For strategies with significant effects, the system includes them in the preferred strategy library; for strategies with unsatisfactory effects, the system analyzes the reasons and makes adjustments and optimizations. This continuous improvement mechanism ensures the adaptability and effectiveness of the detection strategies. To ensure the accuracy of strategy execution, the system also sets up multiple verification mechanisms. Before automatically executing a calibration strategy, the system requires operators to confirm the strategy suggestions to avoid the risks that may arise from relying entirely on automation. At the same time, the system regularly reviews and updates the strategy library, removing outdated strategies and adding new optimized strategies. This combination of manual and automated methods ensures both detection efficiency and operational safety.

[0073] The data management during implementation employs a distributed storage architecture. Historical test data is stored in a relational database, while real-time data is processed efficiently using a time-series database. The data query and analysis interfaces are standardized, supporting direct access to various data analysis tools. This data management approach ensures efficient processing and rapid access to massive amounts of test data, providing reliable data support for real-time decision-making. The entire implementation system also considers the characteristics of different production equipment. For different types of spinning equipment, the system loads corresponding equipment parameter models and strategy sets. When production equipment is changed or product specifications are adjusted, the system can quickly switch to the appropriate test strategy set, maintaining the continuity and accuracy of quality testing. This flexible design allows this implementation method to adapt to diverse production environments and product requirements.

[0074] Example 4: See Figure 4 The systematic implementation of the calibration feedback process requires the establishment of a comprehensive data recording and comparative analysis mechanism. The execution results of the yarn appearance layer calibration strategy are recorded through a distributed data acquisition system, including changes in surface image parameters before and after calibration, equipment adjustment parameters, environmental condition data, and evaluation values ​​obtained from re-testing. This data is stored in a high-frequency sampling database in time-series format, retaining complete operation timestamps and equipment status information. The execution data of the fiber structure layer calibration strategy is collected through a dedicated structure monitoring module, recording microscopic changes in fiber arrangement, torsion characteristics, and cross-sectional morphology, while also preserving detailed records of process parameter adjustments. The execution data of the molecular chain layer calibration strategy originates from a molecular motion monitoring system, recording the impact of environmental parameter adjustments such as temperature and humidity on molecular chain motion characteristics, including changes in vibrational frequency, orientation, and crystallinity.

[0075] The expected results data for the target detection strategy combination are derived from simulation predictions and historical experience values ​​before strategy execution. This data is stored in the expected effect database of the strategy management system. Expected data includes the theoretical improvement values ​​of each layer's parameters, the expected range of changes in evaluation values, and the estimated improvement in quality indicators. The comparative analysis of the execution result data and the expected result data employs a multi-dimensional difference analysis method, performing time series alignment to ensure temporal consistency in the data comparison. The deviation between the actual and expected changes of each parameter is calculated, and statistically significant difference features are extracted using pattern recognition algorithms.

[0076] The process of extracting discrepancies employs root cause analysis, examining potential causes from multiple dimensions, including equipment status, process parameters, environmental conditions, and raw material characteristics. For example, when the improvement in surface hairiness density after yarn appearance layer calibration does not meet expectations, the system analyzes the influence weights of factors such as spinning tension, yarn guide wear, and workshop cleanliness. For differences in fiber structure layer calibration effects, the system focuses on equipment factors such as the accuracy of the drawing frame drafting device and the condition of the roving frame twisting mechanism. Molecular chain layer calibration discrepancy analysis emphasizes the impact of environmental control system stability and the accuracy of heat setting process parameters.

[0077] Table 1: Comparison and Analysis of Calibration Strategy Execution Results and Expected Data

[0078]

[0079] The strategy update step modifies the weight allocation rules of the benchmark detection strategy set based on the extracted calibration difference factors. The modification of the weight allocation rules employs an adaptive learning algorithm, dynamically adjusting the weight coefficients of each parameter according to the degree of influence and frequency of occurrence of the difference factors. For example, when yarn guide wear is identified as a major difference factor affecting the calibration effect of the appearance layer, the system will reduce the proportion of hairiness density in the weight allocation, while increasing the weights of other parameters less affected by this factor. A sliding window mechanism is used during the modification process, considering only difference factor data within a recent period to ensure that the weight adjustment reflects the current production status.

[0080] The confidence parameter adjustment for the compensation detection strategy set is based on the statistical analysis results of the discrepancy factors. The confidence parameter of each compensation strategy is dynamically updated according to its historical execution performance. When a strategy repeatedly produces execution results that significantly deviate from the expected data, the system automatically lowers its confidence rating. Confidence adjustment uses a Bayesian update algorithm, combining prior probabilities and new discrepancy data to calculate the posterior confidence. For strategies with low confidence, the system will limit their usage frequency or require manual confirmation before execution, while simultaneously triggering the strategy optimization process.

[0081] A strategy effectiveness evaluation mechanism was established during the strategy update process. After each correction of weight rules and adjustment of confidence levels, the system tracks the strategy execution effect and evaluates the effectiveness of the updated measures. The effectiveness evaluation uses A / B testing to compare the success rate and improvement effect of strategy execution before and after the update, ensuring the correctness of the update direction. For updates with significant effects, the system will solidify them as long-term rules; for updates with unsatisfactory effects, the system will roll back to the previous version and re-analyze the differences. The entire strategy update system adopts version control management. All changes to weight rules and confidence parameters are recorded with detailed version information, including update time, reason for change, impact assessment, and other metadata. This version management facilitates the tracing of historical change records and enables rapid identification of causes and remedial measures when problems occur. The system regularly generates strategy update reports, summarizing and analyzing recent major changes and their effects, providing a reference for strategy optimization. The strategy update process also considers the characteristic differences of different production batches and product specifications. For special batches or products with special specifications, the system will establish independent strategy subsets to avoid negative impacts of general strategy updates on special production situations. These special policy subsets remain relatively independent of the main policy set, but they also participate in the overall update and learning process to ensure the integrity and consistency of the policy system.

[0082] Data security measures implemented during the process include simulation testing and verification before policy updates. All changes to weight rules and confidence parameters must be fully verified in a test environment before deployment to the production system. Update operations employ a canary release mechanism, first piloting the update in a small number of production units to confirm its effectiveness before full rollout. This cautious update mechanism ensures production security and system stability during the policy update process.

[0083] Example 5: The system implementation of the data closed-loop step is based on a sound data management system, updating the revised weight allocation rules and adjusted strategy confidence parameters to the historical detection dataset. This process is achieved through a specially designed data synchronization module, which is responsible for coordinating the data flow between the real-time production database and the historical archive database, ensuring that all strategy update information is completely and accurately recorded in the long-term storage system. The historical detection dataset adopts a distributed architecture for storage and includes strategy version information, weight coefficient change records, confidence parameter adjustment logs, and corresponding difference factor analysis reports. All data has precise timestamps and version identifiers, supporting data traceability and version rollback by time dimension. A strict consistency verification mechanism is implemented during the data update process, using checksum verification and data integrity checks to prevent data corruption or loss during transmission or storage.

[0084] When performing simultaneous multi-performance index testing in the next testing cycle based on the updated historical testing dataset, the system loads the latest version of the strategy configuration information. At the start of the testing cycle, the data acquisition module acquires real-time data of parameters for each layer according to a preset sampling frequency. This data includes surface image information acquired through a high-resolution industrial camera, fiber structure data acquired through a microscopic imaging system, and molecular motion parameters measured through a spectral analysis device. All acquired raw data undergoes a preprocessing process, including data cleaning, noise filtering, and outlier removal, to ensure the quality and reliability of the input data. The preprocessed data is then sent to the hierarchical analysis module, which uses updated weight allocation rules to calculate the performance status evaluation values ​​of each layer. These calculation processes fully consider the impact of the latest strategy adjustments.

[0085] After the evaluation value is calculated, the system matches the current result with the updated performance status evaluation benchmark range. This matching process not only compares the numerical values ​​but also analyzes the data's changing trends and fluctuation patterns, using pattern recognition algorithms to identify potential anomalies. When an evaluation value deviates from the normal range, the system selects an appropriate calibration strategy based on the updated strategy confidence parameters. These strategies prioritize those with higher confidence while also considering applicability and execution cost. During strategy execution, the system monitors the execution effect in real time, collecting data on parameter changes after calibration to provide feedback for strategy optimization.

[0086] Multi-layered data quality control measures are implemented throughout the entire data closed-loop process. During data acquisition, equipment calibration and signal conditioning ensure the accuracy of raw data. In data processing, redundant computation and result verification mechanisms guarantee the reliability of the computation process. During data storage, regular data backups and integrity checks maintain data security. These quality control measures form a complete data assurance system, ensuring the stable operation of the entire closed-loop process from data acquisition to policy updates.

[0087] The operation and monitoring of the data closed-loop system are achieved through a dedicated monitoring platform, which displays the data flow status, strategy execution, and system performance indicators in real time. Monitoring data includes system indicators such as data transmission rate, processing latency, and storage utilization, as well as business indicators such as parameter detection values, evaluation results, and strategy execution effectiveness at each layer. When anomalies are detected, the system triggers an alarm mechanism to notify relevant personnel for timely intervention. The monitoring data itself is also incorporated into historical datasets for analyzing system operation and optimizing system performance. A comprehensive data governance mechanism has also been established during the implementation of the data closed-loop system. Data classification and grading standards have been formulated, clarifying the storage periods and usage permissions for different types of data; a data quality assessment system has been established, and historical datasets are regularly audited for quality; data lifecycle management is implemented, and expired data is archived or cleaned up. These governance measures ensure that historical datasets maintain good availability and integrity, providing a reliable data foundation for continuous simultaneous monitoring of multiple performance indicators.

[0088] The system's continuous learning capability is enhanced through a data closed-loop mechanism. After each detection cycle, the system analyzes the new data generated during the detection process, evaluates the effectiveness of the current strategy, and further optimizes the weight rules and confidence parameters based on the evaluation results. This continuous optimization process adopts an incremental learning approach, gradually improving the detection strategy without affecting the normal operation of the system, enabling the system to adapt to changes in production processes and equipment status evolution. Special attention is paid to avoiding overfitting during the optimization process, and regularization techniques and cross-validation are used to ensure the generalization ability of the strategy.

[0089] The implementation of the data closed-loop system also considers the specific needs of different production scenarios. The system supports parallel management of multiple strategy configurations, allowing selection of different strategy sets based on product specifications, equipment models, or process requirements. When switching production scenarios, the system automatically loads the corresponding historical data and strategy configurations, ensuring the continuity and accuracy of the detection process. This flexible design enables the data closed-loop system to adapt to diverse production environments, providing reliable multi-performance index synchronous detection services for various types of cotton textile yarns. The overall operational efficiency of the data closed-loop system is improved through optimized algorithms and parallel processing technology. The data update process uses an incremental update method, transmitting only the changed data to reduce network load; the data processing process utilizes a distributed computing framework, distributing computational tasks to multiple nodes for parallel execution; and the data storage process uses columnar storage and compression technology. These technical measures ensure that the system maintains high operational performance in environments with large amounts of data, meeting the needs of real-time detection.

[0090] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0091] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for simultaneously detecting multiple performance indicators of cotton textile yarns, characterized in that, The method comprises the following steps: obtaining current surface image parameters of a yarn body appearance layer, current internal structure parameters of a fiber structure layer and current molecular motion parameters of a molecular chain layer corresponding to a current production batch of yarn to be detected; performing layered performance analysis on the current surface image parameters of the yarn body appearance layer, the current internal structure parameters of the fiber structure layer and the current molecular motion parameters of the molecular chain layer respectively to obtain a first current performance state evaluation value of the yarn body appearance layer, a second current performance state evaluation value of the fiber structure layer and a third current performance state evaluation value of the molecular chain layer; matching the first current performance state evaluation value, the second current performance state evaluation value and the third current performance state evaluation value with a preset performance state evaluation benchmark range, and triggering a corresponding detection calibration operation according to a matching judgment result; the current surface image parameters of the yarn body appearance layer specifically include yarn surface hairiness density value, yarn surface texture definition value and yarn surface color difference distribution value; the current internal structure parameters of the fiber structure layer specifically include fiber arrangement tightness value, fiber torsion angle value and fiber cross-section uniformity value; the current molecular motion parameters of the molecular chain layer specifically include molecular chain vibration frequency value, molecular chain orientation value and molecular chain crystallinity value; the specific steps for obtaining the first current performance state evaluation value of the yarn body appearance layer are as follows: performing normalization conversion processing on the yarn surface hairiness density value, the yarn surface texture definition value and the yarn surface color difference distribution value; based on the yarn surface hairiness density value, the yarn surface texture definition value and the yarn surface color difference distribution value after normalization conversion processing, and combining a preset yarn body appearance layer weight distribution strategy for comprehensive calculation, the first current performance state evaluation value of the yarn body appearance layer is generated; the specific steps for obtaining the second current performance state evaluation value of the fiber structure layer are as follows: performing standardization conversion processing on the fiber arrangement tightness value, the fiber torsion angle value and the fiber cross-section uniformity value; based on the fiber arrangement tightness value, the fiber torsion angle value and the fiber cross-section uniformity value after standardization conversion processing, and combining a preset fiber structure layer dynamic influence factor for weighted fusion calculation, the second current performance state evaluation value of the fiber structure layer is generated; the specific steps for obtaining the third current performance state evaluation value of the molecular chain layer are as follows: performing feature quantization processing on the molecular chain vibration frequency value, the molecular chain orientation value and the molecular chain crystallinity value; based on the molecular chain vibration frequency value, the molecular chain orientation value and the molecular chain crystallinity value after feature quantization processing, and combining a preset molecular chain layer coupling coefficient for multi-dimensional aggregation calculation, the third current performance state evaluation value of the molecular chain layer is generated.

2. The method according to claim 1, wherein, The specific steps for performing matching judgment operation are as follows: construct a benchmark detection strategy set based on historical detection data set, and construct a compensation detection strategy set based on real-time detection feedback data; the benchmark detection strategy set contains parameter detection values of each layer of yarn of different batches and corresponding quality rating results; Through cluster analysis and pattern recognition, a standardized detection strategy set is formed, each strategy being associated with parameter ranges and processing suggestions; the benchmark detection strategy set mainly provides dynamic adjustment schemes for transient fluctuations and abnormal conditions in the production process; According to the first current performance state evaluation value of the yarn body appearance layer, the second current performance state evaluation value of the fiber structure layer and the third current performance state evaluation value of the molecular chain layer, a target detection strategy combination is selected from the benchmark detection strategy set and the compensation detection strategy set, specifically including: Using a rule-based matching algorithm, first, the first current performance state evaluation value of the yarn body appearance layer, the second current performance state evaluation value of the fiber structure layer and the third current performance state evaluation value of the molecular chain layer are compared with the threshold range in the benchmark strategy set to identify the benchmark strategy that meets the standard range; at the same time, the system analyzes the abnormal fluctuation pattern in the real-time data to select an adjustment strategy from the compensation strategy set that is suitable for the current production condition; finally, through a strategy fusion algorithm, the benchmark strategy and the compensation strategy are combined into a target detection strategy combination suitable for the current situation.

3. The method according to claim 2, wherein, The specific steps of triggering the corresponding detection calibration operation according to the matching judgment are as follows: If the first current performance state evaluation value of the yarn body appearance layer exceeds the preset performance state evaluation benchmark range of the yarn body appearance layer, the yarn body appearance layer calibration strategy is started; If the second current performance state evaluation value of the fiber structure layer exceeds the preset performance state evaluation benchmark range of the fiber structure layer, the fiber structure layer calibration strategy is started; If the third current performance state evaluation value of the molecular chain layer exceeds the preset performance state evaluation benchmark range of the molecular chain layer, the molecular chain layer calibration strategy is started.

4. The method according to claim 3, wherein, It also includes a calibration feedback step: Record the execution result data of the yarn body appearance layer calibration strategy, the fiber structure layer calibration strategy and the molecular chain layer calibration strategy; Compare the execution result data with the expected result data of the target detection strategy combination to extract calibration difference factors.

5. The method according to claim 4, wherein, It also includes a strategy updating step: Based on the calibration difference factors, the weight distribution rules of the benchmark detection strategy set are corrected; Based on the calibration difference factors, the strategy confidence parameters of the compensation detection strategy set are adjusted.

6. The method according to claim 5, wherein, It also includes a data closed loop step: The corrected weight distribution rules and the adjusted strategy confidence parameters are updated to the historical detection data set; Based on the updated historical detection data set, the multi-performance index synchronous detection in the next detection cycle is carried out.

Images