A non-contact method of detecting a rotating braided yarn

Abstract

The application discloses a kind of non-contact rotary braided yarn detection methods, comprising the following steps: (1) establish reserve knowledge base, complete yarn type identification and extract the original image of carrier on rotary braiding machine;(2) the original image extracted above is preprocessed and the image after pre-processing is collected;(3) the defect of above-mentioned yarn is defined, including yarn fiber breakage, yarn fiber knotting and yarn fiber crossing;(4) according to the image after pre-processing, the feature recognition of the defect of yarn is carried out;(5) yarn detection is carried out using binocular camera, and the weaving state of yarn is analyzed according to the definition of above-mentioned yarn defect;(6) the data of yarn weaving state is extracted;(7) if the number of yarn fiber breakage, yarn fiber knotting and crossing is greater than the preset value, the rotary braiding machine stops working.The application detects the rotary braided yarn by image detection technology, effectively improves the detection efficiency, reduces the machine error, and improves the product quality.

Description

Technical Field

[0001] This invention relates to the field of rotary three-dimensional knitting technology, and more particularly to the detection of rotary three-dimensional knitted yarns, specifically a non-contact rotary knitted yarn detection method. Background Technology

[0002] Currently, most textile processes use contact-type yarn detectors. During the detection process, the sensor element is in direct contact with the yarn. The instrument surface is in prolonged contact with the moving yarn, which not only causes wear and shortens the instrument's lifespan but also affects the yarn's running condition, failing to accurately reflect yarn quality and its changes. Non-contact yarn detection methods can significantly reduce the instrument's influence on the yarn's transport status, thereby improving the accuracy of yarn detection. Summary of the Invention

[0003] Purpose of the invention: To address the shortcomings of existing technologies, this invention provides a non-contact rotary braided yarn detection method, which solves the problems of yarn wear and inaccurate detection in contact rotary braided yarn detection, thereby significantly improving the accuracy of rotary braided yarn detection.

[0004] Technical Solution: To solve the above problems, the non-contact rotary knitting yarn detection method provided by this invention adopts the following steps:

[0005] A non-contact rotary knitting yarn detection method, characterized by comprising the following steps:

[0006] S1 establishes a knowledge base, completes yarn type identification, and extracts the original image of the yarn carrier on the rotary braiding machine: the yarn type image is processed and imported into the knowledge base, and the system automatically identifies the yarn type; the camera captures the original image of the yarn carrier's movement in real time;

[0007] S2 preprocesses the extracted original image and collects the preprocessed image, including performing image digitization processing on the original image through an A / D converter to obtain a digital image, and preprocessing the digital image to remove noise and enhance image features;

[0008] S3 defines yarn defects, including yarn fiber breakage, yarn fiber knotting, and yarn fiber crossing.

[0009] S4 performs feature recognition on yarn defects in the preprocessed digital image: the preprocessed digital image is segmented using a threshold segmentation method, features are extracted from the segmented regions and compared with the yarn defect features already input into the yarn detection system to identify the types of defects in the yarn on the yarn carrier.

[0010] S5 uses a camera to inspect yarn and analyzes the yarn weaving state based on the above definition of yarn defects: the camera is placed at the yarn guide opening of the yarn carrier, and the image captured by the camera is used to identify the types of defects in the yarn on the yarn carrier through the above defect feature recognition.

[0011] S6 Extract the data on the above yarn weaving state: Determine the yarn weaving state according to the above steps, and then obtain the degree of yarn fiber breakage, the number of yarn fiber knots and crosses through data post-processing;

[0012] S7 If the above-mentioned yarn fibers experience cross-sectional breakage, knotting, or a cross count exceeding the preset value, the rotary braiding machine will stop working.

[0013] Furthermore, the definition of the defects in the yarn in step S3 also includes: memorizing and feeding back the yarn defects based on a deep learning neural network.

[0014] Furthermore, in the threshold analysis in step S4, if more than half of the volume of an image pixel is air, the yarn defect is identified as yarn breakage; if the volume of an image pixel is continuously filled, it is identified as yarn knotting and crossing.

[0015] Furthermore, the data on the yarn weaving state extracted in step S6 also includes: calculating the degree of yarn fiber breakage, the number of yarn fibers knotted and crossed based on deep learning neural network analysis.

[0016] Furthermore, step S7, in which the rotary braiding machine stops working, also includes: real-time feedback of the yarn status of the rotary braiding machine, adjusting the motion parameters of the rotary braiding machine chassis and the yarn path planning through the yarn status data, realizing the self-detection feedback adjustment of the rotary braiding machine, and stopping the rotary braiding machine if it exceeds the autonomous adjustment range.

[0017] This invention provides a non-contact method for detecting rotary braided yarns. Compared with existing technologies, it has the following advantages:

[0018] (1) The non-contact rotary knitting yarn detection method constructs a yarn reserve knowledge base and defines yarn defect types, solving the problem of non-contact yarn detection that is complex to operate and has low detection efficiency.

[0019] (2) The non-contact rotary braiding yarn detection method uses a computer vision system and a deep learning-based neural network to perform data analysis and processing on the image, which can realize yarn detection of non-contact rotary braiding machine. Attached Figure Description

[0020] Figure 1 The flowchart shown is a non-contact rotary braided yarn detection method of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] As shown in the figure, an embodiment of the present invention provides a technical solution, including the following steps:

[0023] S1. Establish a reserve knowledge base, complete yarn type identification and extract the original image of the yarn carrier on the rotary braiding machine: The yarn type image is processed and imported into the reserve knowledge base. The system automatically identifies the yarn type to facilitate the next step. If the yarn cannot be identified, the reserve knowledge base is added or it is identified again. Then, the original image of the yarn carrier movement is captured in real time by the binocular camera.

[0024] S2. Preprocess the extracted original images and collect the preprocessed images: The original images are digitized by an A / D converter, and then the digitized images are preprocessed to remove noise and enhance image features.

[0025] S3. Define the defects of the above yarns, including yarn fiber breakage, yarn fiber knotting and yarn fiber crossing: Summarize and analyze the problems that occur in the yarns that are working normally on the rotary braiding machine, and then define the defects of the yarns based on the literature review and input them into the yarn detection system.

[0026] S4. Based on the preprocessed image, identify the defects of the yarn: Use the threshold segmentation method to extract features from the preprocessed segmented region and compare them with the yarn defect features already input into the yarn detection system for identification.

[0027] S5. Use binocular cameras to inspect yarn and analyze the yarn weaving state according to the above definition of yarn defects: Place four binocular cameras at the yarn guide opening of the yarn carrier, and use the above defect feature identification to determine the type of defect in the yarn on the yarn carrier.

[0028] S6. Extract the data on the yarn weaving state: Determine the yarn weaving state according to the above steps, and then obtain the degree of yarn fiber breakage, the number of yarn fiber knots and crosses through data post-processing.

[0029] S7. If more than 50% of the cross-section of the above-mentioned yarn fibers is broken, or the number of knots and crosses in the yarn fibers exceeds the preset value, the rotary braiding machine will stop working.

[0030] In this embodiment, the definition of yarn defects in step S3 further includes: memorizing and feeding back yarn defects using a deep learning-based neural network. The deep learning-based neural network is designed to address the characteristics of yarn fiber image data, constructing a corresponding yarn detection network to obtain a yarn defect detection model. The yarn detection network employs a feature pyramid network, which enhances the convolutional neural network through top-down paths and lateral connections. It constructs a multi-scale feature pyramid from the input image captured by the camera, with each layer of the pyramid used to detect yarn defects at different scales. The analyzed anomaly results are then fed back in real time. This technique of applying a feature pyramid network to a yarn detection network is existing technology and can be found in Chinese patent application CN 113592852 A, which will not be elaborated upon here.

[0031] The threshold analysis in S4 also includes a novel 50% threshold analysis method, which identifies yarn defects as yarn breakage if more than half of the volume of an image pixel is air, and yarn knots and crosses if the volume of an image pixel is continuously filled.

[0032] The data extracted from the yarn weaving state in S6 also includes: calculating the degree of yarn fiber breakage, the number of yarn fiber knots and crosses based on deep learning neural network analysis.

[0033] The S7 step of stopping the rotary braiding machine also includes: real-time feedback of the yarn status of the rotary braiding machine, adjusting the motion parameters of the rotary braiding machine chassis and the yarn path planning through the yarn status data, realizing the self-detection feedback adjustment of the rotary braiding machine, and stopping the rotary braiding machine if it exceeds the autonomous adjustment range.

Claims

1. A non-contact rotary knitting yarn detection method, characterized in that, Includes the following steps: S1 establishes a reserve knowledge base, completes yarn type identification, and extracts the original image of the yarn carrier on the rotary braiding machine: the yarn type image is processed and imported into the reserve knowledge base, and the system automatically identifies the yarn type; The camera captures raw images of the yarn carrier's movement in real time. S2 preprocesses the extracted original image and collects the preprocessed image, including performing image digitization processing on the original image through an A / D converter to obtain a digital image, and preprocessing the digital image to remove noise and enhance image features; S3 defines yarn defects, including yarn fiber breakage, yarn fiber knotting, and yarn fiber crossing. Yarn defects are stored in a deep learning neural network, which memorizes and feeds back the yarn defects. S4 performs feature recognition on yarn defects in the preprocessed digital image: The preprocessed digital image is segmented using a threshold segmentation method, and features are extracted from the segmented regions and compared with the yarn defect features input into the yarn detection system to determine the type of yarn defect on the yarn carrier; if more than half of the volume of the image pixel is air, the yarn defect is identified as yarn breakage; if the image pixel volume is continuously filled, the defect is identified as yarn knotting and crossing. S5 uses a camera to inspect yarn and analyzes the yarn weaving state based on the above definition of yarn defects: the camera is placed at the yarn guide opening of the yarn carrier, and the image captured by the camera is used to identify the types of defects in the yarn on the yarn carrier through the above defect feature recognition. S6 Extracts the data on the yarn weaving state: Based on the above steps, the yarn weaving state is determined, and then the degree of yarn fiber breakage, the number of yarn fiber knots and crosses are obtained through data post-processing. This includes the calculation of the degree of yarn fiber breakage, the number of yarn fiber knots and crosses based on deep learning neural network analysis; the weaving state includes whether the yarn weaving is normal and whether it contains the previously defined yarn defects. S7 If more than 50% of the cross-section of the above-mentioned yarn fibers is broken, or the number of knotted and crossed yarn fibers is greater than the preset value, the rotary braiding machine will stop working; The shutdown of the rotary braiding machine also includes: real-time feedback of the yarn status of the rotary braiding machine, adjusting the motion parameters of the rotary braiding machine chassis and the yarn path planning based on the yarn status data, realizing the self-detection feedback adjustment of the rotary braiding machine, and stopping the rotary braiding machine if it exceeds the autonomous adjustment range.

2. The non-contact rotary braided yarn detection method according to claim 1, characterized in that, In step S1, the original images of different types of yarn are input into the knowledge base. The system captures the features of these images and then classifies the yarns according to the features.

Images