Textile product production data real-time acquisition method and system

Abstract

The invention relates to the technical field of data processing, in particular to a textile product production data real-time acquisition method and system, and the method comprises the steps: carrying out the multi-label labeling of a surface image of a textile at each historical moment, obtaining a quality data set of a corresponding label, obtaining the attention degree of each label type, and obtaining a quality data set of the corresponding label; taking the attention degree of each label type as the weight of the data of each label type to perform weighted summation, and taking the weighted summation as the comprehensive defect degree of the corresponding quality data set at each moment; and dividing defect samples according to the comprehensive defect degree, calculating an average value of loss contribution coefficients of the defect samples, taking a reciprocal of the average value as an adjustment weight of an initial time interval of the collected data, and obtaining an optimal collection interval of real-time collection of the textile product production data according to the adjustment weight. By calculating the loss contribution coefficient of the defect sample, the time interval of data acquisition is dynamically adjusted, and the data acquisition efficiency in the production process is improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and more specifically, to a method and system for acquiring textile product production data in real time. Background Art

[0002] Textile products are products processed through textile technology and are used in many fields such as clothing, home furnishing, industry, and medical treatment. These products include but are not limited to clothing fabrics, bedding, curtains, carpets, filter materials, insulation materials, safety protection equipment, etc. The production process of textile products involves multiple steps, each of which has an important impact on the quality of the final product, so it is necessary to conduct quality inspection of textile products; in the modern textile industry, quality inspection not only relies on manual inspection, but also increasingly adopts automation and digitalization technologies, such as machine vision systems, spectral analysis, automatic measuring equipment, etc., to capture key quality indicators such as color, tension, size, etc., to improve the accuracy and efficiency of inspection.

[0003] The existing Chinese patent application document with publication number CN118468008A discloses a real-time detection method and device for hose production line data. The method includes: firstly obtaining the production data points and the corresponding production data vectors corresponding to the production time, obtaining the eigenvectors corresponding to all the eigenvalues ​​of the covariance matrix according to the production data vectors, and recording the eigenvalues ​​as the first characterization values ​​of the eigenvectors corresponding to the corresponding eigenvalues; obtaining the second characterization value and the third characterization value of each eigenvector according to the number of projection points on the straight line where each eigenvector is located and the distance between the adjacent projection points on the straight line where each eigenvector is located; obtaining the target eigenvector according to the first characterization value, the second characterization value and the third characterization value of the eigenvector, obtaining the compressed data according to the target eigenvector, and detecting the production process of the hose production line according to the compressed data.

[0004] The application document can reliably detect the production process of the hose production line in real time. However, in the process of obtaining production data of textile products in real time, a time interval for collecting data is required. However, since most textile products are defect-free and a small number are defective in the actual production process, if a fixed time interval is used to obtain production quality data or the collection time interval is large, it is easy to miss the production data of defective textile products. When the collection time interval is small, the efficiency of collecting production data of textile products is low. Summary of the invention

[0005] In order to solve the problem that when obtaining production quality data at fixed time intervals or when the collection time interval is large, it is easy to miss the production data of defective textile products, and when the collection time interval is small, the efficiency of collecting production data of textile products is low, the present invention provides solutions in the following aspects.

[0006] In a first aspect, a method for real-time acquisition of textile product production data includes: performing multi-label annotation on the surface images of textiles at various historical moments, obtaining a quality data set of corresponding labels, and performing preprocessing; obtaining the degree of attention of each label type according to the abnormal distribution of the same label type in the quality data set, taking the degree of attention of each label type as the weight of the data of each label type for weighted summation as the comprehensive defect degree of the corresponding quality data set at each moment; dividing defect samples according to the comprehensive defect degree, calculating the average value of the loss contribution coefficients of the defect samples, taking the inverse of the average value as the adjustment weight for the initial time interval of the collected data, and obtaining the optimal collection interval for real-time collection of textile product production data according to the adjustment weight; wherein the optimal collection interval satisfies the following relationship: , where represents the optimal collection interval, represents the average value of the loss contribution coefficient of the defective samples, Indicates the initial time interval.

[0007] The effect is that by multi-labeling the textile surface images at each historical moment, various types of textile defects can be fully captured, and the average value of the loss contribution coefficient of the defective samples can be used to dynamically adjust the time interval for collecting data, making the collection process more flexible and efficient. The adaptive collection interval can optimize the data collection frequency according to the actual defect situation in the production process, thereby improving production efficiency while ensuring the quality of inspection. By calculating the degree of attention of each label type and taking it as a weighted sum, the comprehensive defect degree of the corresponding quality data set at each moment can be obtained, which can be more concentratedly allocated to the defect types that have a greater impact on product quality, thereby achieving more effective quality control in the production process.

[0008] Preferably, the labels are: color difference, damage, stains, size deviation, and the data of labeling each surface image is obtained to obtain the quality data set at each historical moment; A median filter algorithm is used to remove noise from the quality data set at each moment, and a nearest neighbor interpolation method is used to supplement missing values ​​of the same defect type, so that the quality inspection monitoring data is aligned in time and space, and the format of the quality inspection data is determined.

[0009] Preferably, the degree of concern includes: Taking any moment as the target moment, taking any label in the surface image corresponding to the target moment as the salient label, calculating the absolute value of the difference between the maximum value of the salient label data at the target moment and the salient label data at each historical moment; Calculate the average Euclidean distance between the data of the significant labels in all historical moments and the data of the significant labels in the target moment; The sum of the absolute value of the difference and the average distance is negatively calculated and then exponentially decayed to obtain the ratio of the number of data with significant labels in the historical moments and the number of data with significant labels in the target moment within a preset difference range to the total number of data with significant labels in all historical moments, thereby obtaining a frequency factor; The sum of the frequency factor and the result after exponential decay is normalized and used as the first attention degree of the data of the significant label at the target moment, and the first attention degree is used as the attention degree.

[0010] The effect is as follows: by calculating the absolute value of the difference between the significant label data at the target moment and the maximum value of the historical data, the significant defects that appear at a specific moment can be identified. The average value of the Euclidean distance is used to measure the similarity between the defects at the target moment and the historical data, which helps to evaluate the prevalence and trend of the defects. By exponentially decaying the negative value of the sum of the absolute value of the difference and the average distance, the severity and prevalence of the defects can be weighted, so that more serious or more prevalent defects receive more attention.

[0011] Preferably, the degree of attention also includes: Based on the historical sales data and customer feedback data of the same batch of textile products, the probability of each label type leading to textile product returns or complaints is obtained, and a weight is preset for each label, and the weighted sum is performed to obtain the second attention degree of each label data; The first attention degree and the second attention degree are weightedly summed respectively to obtain a fused attention degree, and the fused attention degree is used as the attention degree of the corresponding tag type.

[0012] The effect is: by analyzing historical sales data and customer feedback, it is possible to identify which types of defects are more likely to lead to returns or complaints, thereby more accurately evaluating the actual impact of defects on product quality. Presetting weights for each label type and taking weighted sums can ensure that the quality control process pays more attention to the types of defects that have the greatest impact on consumer satisfaction. Integrating attention can help production managers allocate resources more reasonably, give priority to those defects that have the greatest impact on product quality and customer satisfaction, improve production efficiency and reduce costs.

[0013] Preferably, the loss contribution coefficient includes: Taking any moment as the target moment, calculate the ratio between the comprehensive defect degree of the quality data set corresponding to the label at the target moment and the average value of the comprehensive defect degree of all quality data sets, and obtain the relative defect degree of the quality data set corresponding to the label at the target moment; Calculate the reciprocal of the absolute value of the difference between the comprehensive defect degree at the target moment and the defect threshold to obtain the defect sensitivity; The sum of the defect sensitivity and the relative defect degree is normalized and used as the loss contribution coefficient of the quality data set corresponding to the label at the target moment.

[0014] The effect is: by calculating defect sensitivity, emphasizing defects that are close to or exceed the preset defect threshold, these key defects receive more attention in quality control. Combining relative defect degree and defect sensitivity can balance the impact of different defect types and severity on overall quality, ensuring that all key defects are properly considered.

[0015] Preferably, the loss contribution coefficient also includes: Taking any moment as the target moment, calculate the ratio between the comprehensive defect degree at the target moment and the comprehensive defect degree of the quality data set corresponding to the label at all moments to obtain the relative defect degree; Calculate the reciprocal of the absolute value of the difference between the comprehensive defect degree at the target moment and the defect threshold to obtain the defect sensitivity; The sum of the defect sensitivity and the relative defect degree is normalized and used as the loss contribution coefficient of the quality data set corresponding to the label at the target moment.

[0016] The effect is that by calculating the defect sensitivity and considering the difference between the comprehensive defect level at the target moment and the defect threshold, defects that are close to or exceed the quality standard can be identified, thereby giving these defects a higher priority in quality control.

[0017] Preferably, the dividing the defect samples according to the comprehensive defect degree includes: The quartiles of the comprehensive defect degree of the quality data set corresponding to all moments are obtained, and the quartiles are used as the defect threshold. If the comprehensive defect degree is less than or equal to the defect threshold, it is classified as a defective sample, otherwise it is a normal sample.

[0018] In a second aspect, a system for real-time acquisition of textile product production data includes: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned method for real-time acquisition of textile product production data is implemented.

[0019] The present invention has the following effects: 1. The present invention calculates the average value of the loss contribution coefficient of the defective samples and uses its reciprocal as an adjustment weight to dynamically adjust the time interval for data collection, which is beneficial to balance the meticulousness and efficiency of detection, ensuring that while maintaining high detection accuracy, the data collection efficiency in the production process is improved.

[0020] 2. The present invention can assign different levels of attention and weights to each defect type by comprehensively considering the defect distribution, sales data, customer feedback and defect thresholds, and pay more attention to defects corresponding to label types that have a greater impact on product quality, thereby improving the ability to identify and handle key defects. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein: Figure 1 It is a method flow chart of steps S1 to S3 in a method for real-time acquisition of textile product production data in an embodiment of the present invention.

[0022] Figure 2 It is a structural block diagram of a system for real-time acquisition of textile product production data according to an embodiment of the present invention. DETAILED DESCRIPTION

[0023] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0024] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0025] Reference Figure 1 A method for real-time acquisition of textile product production data includes steps S1 to S3, which are specifically as follows: S1: Perform multi-label annotation on the surface images of textiles at various historical moments, obtain the quality data set of the corresponding labels, and perform preprocessing.

[0026] Among them, the labels are: color difference, damage, stains, size deviation, and the data of labeling each surface image is obtained, and the quality data set at each historical moment is obtained; specifically, the labeling can be performed through a colorimeter, a visual inspection instrument, a dimension measuring instrument, etc.

[0027] Specific steps of preprocessing: The median filter algorithm is used to remove noise from the quality data set at each moment, and the nearest neighbor interpolation method is used to supplement the missing values ​​of the same defect type, so that the quality inspection monitoring data is aligned in time and space, and the format of the quality inspection data is determined.

[0028] It should be noted that the unit of the collection time interval is seconds. For example, if the preset initial collection time interval is 10 seconds, the first collection time is the 10th second, and the following Moment indicates the moment corresponding to each collection of the surface image of the textile product, that is, a set of quality data in the corresponding surface image is obtained once, which respectively correspond to color difference, damage, stains, and size deviation. The following is an analysis of the quality data set corresponding to each collection moment.

[0029] S2: According to the abnormal distribution of the same label type in the quality data set, the attention level of each label type is obtained, and the attention level of each label type is used as the weight of the data of each label type for weighted summation as the comprehensive defect level of the corresponding quality data set at each moment.

[0030] It should be noted that the abnormal distribution conditions include but are not limited to the following aspects: the frequency of occurrence of label types, the impact of each label type on the quality of textile products, the distribution area of ​​label types and the distribution rules, etc., as the object to be analyzed, the degree of attention of each label is obtained.

[0031] Level of concern, including: Taking any moment as the target moment, taking any label in the surface image corresponding to the target moment as the salient label, calculating the absolute value of the difference between the maximum value of the salient label data at the target moment and the salient label data at each historical moment; Calculate the average Euclidean distance between the data of the significant labels in all historical moments and the data of the significant labels in the target moment; The sum of the absolute value of the difference and the average distance is negatively calculated and then exponentially decayed to obtain the ratio of the number of data with significant labels in the historical moments and the number of data with significant labels in the target moment within the preset difference range to the total number of data with significant labels in all historical moments, and obtain the frequency factor; The sum of the frequency factor and the result after exponential decay is normalized and used as the first attention degree of the data of the significant label at the target moment, and the first attention degree is used as the attention degree.

[0032] Specifically, the first attention degree satisfies the following relationship: ; In the formula, Indicates The moment The first attention level of the data corresponding to each tag type, Indicates The moment Data of tag type, Indicates the first moment in history The maximum value of the data corresponding to each tag type. Indicates the quality data set corresponding to the historical moment The tag type and The quality data set at the moment The number of tag types within the preset difference range, Indicates the history The moment corresponds to the first The data of the first label type is The quality data set at the moment The Euclidean distance between the data of the label type, Indicates the quality data set corresponding to all historical moments The total number of data corresponding to each tag type, represents the normalization function, Represented by natural numbers An exponential function with base .

[0033] That is to say, Indicates the average value of the Euclidean distance, reflecting the The moment Each label type corresponds to the density of the data. Indicates the data set of all samples with The moment The difference in data corresponding to each tag type is The ratio of the number corresponding to the range to the total number of samples reflects its frequency of occurrence. The moment The label type corresponds to the data of all samples The smaller the difference between the first label type, the The smaller the average distance of the data of the label type within the allowable difference range and the higher the frequency of occurrence, the The moment The greater the severity of the corresponding data, the larger the label type, and vice versa.

[0034] It should also be noted that the preset difference range is: ,in, Indicates the first The mean of the data corresponding to each label type, Indicates the first The standard deviation of the data corresponding to each label type, The constant is 1, indicating that the difference range is 1 standard deviation above and below the mean.

[0035] In addition, another embodiment further includes: Based on the historical sales data and customer feedback data of the same batch of textile products, the probability of each label type leading to textile product returns or complaints is obtained, and a weight is preset for each label, and the weighted sum is performed to obtain the second attention degree of each label data; Specifically, the second attention degree satisfies the following relationship: ; In the formula, Indicates The second most popular degree of data corresponding to each tag type is Indicates The weight of the probability of a return or complaint caused by the reason for each label type. Indicates The probability of a return or complaint caused by the reason for the data corresponding to each tag type.

[0036] For example, since size deviation has the smallest impact on the quality of textile products, color difference has a smaller impact on the quality of textile products, stains have a greater impact on the quality of textile products, and damage has the greatest impact on the quality of textile products, the weight should be set to , , , , .

[0037] The first attention degree and the second attention degree are weightedly summed respectively to obtain a fused attention degree, and the fused attention degree is used as the attention degree of the corresponding tag type.

[0038] Specifically, the fusion attention satisfies the following relationship: ; In the formula, Indicates The fusion attention of the data corresponding to each label type, Indicates the weight of the first attention, Indicates The moment The first attention level of the data corresponding to each tag type, Indicates the weight of the second attention level, Indicates The second most popular degree of data corresponding to each tag type.

[0039] For example, , .

[0040] Further analysis shows that the purpose of real-time acquisition of textile product production data is to better identify defects in textile products. According to the logic of defect detection, it is necessary to build a convolutional neural network model and select samples of multiple defect types to train the convolutional neural network model. Therefore, in order to better process these samples, a loss contribution coefficient can be introduced to adjust the contribution of these samples to the overall loss. However, the present invention only relates to a method for real-time acquisition of textile product production data, and therefore, how to build a convolutional neural network model will not be described in detail.

[0041] S3: Divide the defect samples according to the comprehensive defect degree, calculate the average value of the loss contribution coefficient of the defect samples, use the reciprocal of the average value as the adjustment weight for the initial time interval of data collection, and obtain the optimal collection interval for real-time collection of textile product production data based on the adjustment weight.

[0042] Defect samples are divided according to the comprehensive defect degree, including: The quartiles of the comprehensive defect degree of the quality data set corresponding to all moments are obtained, and the quartiles are used as the defect threshold. If the comprehensive defect degree is less than or equal to the defect threshold, it is classified as a defective sample, otherwise it is a normal sample.

[0043] Loss contribution factors, including: Taking any moment as the target moment, calculate the ratio between the comprehensive defect degree of the quality data set corresponding to the label at the target moment and the average value of the comprehensive defect degree of all quality data sets, and obtain the relative defect degree of the quality data set corresponding to the label at the target moment; Calculate the reciprocal of the absolute value of the difference between the comprehensive defect degree at the target moment and the defect threshold to obtain the defect sensitivity; The sum of defect sensitivity and relative defect degree is normalized and used as the loss contribution coefficient of the quality data set corresponding to the label at the target moment.

[0044] Specifically, the loss contribution coefficient satisfies the following relationship: ; In the formula, Indicates The loss contribution coefficient of the quality data set corresponding to the label at the moment, Indicates The comprehensive defect level of the quality data set corresponding to the label at the moment, represents the total number of all quality data sets, Indicates The comprehensive defect level of the quality data set corresponding to the label at the moment, Indicates the defect threshold.

[0045] That is to say, Indicates The difference between the comprehensive defect degree of the quality data set corresponding to the label at the moment and the median of the comprehensive defect degree at all moments is The greater the comprehensive defect level at the moment, the greater the difference The smaller the time, the The loss contribution coefficient of the sample with defects at all times is larger, and vice versa.

[0046] In addition, another embodiment further includes: Taking any moment as the target moment, calculate the ratio between the comprehensive defect degree at the target moment and the comprehensive defect degree of the quality data set corresponding to the label at all moments to obtain the relative defect degree; Calculate the reciprocal of the absolute value of the difference between the comprehensive defect degree at the target moment and the defect threshold to obtain the defect sensitivity; The sum of defect sensitivity and relative defect degree is normalized and used as the loss contribution coefficient of the quality data set corresponding to the label at the target moment.

[0047] Specifically, the loss contribution coefficient satisfies the following relationship: ; In the formula, Indicates The loss contribution coefficient of the quality data set corresponding to the label at the moment, Indicates The comprehensive defect level of the quality data set corresponding to the label at the moment, Represents the maximum comprehensive defect level of the quality data set corresponding to the label at all times, Indicates the defect threshold.

[0048] That is to say, Indicates The ratio of the comprehensive defect level at the moment to the maximum defect level, thus reflecting the The relative position of the comprehensive defect degree at a certain moment in all samples; Indicates The inverse of the difference between the comprehensive defect level at the time and the defect threshold. near , this reciprocal term will increase, indicating that the defect level of the sample is close to or exceeds the preset threshold and requires more attention.

[0049] The optimal collection interval satisfies the following relationship: ; In the formula, represents the optimal collection interval, represents the average value of the loss contribution coefficient of the defective samples, Indicates the initial time interval.

[0050] The present invention also provides a system for acquiring textile product production data in real time. Figure 2 As shown, the system includes a processor and a memory, the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a method for real-time acquisition of textile product production data according to the first aspect of the present invention is implemented.

[0051] The system also includes other components familiar to those skilled in the art, such as a communication bus and a communication interface. The configuration and functions of these components are known in the art and will not be described in detail here.

[0052] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus or device. For example, a computer-readable storage medium may be any appropriate magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM (Resistive Random Access Memory), a dynamic random access memory DRAM (Dynamic Random Access Memory), a static random access memory SRAM (Static Random-Access Memory), an enhanced dynamic random access memory EDRAM (Enhanced Dynamic Random Access Memory), a high-bandwidth memory HBM (High-Bandwidth Memory), a hybrid memory cube HMC (Hybrid Memory Cube), etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of a device or accessible or connectable to a device. Any application or module described in the present invention may be implemented using computer-readable / executable instructions that may be stored or otherwise maintained by such a computer-readable medium.

[0053] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0054] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A method for real-time acquisition of textile product production data, characterized in that: include: Perform multi-label annotation on the surface images of textiles at various historical moments, obtain the quality data set of the corresponding labels, and perform preprocessing; According to the abnormal distribution of the same label type in the quality data set, the attention level of each label type is obtained, and the attention level of each label type is used as the weight of the data of each label type for weighted summation as the comprehensive defect level of the corresponding quality data set at each moment; Classifying defective samples according to comprehensive defect levels, calculating an average value of loss contribution coefficients of defective samples, using the reciprocal of the average value as an adjustment weight for an initial time interval for collecting data, and obtaining an optimal collection interval for real-time collection of textile product production data according to the adjustment weight; The optimal collection interval satisfies the following relationship: , where represents the optimal collection interval, represents the average value of the loss contribution coefficient of the defective samples, Indicates the initial time interval.

2. A method for real-time acquisition of textile product production data according to claim 1, characterized in that: in, The labels are: color difference, damage, stains, size deviation, and the data of labeling each surface image is obtained to obtain the quality data set at each moment in history; A median filter algorithm is used to remove noise from the quality data set at each moment, and a nearest neighbor interpolation method is used to supplement missing values ​​of the same defect type, so that the quality inspection monitoring data is aligned in time and space, and the format of the quality inspection data is determined.

3. A method for real-time acquisition of textile product production data according to claim 1, characterized in that: The attention degree includes: taking any moment as the target moment, taking any label in the surface image corresponding to the target moment as the significant label, calculating the absolute value of the difference between the data of the significant label at the target moment and the maximum value of the data of the significant label at each historical moment; calculating the average distance of the Euclidean distance between the data of the significant label in all historical moments and the data of the significant label in the target moment; taking the sum of the absolute value of the difference and the average distance as a negative value and then performing exponential decay, obtaining the ratio of the number of data of the significant label in the historical moment and the data of the significant label in the target moment within a preset difference range to the total number of data of the significant label in all historical moments, obtaining a frequency factor, adding the frequency factor and the result after exponential decay and normalizing the sum as the first attention degree of the data of the significant label in the target moment, and taking the first attention degree as the attention degree.

4. The method for real-time acquisition of textile product production data according to claim 1, characterized in that: The degree of concern also includes: Based on the historical sales data and customer feedback data of the same batch of textile products, the probability of each label type leading to textile product returns or complaints is obtained, and a weight is preset for each label, and the weighted sum is performed to obtain the second attention degree of each label data; The first attention degree and the second attention degree are weightedly summed respectively to obtain a fused attention degree, and the fused attention degree is used as the attention degree of the corresponding tag type.

5. The method for real-time acquisition of textile product production data according to claim 1, characterized in that: The loss contribution coefficient includes: Taking any moment as the target moment, calculate the ratio between the comprehensive defect degree of the quality data set corresponding to the label at the target moment and the average value of the comprehensive defect degree of all quality data sets, and obtain the relative defect degree of the quality data set corresponding to the label at the target moment; Calculate the reciprocal of the absolute value of the difference between the comprehensive defect degree at the target moment and the defect threshold to obtain the defect sensitivity; The sum of the defect sensitivity and the relative defect degree is normalized and used as the loss contribution coefficient of the quality data set corresponding to the label at the target moment.

6. A method for real-time acquisition of textile product production data according to claim 1, characterized in that: The loss contribution coefficient also includes: Taking any moment as the target moment, calculate the ratio between the comprehensive defect degree at the target moment and the comprehensive defect degree of the quality data set corresponding to the label at all moments to obtain the relative defect degree; Calculate the reciprocal of the absolute value of the difference between the comprehensive defect degree at the target moment and the defect threshold to obtain the defect sensitivity; The sum of the defect sensitivity and the relative defect degree is normalized and used as the loss contribution coefficient of the quality data set corresponding to the label at the target moment.

7. A method for real-time acquisition of textile product production data according to claim 1, characterized in that: The method of dividing defect samples according to the comprehensive defect degree includes: The quartiles of the comprehensive defect degree of the quality data set corresponding to all moments are obtained, and the quartiles are used as the defect threshold. If the comprehensive defect degree is less than or equal to the defect threshold, it is classified as a defective sample, otherwise it is a normal sample.

8. A real-time acquisition system for textile product production data, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the method for real-time acquisition of textile product production data according to any one of claims 1-7 is implemented.

Images