CN117273554B - Textile production quality prediction method based on data identification - Google Patents
Textile production quality prediction method based on data identification Download PDFInfo
- Publication number
- CN117273554B CN117273554B CN202311574158.5A CN202311574158A CN117273554B CN 117273554 B CN117273554 B CN 117273554B CN 202311574158 A CN202311574158 A CN 202311574158A CN 117273554 B CN117273554 B CN 117273554B
- Authority
- CN
- China
- Prior art keywords
- data
- textile
- quality
- raw material
- sample
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 239000004753 textile Substances 0.000 title claims abstract description 91
- 238000000034 method Methods 0.000 title claims abstract description 62
- 238000004519 manufacturing process Methods 0.000 title claims abstract description 41
- 239000002994 raw material Substances 0.000 claims abstract description 122
- 238000001514 detection method Methods 0.000 claims abstract description 50
- 238000012545 processing Methods 0.000 claims abstract description 25
- 238000004458 analytical method Methods 0.000 claims abstract description 22
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 17
- 238000012549 training Methods 0.000 claims abstract description 15
- 230000008569 process Effects 0.000 claims abstract description 14
- 230000002159 abnormal effect Effects 0.000 claims abstract description 12
- 238000003066 decision tree Methods 0.000 claims abstract description 11
- 238000003064 k means clustering Methods 0.000 claims abstract description 10
- 239000000463 material Substances 0.000 claims abstract description 8
- 230000005856 abnormality Effects 0.000 claims abstract description 6
- 239000000835 fiber Substances 0.000 claims description 69
- 238000012360 testing method Methods 0.000 claims description 24
- 229920000742 Cotton Polymers 0.000 claims description 11
- 210000002268 wool Anatomy 0.000 claims description 11
- 238000009826 distribution Methods 0.000 claims description 10
- 239000004744 fabric Substances 0.000 claims description 10
- 230000000704 physical effect Effects 0.000 claims description 10
- 244000025254 Cannabis sativa Species 0.000 claims description 9
- 235000012766 Cannabis sativa ssp. sativa var. sativa Nutrition 0.000 claims description 9
- 235000012765 Cannabis sativa ssp. sativa var. spontanea Nutrition 0.000 claims description 9
- 235000009120 camo Nutrition 0.000 claims description 9
- 235000005607 chanvre indien Nutrition 0.000 claims description 9
- 239000011487 hemp Substances 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 8
- 238000005520 cutting process Methods 0.000 claims description 8
- 238000005286 illumination Methods 0.000 claims description 8
- 238000002790 cross-validation Methods 0.000 claims description 6
- 238000007405 data analysis Methods 0.000 claims description 5
- 238000004043 dyeing Methods 0.000 claims description 5
- 238000011496 digital image analysis Methods 0.000 claims description 4
- 238000005516 engineering process Methods 0.000 claims description 4
- 238000009864 tensile test Methods 0.000 claims description 4
- 238000012795 verification Methods 0.000 claims description 4
- 238000010191 image analysis Methods 0.000 claims description 3
- 230000014759 maintenance of location Effects 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 230000035699 permeability Effects 0.000 claims description 3
- 238000009987 spinning Methods 0.000 claims description 3
- 230000000844 anti-bacterial effect Effects 0.000 claims description 2
- 238000005452 bending Methods 0.000 claims description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 2
- 230000009471 action Effects 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 230000007547 defect Effects 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 238000007781 pre-processing Methods 0.000 description 3
- 238000010200 validation analysis Methods 0.000 description 3
- 230000000845 anti-microbial effect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000004140 cleaning Methods 0.000 description 2
- 238000003708 edge detection Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000010998 test method Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 229920004933 Terylene® Polymers 0.000 description 1
- 238000005299 abrasion Methods 0.000 description 1
- 230000002547 anomalous effect Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 239000005020 polyethylene terephthalate Substances 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000003908 quality control method Methods 0.000 description 1
- 238000000611 regression analysis Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000012731 temporal analysis Methods 0.000 description 1
- 238000000700 time series analysis Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06395—Quality analysis or management
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/27—Regression, e.g. linear or logistic regression
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/04—Manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Data Mining & Analysis (AREA)
- General Business, Economics & Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Educational Administration (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Game Theory and Decision Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Quality & Reliability (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Operations Research (AREA)
- Primary Health Care (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Manufacturing & Machinery (AREA)
- Treatment Of Fiber Materials (AREA)
Abstract
The invention relates to the technical field of textile, and discloses a textile production quality prediction method based on data identification, which comprises the following steps: s1: collecting data related to textile quality, comprising: textile materials, production process parameters and textile machinery sensor data; s2: the data acquired in the step S1 are cleaned and preprocessed, and the method comprises the following steps: processing the missing values: filling and deleting missing data; processing outliers: identifying and processing outlier data points; characteristic engineering: according to the characteristics of the textile, adopting K-means clustering analysis to analyze the textile data and creating new characteristic data; s3: establishing a prediction model for textile production quality: the method comprises the steps of adopting a decision tree model to carry out model training on new feature data created in the step S2, and importing raw material feature data to obtain a textile quality prediction result; s4: for raw materials which are not in A, B, C and I, II and III in S2, an abnormality detection algorithm is adopted to identify and process abnormal raw materials.
Description
Technical Field
The invention relates to the technical field of textile quality prediction, in particular to a textile production quality prediction method based on data identification.
Background
The textile production quality prediction method based on data identification is a technology for predicting the textile production quality by using data analysis, machine learning and statistical methods. This approach improves quality and production efficiency by collecting and analyzing data related to the production process and product quality to identify potential quality problems ahead of time, taking appropriate measures.
Modern textile production typically uses various sensors to monitor critical parameters in the production process, such as temperature, humidity, tension, pressure, etc. These sensors can collect data in real time for monitoring and controlling the production process, and subsequent quality predictions.
The collected data may be processed by machine learning and data analysis techniques. Common methods include regression analysis, classification, clustering, time series analysis, and the like. These methods can be used to identify patterns and trends related to production quality.
Image processing techniques are also important for quality prediction of textiles. Any defect or area of failure can be detected by taking or scanning an image of the textile surface. Computer vision and deep learning techniques may be used for automated defect detection.
Data-based textile production quality prediction methods are often combined with quality control and process optimization. These methods can be used not only to detect problems, but also to provide feedback to improve the manufacturing process and reduce defect rates.
From the historical data and the real-time data, a predictive model can be constructed to predict the quality level of the textile production. These models may be used to aid in decisions such as adjusting production parameters or taking steps in advance to avoid quality problems.
However, a disadvantage of the prior art is that for conventional products, there is typically a large amount of historical data available for modeling and analysis so that accurate quality prediction models can be constructed. However, for non-conventional or custom products, there may be insufficient historical data that can affect the performance and reliability of the model.
Disclosure of Invention
The invention provides a textile production quality prediction method based on data identification, which is used for solving the problems in the background technology.
The invention provides the following technical scheme: the textile production quality prediction method based on data identification is characterized by comprising the following steps of:
S1: collecting data related to textile quality, comprising: textile materials, production process parameters and textile machinery sensor data;
Wherein the textile material: including the kind and quality detection grade information of the raw materials;
production process parameters: recording temperature, humidity and dyeing process data in the spinning process;
Textile machine sensor data: comprises the steps of acquiring tension, pressure and speed sensor data from textile machinery equipment;
s2: the data acquired in the step S1 are cleaned and preprocessed, and the method comprises the following steps:
Processing the missing values: filling and deleting missing data;
processing outliers: identifying and processing outlier data points;
Characteristic engineering: according to the characteristics of the textile, analyzing the textile data by adopting K-means clustering analysis to create new characteristic data, dividing the types of raw materials into A, B, C types under the conventional state, and dividing the quality detection class of each raw material into I, II and III types;
s3: establishing a prediction model for textile production quality: the method comprises the steps of adopting a decision tree model to carry out model training on new feature data created in the step S2, carrying out model tuning by using verification data, importing raw material feature data, and obtaining a textile quality prediction result;
s4: for raw materials which are not in A, B, C and I, II and III in S2, an abnormality detection algorithm is adopted to identify and process abnormal raw materials.
Preferably, in step S1, the quality detection grade information includes detection of physical properties of the textile raw material;
Wherein the physical property information detection comprises a tensile strength test, a fiber thickness test and a pilling performance density test.
Preferably, the types of the characteristic engineering binding raw materials in the step S2 are classified into A, B, C types, wherein the type a is cotton, the type B is hemp, and the type C is wool:
Selecting characteristics, namely selecting characteristics related to the characteristics of the raw materials according to the physical property data of each type of raw materials;
For the A-class cotton, selecting characteristics related to softness and water absorbability in a characteristic engineering, and specifically selecting fiber length and fiber bending characteristics;
for the type B hemp, selecting characteristics related to wear resistance, air permeability and antibacterial property by a characteristic engineering, and specifically selecting fiber strength, fiber thickness and fabric porosity;
and for the C-type wool, the characteristic engineering selects characteristics related to warmth retention and elasticity, such as fiber diameter and fiber elastic modulus.
Preferably, the method comprises the steps of analyzing textile data by adopting K-means clustering analysis according to the characteristics of the textile, and creating new characteristic data, wherein the method specifically comprises the following steps:
a: preparing a data set containing raw material types and quality detection grades;
b: encoding the types and quality detection grades of raw materials, and converting the classification variables into a numerical form by using a One-Hot Encoding (One Encoding) method;
c: data normalization is performed to ensure that the values of the different features lie within the same range;
d: clustering the data by using K-means clustering analysis, dividing the data points into K clusters, wherein K is the selected cluster number, and setting K as the category number, namely, the raw material comprises A, B, C categories one by one, and the quality grades I, II and III category one by 9 clusters in total;
e: carrying out cluster allocation on each data point, determining which cluster the data point belongs to, and representing the raw material type and quality detection grade cluster to which each data point belongs;
f: creating new features to represent the types and quality detection levels of raw materials based on cluster allocation, including creating two new binary features to represent the types and quality detection levels of raw materials respectively, wherein the value of each feature is 0 or 1 to represent that the raw materials belong to a certain category or not;
g: the new characteristic data are used for subsequent data analysis and modeling to determine the influence of the raw material types and quality detection grades on the textile.
Preferably, the weight of the feature is adjusted according to the importance of different raw material types and quality grades;
wherein, raw material types A, B, C and three quality grades I, II, III, the importance of each characteristic is evaluated, and the importance is divided into three grades of high, medium and low, and the formula is as follows:
Feature weight = original weight raw material class weight quality grade weight
Wherein the raw material type weight and the quality grade weight are set according to the importance of different raw material types and quality grades, the weights adopt real numbers from 0 to 1, wherein 1 represents the highest weight, and 0 represents the lowest weight;
Specifically, for class a raw materials, the softness characteristics are important in all quality classes, then the original weight is set to 0.8, and the class a raw material class weight is set to 1;
For the B-class and C-class raw materials, different raw material class weights are set according to the importance of the raw materials, and for different quality grades, different quality grade weights are set.
Preferably, the missing values are processed in S2, and the data is filled with the mean, median or mode according to the data distribution and the distribution of the missing values; if the influence of the missing data on the analysis result is larger, deleting the characteristics containing the missing value;
Processing outliers and detecting outliers using a Z-Score anomaly detection algorithm, comprising: the deviation of the data points from the mean value of the data set is measured, the Z-Score (standard Score) for each data point is calculated, a threshold is set, and data points exceeding the threshold are considered to be abnormal.
Preferably, in S3, a decision tree model is used to train the model, which specifically includes the following steps:
Root node: initially, placing all collected data related to the quality of the textile into a root node;
node splitting: selecting, at each node, a feature and a corresponding threshold to divide the data of the current node into two subsets, the selection of splits being based on the measure of unrepeace, the goal being to select features and thresholds that minimize unrepeace;
creating a child node: each split creates two child nodes, and the feature selection and splitting of the next round is continued, which forms a branch structure of the tree until a stop condition is reached;
And (3) recursion: recursively splitting the child nodes until a stop condition is reached, which may include the depth of the tree reaching a predetermined value, the number of samples contained by the node being less than a threshold;
leaf node marking: when the stop condition is met, the samples in the leaf nodes are marked as a specific class, namely the quality class of the textile, wherein 1 stands for class i, 2 stands for class ii, and 3 stands for class iii.
Preferably, the feature threshold specifically includes:
Fiber length
Threshold value: the threshold is set at 100 mm, when a sample with a fiber length less than 100 mm will enter one subnode, and a sample with a length greater than or equal to 100 mm will enter the other subnode;
Fiber strength
Threshold value: the threshold is set to 50 newtons, when a sample with a fiber strength less than 50 newtons will enter one child node, and a sample with a fiber strength greater than or equal to 50 newtons will enter the other child node;
Porosity of fabric
Threshold value: the threshold is set to 0.2 when a sample with a porosity of the fabric less than 0.2 will enter one child node and a sample greater than or equal to 0.2 will enter the other child node.
Preferably, wherein the tensile strength testing method comprises:
preparing a sample: cutting out a standard-sized sample from the raw material;
clamping a sample: clamping the sample in a clamp of a tensile testing machine;
applying a stretching force: gradually applying a tensile force, and recording data of load and sample elongation;
calculation results: calculating tensile strength and elongation indexes according to the load and elongation data;
The fiber thickness testing method comprises the following steps:
preparing a sample: obtaining a fiber sample from the raw material and preparing an appropriate slice or cross section;
and (3) observing the fiber: measuring the diameter of the fiber using a digital image analysis system;
calculation results: calculating an average thickness index of the fiber according to the measured data;
the specific method for testing the pilling performance density comprises the following steps:
Cutting out a sample with standard size from a textile to be tested, and clamping two ends of the sample;
Determining illumination conditions, including brightness and illumination time of the light source;
a light source is arranged to uniformly irradiate the upper side surface of the sample;
A whiteboard for projection is arranged on the lower surface of the sample;
Shooting a shadow image of the sample projected on the whiteboard by using a camera or an image acquisition device;
Importing the shot image into a computer, and using image analysis software to analyze the quantity and degree of pilling phenomena and pilling areas in the image, and simultaneously analyzing the density distribution of textiles;
the following are the specific steps of the analysis:
(1) Shadow difference analysis: analyzing the shadow difference according to the shadow range and depth of the pixels acquired by the image, wherein the pixels exceeding a preset threshold value represent that the shadow difference exists, and indicate that the density is uneven;
(2) Detecting a pilling position: the pilling phenomenon is evaluated according to the number and degree of the pilling sites by detecting the pilling sites in the image using an image analysis algorithm by searching for high contrast areas in shadows.
The invention has the following beneficial effects:
1. quality-related data is collected from textile production processes, including textile materials, production process parameters, and textile machinery sensor data. These data include textile material type and quality inspection grade information, production process parameters (e.g., temperature, humidity, dyeing process data), and textile machine sensor data (e.g., tension, pressure, speed).
And cleaning and preprocessing the acquired data, including processing missing values, processing abnormal values and performing feature engineering. The feature engineering selects features related to the raw materials according to the characteristics of different raw material types (A, B, C) and quality grades (I, II and III) to analyze and create new feature data.
Training the processed data by adopting a decision tree model, performing model tuning by using verification data, and importing raw material characteristic data to predict the quality grade of textiles.
For raw materials other than A, B, C and of classes I, II and III, an anomaly detection algorithm is used to identify and process anomalous raw materials.
The weights of the features are adjusted according to the importance of different raw material types and quality grades to more accurately influence the prediction of the quality of the textile.
In the data cleaning and preprocessing steps, missing values are filled in by means of average, median or mode, and abnormal values are detected by means of a Z-Score abnormality detection algorithm.
The establishment of the decision tree model includes the creation of root nodes, the selection of node splits is based on an unrepeace metric, child nodes are created until a stop condition is reached, and finally leaf nodes are marked to represent the quality level of the textile.
Threshold settings for different features, such as fiber length, fiber strength, fabric porosity, etc., are used to divide the data into different sub-nodes.
Including tensile strength testing, fiber thickness testing, and pilling performance density testing, are used to evaluate the physical properties of the textile raw materials. The method predicts the quality grade of the textile by analyzing and processing data related to the production of the textile, including raw materials, process parameters, and sensor data. The method is beneficial to improving the prediction and control of the quality of the textile and reducing the defective rate, thereby improving the production quality, adopting machine learning technologies such as a decision tree model and the like, automatically processing a large amount of data and making decisions according to the prediction result of the model. This helps to achieve automated and intelligent management of the production process, reducing the need for human intervention.
Drawings
FIG. 1 is a schematic flow chart of the system of the present invention.
FIG. 2 is a schematic diagram of a pilling performance density test according to the present invention;
fig. 3 is a schematic view of the shadow part of the plate of the present invention.
Description of the embodiments
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Examples
Referring to fig. 1-3, the application discloses a textile production quality prediction method based on data identification, which comprises the following steps:
S1: collecting data related to textile quality, comprising: textile materials, production process parameters and textile machinery sensor data;
Wherein, raw material types: the quality detection grade information of cotton, hemp and wool is divided into: class I, class II, class III.
Production process parameter data:
and (3) temperature recording: average temperature during spinning (degrees celsius).
Humidity recording: average relative humidity (percent) during production.
Dyeing process data: parameters of the dyeing process, such as dye type, concentration, etc.
Textile machine sensor data:
Tension sensor data: the change in tension per minute (newtons) was recorded.
Pressure sensor data: the pressure change per minute (pascal) was recorded.
Speed sensor data: the production rate change per minute (meters per second) was recorded.
S2: the data acquired in the step S1 are cleaned and preprocessed, and the method comprises the following steps:
Processing the missing values: if the humidity data is missing, the missing values are filled with the mean value. For other missing data, the corresponding sample is selected for deletion due to the smaller number.
Processing outliers: in the temperature, humidity, tension, pressure and velocity data, abnormal data points were identified using the Z-Score anomaly detection algorithm and replaced with an average value.
Characteristic engineering: based on the textile characteristics, K-means clustering analysis is adopted to analyze the data. The results show that the raw materials are classified into class A (cotton), class B (silk) and class C (terylene), and the quality detection class is classified into class I, class II and class III, and new characteristic data are created to represent the class and quality grade of the raw materials.
S3: establishing a prediction model for textile production quality: the method comprises the steps of adopting a decision tree model to carry out model training on new feature data created in the step S2, carrying out model tuning by using verification data, importing raw material feature data, and obtaining a textile quality prediction result;
s4: aiming at raw materials which are not in A, B, C and I, II and III in S2, adopting a K-means outlier anomaly detection algorithm to identify and process the anomaly raw materials, wherein the method comprises the following steps of:
The data has been cleaned and preprocessed in step S2, ensuring the data quality. The characteristic data of the raw material not in S2 is integrated with the previous processing result so as to perform abnormality detection.
Training raw material data by using an anomaly detection algorithm, and establishing a model to identify the anomaly raw material. At this stage, cross-validation techniques are used to tune the model parameters to ensure their performance in different contexts.
Unknown raw material data is predicted using a trained anomaly detection model. The model will assign an anomaly score or probability to each raw material data point. By setting the appropriate threshold, it is determined which raw materials are considered abnormal.
Once an abnormal raw material is identified, one of the following actions may be taken:
Removing abnormal raw materials: abnormal raw materials are removed from the production process to ensure the quality of the product.
Further investigation and processing: for abnormal raw materials, the cause thereof is further analyzed and necessary measures are taken to correct the problem to prevent the occurrence of similar problems in the future.
Wherein: the cross-validation adopts K-fold cross-validation, and specifically comprises the following steps:
Data preparation:
the acquired data set is divided into a training set and a testing set.
The value of K, i.e. the number of folds K, is chosen to be 5 or 10, depending on the size of the data volume.
Cross-validation:
the training set is divided into K folds of similar size.
For each cross-validation iteration, one of the folds is selected as the validation set and the remaining K-1 folds are selected as the training set.
In each iteration, the decision tree model is trained using a training set and model performance is evaluated using a validation set.
Performance evaluation:
In each iteration, performance indexes of the model, such as accuracy, F1 fraction, mean square error and the like, are recorded.
And (5) averaging the performance indexes of the K iterations to obtain the comprehensive performance evaluation of the model.
And (3) model tuning:
Different model hyper-parameters are tried in each iteration to determine the optimal hyper-parameter configuration to improve model performance.
And (3) training a final model:
After the best model configuration is selected, all training data (including validation data) is used to train the final model.
In step S1, quality detection grade information comprises detection of physical properties of textile raw materials;
Wherein the physical property information detection comprises a tensile strength test, a fiber thickness test and a pilling performance density test.
Wherein the tensile strength test method comprises:
preparing a sample: cutting out a standard-sized sample from the raw material;
clamping a sample: clamping the sample in a clamp of a tensile testing machine;
applying a stretching force: gradually applying a tensile force, and recording data of load and sample elongation;
calculation results: calculating tensile strength and elongation indexes according to the load and elongation data;
The fiber thickness testing method comprises the following steps:
preparing a sample: obtaining a fiber sample from the raw material and preparing an appropriate slice or cross section;
and (3) observing the fiber: measuring the diameter of the fiber using a digital image analysis system;
calculation results: calculating an average thickness index of the fiber according to the measured data;
the specific method for testing the pilling performance density comprises the following steps: cutting out a sample with standard size from a textile to be tested, and tiling the sample;
Arranging a light source to illuminate a side of the sample;
Determining illumination conditions, including brightness and illumination time of the light source;
capturing an image of the side of the sample using a camera or image acquisition device;
The captured image is imported into a computer and the number and extent of pilling events in the image and pilling areas are analyzed using image analysis software.
Wherein the types of the characteristic engineering combined raw materials in the step S2 are classified into A, B, C types, wherein the A type is cotton, the B type is hemp and the C type is wool:
Selecting characteristics, namely selecting characteristics related to the characteristics of the raw materials according to the physical property data of each type of raw materials;
Wherein, the A-class raw material is cotton
For class a cotton, cotton fibers are generally known for softness and absorbency, the following characteristics are selected:
Fiber length: the average length (in millimeters) of cotton fibers, long fibers are generally associated with softness.
Fiber curvature: the flexibility of the fiber in terms of softness. Fibers with less tortuosity are generally softer.
The B-class raw material is hemp
For group B hemp, the hemp fibers are generally known for abrasion resistance, air permeability and antimicrobial properties, the following characteristics are chosen:
Fiber strength: average strength of fibrilia (in newtons). Higher strength is generally associated with wear resistance.
Fiber thickness: average thickness of fibrilia (in microns). Finer fibers are often associated with breathability.
Porosity of the fabric: percentage of air voids in the textile. Higher porosity is generally associated with breathability while also improving antimicrobial properties.
The C-type raw material is wool
For class C wool, wool fibers are generally known for warmth and elasticity, and the following characteristics are selected:
Fiber diameter: average diameter of wool fibers (in microns). Finer fibers are often associated with warmth retention.
Modulus of elasticity of fiber: the elastic modulus of the wool fibers, i.e., the stretchability of the fibers. The higher modulus of elasticity is generally related to elasticity.
According to the characteristics of the textile, adopting K-means clustering analysis to analyze the textile data and creating new characteristic data, wherein the method specifically comprises the following steps:
a: preparing a data set containing raw material types and quality detection grades;
b: encoding the types and quality detection grades of raw materials, and converting the classification variables into a numerical form by using a One-Hot Encoding (One Encoding) method;
c: data normalization is performed to ensure that the values of the different features lie within the same range;
d: clustering the data by using K-means clustering analysis, dividing the data points into K clusters, wherein K is the selected cluster number, and setting K as the category number, namely, the raw material comprises A, B, C categories one by one, and the quality grades I, II and III category one by 9 clusters in total;
e: carrying out cluster allocation on each data point, determining which cluster the data point belongs to, and representing the raw material type and quality detection grade cluster to which each data point belongs;
f: creating new features to represent the types and quality detection levels of raw materials based on cluster allocation, including creating two new binary features to represent the types and quality detection levels of raw materials respectively, wherein the value of each feature is 0 or 1 to represent that the raw materials belong to a certain category or not;
g: the new characteristic data are used for subsequent data analysis and modeling to determine the influence of the raw material types and quality detection grades on the textile.
According to the importance of different raw material types and quality grades, the weight of the characteristics is adjusted;
wherein, raw material types A, B, C and three quality grades I, II, III, the importance of each characteristic is evaluated, and the importance is divided into three grades of high, medium and low, and the formula is as follows:
Feature weight = original weight raw material class weight quality grade weight
Wherein the raw material type weight and the quality grade weight are set according to the importance of different raw material types and quality grades, the weights adopt real numbers from 0 to 1, wherein 1 represents the highest weight, and 0 represents the lowest weight;
Specifically, for class a raw materials, the softness characteristics are important in all quality classes, then the original weight is set to 0.8, and the class a raw material class weight is set to 1;
For the B-class and C-class raw materials, different raw material class weights are set according to the importance of the raw materials, and for different quality grades, different quality grade weights are set.
Processing missing values
In processing missing values, a suitable filling strategy is first selected according to the data distribution and the distribution of missing values. The specific method comprises the following steps:
for numerical features, the filling may be chosen using mean, median or mode, the choice being dependent on the distribution of missing values and the extent of influence on the analysis result.
If the missing data has a large influence on the analysis result, it may be considered to delete the feature containing the missing value to ensure the accuracy of the data.
Handling outliers
In processing outliers, a Z-Score anomaly detection algorithm may be employed to identify and process outliers. The method comprises the following specific steps:
the deviation of each data point from the mean of the dataset was measured and its Z-Score was calculated.
A threshold is set and data points exceeding the threshold are considered outliers.
S3, training a model by adopting a decision tree model, and specifically comprising the following steps of:
Root node: initially, placing all collected data related to the quality of the textile into a root node;
node splitting: selecting, at each node, a feature and a corresponding threshold to divide the data of the current node into two subsets, the selection of splits being based on the measure of unrepeace, the goal being to select features and thresholds that minimize unrepeace;
creating a child node: each split creates two child nodes, and the feature selection and splitting of the next round is continued, which forms a branch structure of the tree until a stop condition is reached;
And (3) recursion: recursively splitting the child nodes until a stop condition is reached, which may include the depth of the tree reaching a predetermined value, the number of samples contained by the node being less than a threshold;
leaf node marking: when the stop condition is met, the samples in the leaf nodes are marked as a specific class, namely the quality class of the textile, wherein 1 stands for class i, 2 stands for class ii, and 3 stands for class iii.
The characteristic threshold specifically includes:
Fiber length
Threshold value: the threshold is set at 100 mm, when a sample with a fiber length less than 100 mm will enter one subnode, and a sample with a length greater than or equal to 100 mm will enter the other subnode;
Fiber strength
Threshold value: the threshold is set to 50 newtons, when a sample with a fiber strength less than 50 newtons will enter one child node, and a sample with a fiber strength greater than or equal to 50 newtons will enter the other child node;
Porosity of fabric
Threshold value: the threshold is set to 0.2 when a sample with a porosity of the fabric less than 0.2 will enter one child node and a sample greater than or equal to 0.2 will enter the other child node.
Wherein the tensile strength test method comprises:
preparing a sample: cutting out a standard-sized sample from the raw material;
clamping a sample: clamping the sample in a clamp of a tensile testing machine;
applying a stretching force: gradually applying a tensile force, and recording data of load and sample elongation;
calculation results: calculating tensile strength and elongation indexes according to the load and elongation data;
The fiber thickness testing method comprises the following steps:
preparing a sample: obtaining a fiber sample from the raw material and preparing an appropriate slice or cross section;
and (3) observing the fiber: measuring the diameter of the fiber using a digital image analysis system;
calculation results: calculating an average thickness index of the fiber according to the measured data;
the specific method for testing the pilling performance density comprises the following steps:
Cutting out a sample with standard size from a textile to be tested, and clamping two ends of the sample;
Determining illumination conditions, including brightness and illumination time of the light source;
a light source is arranged to uniformly irradiate the upper side surface of the sample;
A whiteboard for projection is arranged on the lower surface of the sample;
Shooting a shadow image of the sample projected on the whiteboard by using a camera or an image acquisition device;
Importing the shot image into a computer, and using image analysis software to analyze the quantity and degree of pilling phenomena and pilling areas in the image, and simultaneously analyzing the density distribution of textiles;
the following are the specific steps of the analysis:
Shadow difference analysis: analyzing the shadow difference according to the shadow range and depth of the pixels acquired by the image, wherein the pixels exceeding a preset threshold value represent that the shadow difference exists, and indicate that the density is uneven;
specifically, before shadow difference analysis, firstly preprocessing an image, including removing noise in the image, enhancing contrast and adjusting brightness;
converting the color image into a gray image, wherein the gray image only contains brightness information and is convenient for shadow difference analysis;
Selecting an appropriate threshold, dividing the image into shadow and non-shadow areas, marking areas with pixel values higher than the threshold as shadows, and marking areas lower than the threshold as non-shadows;
(2) Detecting a pilling position: detecting pilling positions in the image by using an image analysis algorithm, wherein the method is realized by searching high-contrast areas in shadows, the shadows of the pilling positions are darker and are in a dot shape, and the pilling phenomenon is evaluated according to the number and the degree of the pilling positions;
Specifically, a Canny edge detection algorithm is used to detect edges in the image. The Canny algorithm comprises the following steps:
And calculating the gradient and the direction of each pixel point in the image.
In the gradient image, only the pixel with the largest local gradient is reserved, and the other pixels are set to zero.
Two thresholds are defined, a higher threshold and a lower threshold. Pixels are classified into three types, strong edges, weak edges and non-edges according to gradient values, wherein strong edges are preserved and weak edges are connected in a subsequent step.
After Canny edge detection, connected edge pixels are obtained, which constitute the contour of the object.
The number of detected contours, each representing a pilling point, is counted.
The area of each profile was measured and the size of the pilling point was assessed. A larger area indicates a more serious pilling problem.
Shape characteristics of the profile, such as roundness, are analyzed to evaluate the degree of pilling. A less regular profile may indicate a more severe pilling problem.
Starting with a strong edge pixel, a complete edge contour is constructed by connecting adjacent weak edge pixels.
It is noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
The foregoing is merely a preferred embodiment of the present invention, and it should be noted that it will be apparent to those skilled in the art that several modifications and variations can be made without departing from the technical principle of the present invention, and these modifications and variations should also be regarded as the scope of the invention.
Claims (3)
1. The textile production quality prediction method based on data identification is characterized by comprising the following steps of:
S1: collecting data related to textile quality, comprising: textile materials, production process parameters and textile machinery sensor data;
Wherein the textile material: including the kind and quality detection grade information of the raw materials;
production process parameters: recording temperature, humidity and dyeing process data in the spinning process;
Textile machine sensor data: comprises the steps of acquiring tension, pressure and speed sensor data from textile machinery equipment;
s2: the data acquired in the step S1 are cleaned and preprocessed, and the method comprises the following steps:
Processing the missing values: filling and deleting missing data;
processing outliers: identifying and processing outlier data points;
Characteristic engineering: according to the characteristics of the textile, analyzing the textile data by adopting K-means clustering analysis to create new characteristic data, dividing the types of raw materials into A, B, C types under the conventional state, and dividing the quality detection class of each raw material into I, II and III types;
s3: establishing a prediction model for textile production quality: the method comprises the steps of adopting a decision tree model to carry out model training on new feature data created in the step S2, carrying out model tuning by using verification data, importing raw material feature data, and obtaining a textile quality prediction result;
S4: aiming at raw materials which are not in A, B, C, I, II and III in S2, an abnormality detection algorithm is adopted to identify and process abnormal raw materials;
The method comprises the following steps:
In the step S2, the data is cleaned and preprocessed, the data quality is ensured, and the characteristic data of the raw materials which are not in the step S2 are integrated with the previous processing result to perform abnormality detection;
Training raw material data by using an anomaly detection algorithm, establishing a model to identify an anomaly raw material, and adjusting model parameters by using a cross-validation technology at the stage;
Predicting unknown raw material data by using a trained anomaly detection model, wherein the model is used for assigning an anomaly value to each raw material data point, and determining which raw materials are regarded as anomalies by setting a threshold value;
In step S1, quality detection grade information comprises detection of physical properties of textile raw materials;
wherein, the physical property information detection comprises a tensile strength test, a fiber thickness test and a pilling performance density test;
the specific method for testing the pilling performance density comprises the following steps:
Cutting out a sample with standard size from a textile to be tested, and clamping two ends of the sample;
Determining illumination conditions, including brightness and illumination time of the light source;
a light source is arranged to uniformly irradiate the upper side surface of the sample;
A whiteboard for projection is arranged on the lower surface of the sample;
Shooting a shadow image of the sample projected on the whiteboard by using a camera or an image acquisition device;
Importing the shot image into a computer, and using image analysis software to analyze the quantity and degree of pilling phenomena and pilling areas in the image, and simultaneously analyzing the density distribution of textiles;
the following are the specific steps of the analysis:
(1) Shadow difference analysis: analyzing the shadow difference according to the shadow range and depth of the pixels acquired by the image, wherein the pixels exceeding a preset threshold value represent that the shadow difference exists, and indicate that the density is uneven;
(2) Detecting a pilling position: detecting pilling positions in the image by using an image analysis algorithm, searching for high-contrast areas in shadows, and evaluating pilling according to the number and degree of the pilling positions;
wherein the types of the characteristic engineering combined raw materials in the step S2 are classified into A, B, C types, wherein the A type is cotton, the B type is hemp and the C type is wool:
Selecting characteristics, namely selecting characteristics related to the characteristics of the raw materials according to the physical property data of each type of raw materials;
For the A-class cotton, selecting characteristics related to softness and water absorbability in a characteristic engineering, and specifically selecting fiber length and fiber bending characteristics;
for the type B hemp, selecting characteristics related to wear resistance, air permeability and antibacterial property by a characteristic engineering, and specifically selecting fiber strength, fiber thickness and fabric porosity;
for the C-type wool, the characteristic engineering selects characteristics related to warmth retention and elasticity, such as fiber diameter and fiber elastic modulus;
according to the characteristics of the textile, adopting K-means clustering analysis to analyze the textile data and creating new characteristic data, wherein the method specifically comprises the following steps:
a: preparing a data set containing raw material types and quality detection grades;
b: encoding the types and quality detection grades of the raw materials, and converting the classification variables into a numerical form by using a single-heat encoding method;
c: data normalization is performed to ensure that the values of the different features lie within the same range;
d: clustering the data by using K-means clustering analysis, dividing the data points into K clusters, wherein K is the selected cluster number, and setting K as the category number, namely, the raw material comprises A, B, C categories one by one, and the quality grades I, II and III category one by 9 clusters in total;
e: carrying out cluster allocation on each data point, determining which cluster the data point belongs to, and representing the raw material type and quality detection grade cluster to which each data point belongs;
f: creating new features to represent the types and quality detection levels of raw materials based on cluster allocation, including creating two new binary features to represent the types and quality detection levels of raw materials respectively, wherein the value of each feature is 0 or 1 to represent that the raw materials belong to a certain category or not;
g: the new characteristic data are used for subsequent data analysis and modeling, and the influence of the raw material types and quality detection grades on the textile is determined;
according to the importance of different raw material types and quality grades, the weight of the characteristics is adjusted;
wherein, raw material types A, B, C and three quality grades I, II, III, the importance of each characteristic is evaluated, and the importance is divided into three grades of high, medium and low, and the formula is as follows:
Feature weight = original weight raw material class weight quality grade weight
Wherein the raw material type weight and the quality grade weight are set according to the importance of different raw material types and quality grades, the weights adopt real numbers from 0 to 1, wherein 1 represents the highest weight, and 0 represents the lowest weight;
Specifically, for class a raw materials, the softness characteristics are important in all quality classes, then the original weight is set to 0.8, and the class a raw material class weight is set to 1;
For the B-class and C-class raw materials, respectively setting different raw material class weights according to the importance of the B-class and C-class raw materials, and setting different quality class weights for different quality classes;
S2, processing missing values, and filling data or deleting characteristics containing the missing values by adopting a mean value, a median value or a mode value according to data distribution and distribution of the missing values;
Processing outliers and detecting outliers using a Z-Score anomaly detection algorithm, comprising: measuring deviations of data points from the average of the data sets, calculating a Z-Score for each data point, setting a threshold, and treating data points exceeding the threshold as anomalies;
S3, training a model by adopting a decision tree model, and specifically comprising the following steps of:
Root node: initially, placing all collected data related to the quality of the textile into a root node;
node splitting: selecting, at each node, a feature and a corresponding threshold to divide the data of the current node into two subsets, the selection of splits being based on the measure of unrepeace, the goal being to select features and thresholds that minimize unrepeace;
creating a child node: each split creates two child nodes, and the feature selection and splitting of the next round is continued, which forms a branch structure of the tree until a stop condition is reached;
and (3) recursion: recursively splitting the child nodes until a stopping condition is reached, wherein the stopping condition comprises that the depth of the tree reaches a preset value and the number of samples contained in the node is less than a preset threshold value;
leaf node marking: when the stop condition is met, the samples in the leaf nodes are marked as a specific class, namely the quality class of the textile, wherein 1 stands for class i, 2 stands for class ii, and 3 stands for class iii.
2. A method for predicting the quality of a textile product based on data recognition according to claim 1, characterized in that the characteristic threshold value comprises in particular:
Fiber length
Threshold value: the threshold is set at 100 mm, when a sample with a fiber length less than 100 mm will enter one subnode, and a sample with a length greater than or equal to 100 mm will enter the other subnode;
Fiber strength
Threshold value: the threshold is set to 50 newtons, when a sample with a fiber strength less than 50 newtons will enter one child node, and a sample with a fiber strength greater than or equal to 50 newtons will enter the other child node;
Porosity of fabric
Threshold value: the threshold is set to 0.2 when a sample with a porosity of the fabric less than 0.2 will enter one child node and a sample greater than or equal to 0.2 will enter the other child node.
3. The method for predicting the quality of textile production based on data identification of claim 2, wherein the method for testing the tensile strength comprises:
preparing a sample: cutting out a standard-sized sample from the raw material;
clamping a sample: clamping the sample in a clamp of a tensile testing machine;
applying a stretching force: gradually applying a tensile force, and recording data of load and sample elongation;
calculation results: calculating tensile strength and elongation indexes according to the load and elongation data;
The fiber thickness testing method comprises the following steps:
preparing a sample: obtaining a fiber sample from the raw material and preparing an appropriate slice or cross section;
and (3) observing the fiber: measuring the diameter of the fiber using a digital image analysis system;
calculation results: and calculating an average thickness index of the fiber according to the measured data.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311574158.5A CN117273554B (en) | 2023-11-23 | 2023-11-23 | Textile production quality prediction method based on data identification |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311574158.5A CN117273554B (en) | 2023-11-23 | 2023-11-23 | Textile production quality prediction method based on data identification |
Publications (2)
Publication Number | Publication Date |
---|---|
CN117273554A CN117273554A (en) | 2023-12-22 |
CN117273554B true CN117273554B (en) | 2024-04-19 |
Family
ID=89204931
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202311574158.5A Active CN117273554B (en) | 2023-11-23 | 2023-11-23 | Textile production quality prediction method based on data identification |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117273554B (en) |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101419706A (en) * | 2008-12-11 | 2009-04-29 | 天津工业大学 | Jersey wear flokkit and balling up grading method based on image analysis |
CN105699231A (en) * | 2016-03-10 | 2016-06-22 | 上海工程技术大学 | Fabric fuzzing and pilling analysis device and method |
CN108764597A (en) * | 2018-04-02 | 2018-11-06 | 华南理工大学 | A kind of product quality control method based on integrated study |
CN109101632A (en) * | 2018-08-15 | 2018-12-28 | 中国人民解放军海军航空大学 | Product quality abnormal data retrospective analysis method based on manufacture big data |
CN113408963A (en) * | 2021-07-28 | 2021-09-17 | 上海致景信息科技有限公司 | Textile yarn quality rating method and device, storage medium and processor |
CN114372621A (en) * | 2021-12-28 | 2022-04-19 | 华南理工大学 | Melt-blown fabric quality online prediction and tracing method |
CN116341758A (en) * | 2023-04-23 | 2023-06-27 | 浙江理工大学 | Fabric quality prediction method based on KNN improved PSO-BP algorithm |
CN116467653A (en) * | 2023-04-03 | 2023-07-21 | 浙江理工大学 | Loom abnormal data processing method based on probability distribution and XGBoost decision algorithm |
CN116468160A (en) * | 2023-04-04 | 2023-07-21 | 大连理工大学 | Aluminum alloy die casting quality prediction method based on production big data |
-
2023
- 2023-11-23 CN CN202311574158.5A patent/CN117273554B/en active Active
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101419706A (en) * | 2008-12-11 | 2009-04-29 | 天津工业大学 | Jersey wear flokkit and balling up grading method based on image analysis |
CN105699231A (en) * | 2016-03-10 | 2016-06-22 | 上海工程技术大学 | Fabric fuzzing and pilling analysis device and method |
CN108764597A (en) * | 2018-04-02 | 2018-11-06 | 华南理工大学 | A kind of product quality control method based on integrated study |
CN109101632A (en) * | 2018-08-15 | 2018-12-28 | 中国人民解放军海军航空大学 | Product quality abnormal data retrospective analysis method based on manufacture big data |
CN113408963A (en) * | 2021-07-28 | 2021-09-17 | 上海致景信息科技有限公司 | Textile yarn quality rating method and device, storage medium and processor |
CN114372621A (en) * | 2021-12-28 | 2022-04-19 | 华南理工大学 | Melt-blown fabric quality online prediction and tracing method |
CN116467653A (en) * | 2023-04-03 | 2023-07-21 | 浙江理工大学 | Loom abnormal data processing method based on probability distribution and XGBoost decision algorithm |
CN116468160A (en) * | 2023-04-04 | 2023-07-21 | 大连理工大学 | Aluminum alloy die casting quality prediction method based on production big data |
CN116341758A (en) * | 2023-04-23 | 2023-06-27 | 浙江理工大学 | Fabric quality prediction method based on KNN improved PSO-BP algorithm |
Non-Patent Citations (1)
Title |
---|
基于KPCA-LSSVM的两阶段纺纱质量的预测模型;雷英娜;何牧泽;;价值工程(第31期);第157-160页 * |
Also Published As
Publication number | Publication date |
---|---|
CN117273554A (en) | 2023-12-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN114549522B (en) | Textile quality detection method based on target detection | |
CN114445387A (en) | Fiberboard quality classification method based on machine vision | |
KR20130118822A (en) | Classifier readiness and maintenance in automatic defect classification | |
CN108444934B (en) | Automatic segmentation and quantification method for debonding defect of composite material | |
CN114723704A (en) | Textile quality evaluation method based on image processing | |
CN114878574A (en) | Cloth defect detection method, device and system based on machine vision | |
CN115576284A (en) | Clothing workshop intelligent management method and system | |
CN109297534B (en) | Environmental parameter weight determination method and system for evaluating indoor environmental quality | |
CN111160451A (en) | Flexible material detection method and storage medium thereof | |
JP2021177380A (en) | Device and method for detecting defect in spinning mills and estimating one or more causes of defect | |
Pereira et al. | Computer vision techniques for detecting yarn defects | |
CN115719344A (en) | Fabric defect analysis and identification method for textile fabric quality inspection | |
CN111275705A (en) | Intelligent cloth inspecting method and device, electronic equipment and storage medium | |
CN114998321A (en) | Textile material surface hairiness degree identification method based on optical means | |
CN110458809B (en) | Yarn evenness detection method based on sub-pixel edge detection | |
CN117218114B (en) | Mesh fabric defect rapid detection method based on image data processing | |
CN116008289A (en) | Nonwoven product surface defect detection method and system | |
CN117273554B (en) | Textile production quality prediction method based on data identification | |
CN111636123B (en) | Intelligent spinning production line fault detection system | |
Li et al. | A direct measurement method of yarn evenness based on machine vision | |
CN116579984A (en) | Intelligent cloth inspection method and system for autonomous fault learning | |
Jmali et al. | Fabrics defects detecting using image processing and neural networks | |
CN115311278A (en) | Yarn cutting method for yarn detection | |
JP6872829B1 (en) | Foreign matter generation cause estimation system and foreign matter generation cause estimation method | |
CN117171670B (en) | Textile production process fault monitoring method, device and system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |