Intelligent floating yarn detection system of computerized flat knitting machine

Abstract

The invention relates to the technical field of textile, in particular to an intelligent floating yarn detection system of a computerized flat knitting machine, which captures an image of a knitting yarn in real time through a camera recognition module, focuses a light beam at the position of a knitting needle on a needle bed through a specific wavelength laser, and irradiates a knitting area; the optical filter is matched with the specific wavelength laser to form an optical filtering system, external light source interference is reduced, and image quality is improved; the image processing module is used for processing the image captured by the camera identification module so as to identify floating yarns; the camera recognition module and the specific wavelength laser work cooperatively, the position of the knitting yarn can be accurately captured, early-stage efficient screening and early warning of the floating yarn are achieved, and the accuracy and timeliness of floating yarn recognition are improved.

Description

Technical Field

[0001] This invention relates to the field of textile technology, and in particular to an intelligent floating yarn detection system for computerized flat knitting machines. Background Art

[0002] As a core piece of equipment in the textile industry, the knitting flat knitting machine directly affects the quality of clothing, home textiles, and other products. With consumers' increasing demands for the refinement of textiles and the development of intelligent manufacturing, the industry's need for automated inspection of the production process is growing, especially for real-time monitoring technology of common defects such as loose yarn.

[0003] Currently, yarn float detection mainly employs mechanical probe-type devices, such as the YFJX series computerized flat knitting machine detection system described in the article "Application of Electronic Control Technology in Computerized Flat Knitting Machines." Its principle involves an adjustable-height probe contacting the yarn; when the yarn float height exceeds a set threshold, the probe triggers a microswitch to send a stop signal. These devices are simple in structure and low in cost, playing a crucial role in early applications.

[0004] However, existing technologies have significant shortcomings: First, mechanical probes require manual height preset, relying on operational experience and are prone to false alarms or missed detections due to improper settings; second, contact-based detection cannot adapt to high-speed weaving or complex patterns, especially when weaving multi-colored yarns or mixed materials, making it difficult to identify floating yarns through single physical contact; furthermore, traditional devices lack image analysis capabilities, failing to achieve early warning and precise location of floating yarns. These problems severely restrict detection efficiency and accuracy, urgently requiring a non-contact, intelligent solution that can dynamically adapt to different weaving conditions. Summary of the Invention

[0005] In view of this, the purpose of this invention is to propose an intelligent floating yarn detection system for computerized flat knitting machines, so as to solve the problem that the detection efficiency and accuracy are severely limited in floating yarn detection.

[0006] To achieve the above objectives, the present invention provides an intelligent yarn floating detection system for computerized flat knitting machines, comprising:

[0007] A camera recognition module is installed on the knitting machine head to capture images of the knitting yarn in real time through the gaps between the knitting machine guide rails.

[0008] A specific wavelength laser is used to focus a beam of light onto the position of the knitting needles on the needle bed to illuminate the knitting area;

[0009] The filter, in conjunction with the laser of the specific wavelength, forms an optical filtering system used to reduce interference from external light sources and improve image quality;

[0010] An image processing module is used to process the images captured by the camera recognition module to identify floating yarn;

[0011] The communication module is used to transmit the test results to the flat knitting machine control system.

[0012] Optionally, the camera recognition module includes:

[0013] U-shaped fixing bracket is used to fix the module on the flat knitting machine;

[0014] L-shaped support frame is used to support the camera recognition module;

[0015] A protective enclosure is provided to house and protect the camera recognition module and filter. An opening is provided at the bottom of the enclosure to allow the camera to capture images.

[0016] Optionally, the U-shaped fixing bracket and L-shaped support bracket can be adjusted by adjusting the installation angle to adjust the horizontal and pitch angles of the camera recognition module, thereby optimizing the imaging effect.

[0017] Optionally, the image processing module supports the following two detection modes:

[0018] Based on the emission intensity detection mode, the floating yarn area is marked by comparing the emission intensity difference between the yarn accumulation area and the non-accumulation area using an image recognition algorithm;

[0019] Based on the knitting template comparison mode, the floating yarn recognition in complex knitting scenarios is realized by comparing the pre-stored template file with the actual acquired image.

[0020] Optionally, the comparison mode based on the knitting template includes: generating a template file during the sampling stage. The template file includes key data in the knitting process, including the knitting head position, number of knitting rows, knitting nozzle, and knitting head speed.

[0021] Optionally, the comparison mode based on the weaving template includes: the image processing module 401 dynamically adjusts the threshold parameter for floating yarn detection using a CNN-LSTM model;

[0022] The input data for the CNN-LSTM model is a one-dimensional sequence including exposure time, sensitivity, camera head position, camera head speed, and reflection intensity.

[0023] During training, the input data is sorted according to the order of the weaving pattern to provide ordered input information for the CNN-LSTM model;

[0024] The output of the CNN-LSTM model includes sensitivity, exposure time, and recognition threshold.

[0025] Optionally, the CNN-LSTM model includes: performing grayscale and normalization processing on the input data before training; grayscale...

[0026] Grayscale processing is achieved by converting a color image into a single-channel grayscale image. The calculation formula is as follows:

[0027] I gray =0.299·R+0.587·G+0.114·B;

[0028] Where R, G, and B are the red, green, and blue channel values ​​of an image pixel, respectively, and I gray These are the pixel values ​​of a grayscale image;

[0029] Normalization is performed on the grayscale image, and the normalization formula is:

[0030]

[0031] Among them: I gray The pixel value after grayscale conversion; min(I gray ) and max(I gray ) are the minimum and maximum gray values ​​of the image, respectively; I norm The normalized pixel value ranges from [0, 1].

[0032] Optionally, the optimization criteria for the CNN-LSTM model include: calculating the squared difference between the recognition thresholds of different parameters for floating yarn images and non-floating yarn images under the same conditions, optimizing the model performance by minimizing this squared difference, thereby determining the most ideal threshold parameters for different yarn structures.

[0033] Optionally, the communication module adopts a CONN-TH_6P interface, supports 24V power supply and CAN communication protocol, and is used for quick access to the flat knitting machine operating system;

[0034] The camera recognition module includes a power conversion module, an image processing module, and a communication processing module. The communication processing module uses an STM32F103 and an SN65HVD chip to implement CAN communication. The power conversion module uses an MP1584 voltage conversion chip to perform voltage reduction and regulation.

[0035] Optionally, a flat knitting machine end operation window is included, which is used to realize data transmission, parameter adjustment, template data viewing and alarm prompt functions. The flat knitting machine end operation window is integrated into the Logica system and supports three working modes: detection mode, recording mode and comparison mode.

[0036] The beneficial effects of this invention are as follows: This invention provides an intelligent floating yarn detection system for computerized flat knitting machines. A camera recognition module captures images of the knitting yarn in real time. A specific wavelength laser focuses a beam onto the position of the knitting needles on the needle bed, illuminating the knitting area. A filter, in conjunction with the specific wavelength laser, forms an optical filtering system to reduce interference from external light sources and improve image quality. An image processing module processes the images captured by the camera recognition module to identify floating yarn. The camera recognition module and the specific wavelength laser work together to accurately capture the position of the knitting yarn, achieving early and efficient identification and warning of floating yarn, thus improving the accuracy and timeliness of floating yarn identification.

[0037] Meanwhile, the detection mode based on reflection intensity can utilize the difference in reflection intensity between the piled and unpiled areas of yarn under laser illumination to quickly mark the floating yarn area through image recognition algorithms, effectively realizing floating yarn detection. The knitting template comparison mode introduces a template storage mechanism to generate template files by combining key knitting data and image reflection features during complex knitting patterns, and then compares them, improving the efficiency and reliability of floating yarn recognition under complex patterns and solving the recognition difficulties caused by the diversity of yarn colors and materials and the complexity of patterns. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0039] Figure 1 This is a schematic diagram of a module of an intelligent floating yarn detection system for a computerized flat knitting machine according to an embodiment of the present invention;

[0040] Figure 2 This is a diagram illustrating the effect of the floating yarn identification marking area of ​​an intelligent floating yarn detection system for a computerized flat knitting machine according to an embodiment of the present invention.

[0041] Figure 3 This is a diagram illustrating the normal knitting area of ​​an intelligent floating yarn detection system for a computerized flat knitting machine according to an embodiment of the present invention.

[0042] Figure 4 This is a flowchart of a computer flat knitting machine intelligent floating yarn detection system based on a knitting template comparison mode, according to an embodiment of the present invention.

[0043] Figure 5 This is a schematic diagram of the operation window of a computerized flat knitting machine intelligent floating yarn detection system according to an embodiment of the present invention;

[0044] Figure 6This is a flowchart of the operation window of a computerized flat knitting machine intelligent floating yarn detection system according to an embodiment of the present invention.

[0045] Figure 7 This is a schematic diagram of the camera recognition module structure of an intelligent floating yarn detection system for a computerized flat knitting machine according to an embodiment of the present invention;

[0046] Figure 8 This is a schematic diagram of the communication module and power conversion module of an intelligent floating yarn detection system for a computerized flat knitting machine according to an embodiment of the present invention.

[0047] The diagram is marked as follows:

[0048] 1011, U-shaped fixing frame; 1012, L-shaped support frame; 1013, protective encapsulation box. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0050] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0051] This invention provides a specific embodiment of an intelligent floating yarn detection system for computerized flat knitting machines, such as... Figure 1 As shown, it includes:

[0052] The camera recognition module 101 is installed on the knitting machine head and is used to capture images of the knitting yarn in real time through the gap between the knitting machine guide rails.

[0053] A specific wavelength laser 201 is used to focus a beam of light onto the position of the knitting needles on the needle bed to illuminate the knitting area;

[0054] The filter 301, together with the laser 201 of the specific wavelength, forms an optical filtering system to reduce interference from external light sources and improve image quality.

[0055] Image processing module 401 is used to process the images captured by the camera recognition module 101 to identify floating yarn;

[0056] The communication module 501 is used to transmit the detection results to the flat knitting machine control system.

[0057] In some optional specific embodiments, such as Figure 6 As shown, the camera recognition module 101 includes:

[0058] U-shaped fixing bracket 1011 is used to fix the module on the flat knitting machine;

[0059] L-shaped support frame 1012 is used to support camera recognition module 101;

[0060] A protective enclosure 1013 is used to house and protect the camera recognition module 101 and the filter 301. An opening is provided at the bottom of the enclosure for the camera to capture images.

[0061] In some optional embodiments, the U-shaped fixing frame 1011 and the L-shaped support frame 1012 can adjust the horizontal and pitch angles of the camera recognition module 101 by adjusting the installation angle, so as to optimize the imaging effect.

[0062] In some optional embodiments, the image processing module 401 supports the following two detection modes:

[0063] Based on the emission intensity detection mode, the floating yarn area is marked by comparing the emission intensity difference between the yarn accumulation area and the non-accumulation area using an image recognition algorithm;

[0064] Based on the knitting template comparison mode, the floating yarn recognition in complex knitting scenarios is realized by comparing the pre-stored template file with the actual acquired image.

[0065] In some optional specific embodiments, the knitting template comparison mode includes: generating a template file during the sampling stage, the template file including key data in the knitting process, including the knitting head position, number of knitting rows, knitting nozzle, and knitting head speed.

[0066] In some optional specific embodiments, the comparison mode based on the weaving template includes: the image processing module 401 dynamically adjusts the threshold parameters for floating yarn detection using a CNN-LSTM model;

[0067] The input data for the CNN-LSTM model is a one-dimensional sequence including exposure time, sensitivity, camera head position, camera head speed, and reflection intensity.

[0068] During training, the input data is sorted according to the order of the weaving pattern to provide ordered input information for the CNN-LSTM model;

[0069] The output of the CNN-LSTM model includes sensitivity, exposure time, and recognition threshold.

[0070] In some optional specific embodiments, the CNN-LSTM model includes: performing grayscale and normalization processing on the input data before training; the grayscale processing is achieved by converting a color image into a single-channel grayscale image, and the calculation formula is as follows:

[0071] I gray =0.299·R+0.587·G+0.114·B;

[0072] Where R, G, and B are the red, green, and blue channel values ​​of an image pixel, respectively, and I gray These are the pixel values ​​of a grayscale image;

[0073] Normalization is performed on the grayscale image, and the normalization formula is:

[0074]

[0075] Among them: I gray The pixel value after grayscale conversion; min(I gray ) and max(I gray ) are the minimum and maximum gray values ​​of the image, respectively; I norm The normalized pixel value ranges from [0, 1].

[0076] In some optional specific embodiments, the optimization criteria of the CNN-LSTM model include: calculating the squared difference of the recognition thresholds of different parameters for floating yarn images and non-floating yarn images under the same conditions, optimizing the model performance by minimizing this squared difference, thereby determining the most ideal threshold parameters for different yarn structures.

[0077] In some optional embodiments, the communication module 501 adopts a CONN-TH_6P interface, supports 24V power supply and CAN communication protocol, and is used for quick access to the flat knitting machine operating system;

[0078] The camera recognition module includes a power conversion module, an image processing module 401, and a communication processing module. The communication processing module uses an STM32F103 and an SN65HVD chip to implement CAN communication. The power conversion module uses an MP1584 voltage conversion chip to perform voltage reduction and regulation.

[0079] In some optional specific embodiments, a flat knitting machine end operation window is included, which is used to realize data transmission, parameter adjustment, template data viewing and alarm prompt functions. The flat knitting machine end operation window is integrated into the Logica system and supports three working modes: detection mode, recording mode and comparison mode.

[0080] Before actual production on a flat knitting machine, a garment sampling stage is usually required to ensure that the produced garments meet the requirements. This invention, combining the concept of pre-production garment sampling, proposes two floating yarn detection modes: a detection mode based on reflection intensity and a knitting template comparison mode.

[0081] Reflection intensity-based detection mode: Applicable to knitting scenarios with relatively simple patterns. When a needle breaks, the yarn cannot form a loop within the needle, leading to yarn accumulation. Under laser illumination, there is a difference in reflection intensity between the accumulated and non-accumulated areas. Using image recognition algorithms, the floating yarn areas can be marked and the flat knitting machine control system can be notified. The recognition effect is as follows: Figure 2 and Figure 3 As shown.

[0082] Based on knitting template comparison mode: Applicable when yarns have diverse colors and materials and complex patterns, especially when multiple colors of yarn are used in the same knitting row, including:

[0083] A template storage mechanism is introduced in the sampling stage. Specifically, during template storage, key data generated during the weaving process, such as the current machine head position, number of weaving rows, weaving yarn feeder, and machine head weaving speed, are transmitted in real time to the image processing module 401. The image processing module 401 combines the key data with the reflection features of the image to extract features and generate a recognition template file. The reflection features of the image include brightness and its range.

[0084] The testing flowchart is as follows Figure 4 As shown, in actual production, a pre-generated template file is loaded and compared with images collected during the production process. Through feature comparison algorithms, floating yarn identification under different weaving structures can be achieved quickly and accurately, thereby improving the detection efficiency and reliability of complex weaving patterns.

[0085] When using the template comparison method for floating yarn detection, a CNN-LSTM model is employed for threshold feature learning. Its core objective is to set optimal threshold parameters for various yarn structures. The model's input is a one-dimensional sequence of feature parameters corresponding to the image acquisition, covering elements such as exposure time, sensitivity, machine head position, machine head speed, and reflection intensity. During training, the input data is arranged according to the weaving pattern sequence, providing ordered input information to the model. The model's output includes sensitivity, exposure time, and recognition threshold. Before converting the image sequence into a parameter sequence as input, the image needs to be normalized and converted to grayscale to reduce feature perturbations caused by external interference. Grayscale conversion can be achieved by converting a color image into a single-channel grayscale image; the specific calculation formula is as follows:

[0086] Igray = 0.299·R + 0.587·G + 0.114·B

[0087] Where R, G, and B are the red, green, and blue channel values ​​of the image pixels, respectively, and I... gray Here are the pixel values ​​of the grayscale image. Next, to avoid gradient instability caused by an excessively large range of feature values, the grayscale image is normalized. The normalization formula is as follows:

[0088]

[0089] Among them: I gray The pixel value after grayscale conversion; min(I gray ) and max(I gray ) are the minimum and maximum gray values ​​of the image, respectively; I norm The normalized pixel value ranges from [0, 1].

[0090] When optimizing the model, the standard followed is to calculate the squared difference of the recognition thresholds of different parameters for floating yarn images and non-floating yarn images under the same conditions. By minimizing this squared difference, the model performance is optimized, thereby finding the most ideal threshold parameters for different yarn structures.

[0091] Using the above model for parameter adjustment not only improves work efficiency but also ensures the accuracy and generalization ability of the parameters under different yarns and weaving scenarios.

[0092] With the increasing popularity of digital flat knitting machines, the diversity of recognition parameters is ensured. By combining pre-trained models with detection data during the sampling process, dynamic threshold production models applicable to actual production can be quickly generated, providing more efficient and accurate support for floating yarn detection in the production process.

[0093] In some optional embodiments, a computerized flat knitting machine intelligent yarn floating detection system includes a yarn floating operation window at the flat knitting machine end, the operation interface as shown in the figure. Figure 5 As shown. The main functions include: image acquisition via the detection module; viewing the template matching database; adjusting detection module parameters; and viewing detection module logs.

[0094] The flat knitting machine requires an operation window for functions such as data transmission, debugging transmission module parameters, and viewing template data. This operation window is integrated into the Logica system and developed using the Delphi language.

[0095] The workflow of the flat knitting machine end yarn floating operation window is as follows: Figure 6 As shown,

[0096] After initialization and software startup, the system first performs a self-check of the hardware status to ensure that the device is working properly.

[0097] The software checks whether a heartbeat signal has been received. If no heartbeat signal is received, the software performs a connection check. If a heartbeat signal is received, the software proceeds to the next step.

[0098] Confirm the knitting information, specifically the current knitting information of the flat knitting machine, and ensure that the machine head running speed and knitting file can be correctly obtained.

[0099] Obtain an image of the woven area.

[0100] The user selects a working mode. The software offers three main modes.

[0101] Detect Mode: Uses a detection mode based on reflection intensity;

[0102] Record Mode: Uses a comparison mode based on the weaving template, which is mainly used to generate inspection templates in the case of sampling;

[0103] Compare Mode: Based on the knitting template comparison mode, this mode is mainly used in actual production situations to realize the detection of floating yarn through template comparison algorithm.

[0104] Adjusting the parameters of the camera recognition module 101: In detection mode and recording mode, the parameters of the camera recognition module 101 need to be adjusted according to the acquired imaging conditions to obtain the best comparison results.

[0105] Select a template file from the database. In comparison mode, you need to select and load the template file from the database.

[0106] Start the inspection. If any abnormality is found, the floating yarn operation window on the flat knitting machine will trigger an alarm.

[0107] In some optional specific implementations, such as Figure 7 As shown, the camera recognition module 101 includes a U-shaped fixing frame 1011 for quick installation on the upper part of the flat knitting machine. The rear of the U-shaped fixing frame 1011 is provided with countersunk screw holes to ensure the stability of the device during use and to facilitate quick installation and disassembly. Adjustable L-shaped support frames 1012 are installed at both ends of the U-shaped fixing frame 1011. A protective encapsulation box 1013 is installed between the two sets of L-shaped support frames 1012. The camera module and filter 301 are installed inside the protective encapsulation box 1013. An opening is provided at the bottom of the protective encapsulation box 1013 for the camera of the camera module to capture images of the knitting area in the gap of the flat knitting machine guide rail.

[0108] During use, the protective encapsulation box 1013 allows for horizontal and tilt angle imaging adjustments, and the U-shaped mounting bracket 1011 allows for quick installation and removal.

[0109] In some optional specific embodiments, such as Figure 8 As shown, this system uses the CONN-TH_6P interface commonly used in flat knitting machines to quickly provide 24V power and achieve CAN communication connection.

[0110] The camera recognition module 101 includes a power conversion module, an image processing module 401, and a communication processing module. The communication module uses an STM32F103 and an SN65HVD chip to implement CAN communication; the power conversion module uses an MP1584 voltage converter chip to perform voltage reduction and regulation.

[0111] The working principle of this invention is as follows: It adopts a technical architecture in which a camera recognition module 101 and a specific wavelength laser 201 work together. The camera recognition module 101 is installed at the center of the knitting machine head and uses the gap between the guide rails as an imaging channel to capture the position of the knitting yarn in real time. The beam emitted by the specific wavelength laser 201 is focused on the position of the knitting needle on the needle bed and forms an optical filtering system with the filter 301 to reduce interference from external light sources and improve image quality, so as to realize the early identification and warning of floating yarn during the knitting process of the knitting machine. The optical filtering system formed by the specific wavelength laser 201 and the filter 301 can effectively reduce interference from external light sources and improve image quality, providing a good image basis for accurate identification of floating yarn and reducing the negative impact of irrelevant light on the identification results.

[0112] Combining the pre-production sampling approach for garments, a detection mode based on reflectivity is proposed. This mode utilizes the difference in reflectivity between yarn-accumulated and non-accumulated areas under laser irradiation and image recognition algorithms to mark floating yarn areas. A knitting template comparison mode is also proposed. By introducing a template storage mechanism in the sampling process, a template file is generated by combining key knitting data and image reflectivity features. During production, the template file is compared with the acquired images to achieve rapid and accurate identification of floating yarn under complex patterns.

[0113] For simpler knitting patterns, the reflection intensity-based detection mode utilizes the difference in reflection intensity between piled and unpiled yarn areas under laser illumination. This allows for rapid marking of floating yarn areas using image recognition algorithms, effectively achieving floating yarn detection. The knitting template comparison mode introduces a template storage mechanism. When knitting complex patterns, it combines key knitting data with image reflection features to generate template files for comparison. This improves the efficiency and reliability of floating yarn identification in complex patterns, solving the identification difficulties caused by diverse yarn colors, materials, and complex patterns.

[0114] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of the invention (including the claims) is limited to these examples; within the framework of the invention, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of the different aspects of the invention as described above, which are not provided in the details for the sake of brevity.

[0115] This invention is intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A computerized flat knitting machine intelligent floating yarn detection system, characterized in that, include: A camera recognition module (101) is installed on the knitting machine head to capture images of the knitting yarn in real time through the gap between the knitting machine guide rails; A specific wavelength laser (201) is used to focus a beam onto the position of the knitting needles on the needle bed to illuminate the knitting area; The filter (301), together with the laser of the specific wavelength (201), forms an optical filtering system for reducing interference from external light sources and improving image quality; An image processing module (401) is used to process the images captured by the camera recognition module (101) to identify floating yarn; The communication module (501) is used to transmit the detection results to the flat knitting machine control system.

2. The intelligent floating yarn detection system for a computerized flat knitting machine according to claim 1, characterized in that, The camera recognition module (101) includes: U-shaped fixing bracket (1011) is used to fix the module on the flat knitting machine; L-shaped support frame (1012) is used to support the camera recognition module (101); A protective enclosure (1013) is provided to house and protect the camera recognition module (101) and the filter (301), and an opening is provided at the bottom of the enclosure for the camera to capture images.

3. The intelligent floating yarn detection system for a computerized flat knitting machine according to claim 2, characterized in that, The U-shaped fixing frame (1011) and L-shaped support frame (1012) can adjust the horizontal and pitch angles of the camera recognition module (101) by adjusting the installation angle, so as to optimize the imaging effect.

4. The intelligent floating yarn detection system for a computerized flat knitting machine according to claim 1, characterized in that, The image processing module (401) supports the following two detection modes: Based on the emission intensity detection mode, the floating yarn area is marked by comparing the emission intensity difference between the yarn accumulation area and the non-accumulation area using an image recognition algorithm; Based on the knitting template comparison mode, the floating yarn recognition in complex knitting scenarios is realized by comparing the pre-stored template file with the actual acquired image.

5. The intelligent floating yarn detection system for a computerized flat knitting machine according to claim 4, characterized in that, The comparison mode based on the knitting template includes: generating a template file during the sampling stage. The template file includes key data in the knitting process, including the knitting head position, number of knitting rows, knitting nozzle, and knitting head speed.

6. The intelligent floating yarn detection system for a computerized flat knitting machine according to claim 4, characterized in that, The comparison mode based on the weaving template includes: the image processing module (401) uses a CNN-LSTM model to dynamically adjust the threshold parameters for floating yarn detection; The input data for the CNN-LSTM model is a one-dimensional sequence including exposure time, sensitivity, camera head position, camera head speed, and reflection intensity. During training, the input data is sorted according to the order of the weaving pattern to provide ordered input information for the CNN-LSTM model; The output of the CNN-LSTM model includes sensitivity, exposure time, and recognition threshold.

7. The intelligent floating yarn detection system for a computerized flat knitting machine according to claim 6, characterized in that, The CNN-LSTM model includes: performing grayscale and normalization processing on the input data before training; grayscale... Grayscale processing is achieved by converting a color image into a single-channel grayscale image. The calculation formula is as follows: I gray =0.299·R+0.587·G+0.114·B; Where R, G, and B are the red, green, and blue channel values ​​of an image pixel, respectively, and I gray These are the pixel values ​​of a grayscale image; Normalization is performed on the grayscale image, and the normalization formula is: Among them: I gray The pixel value after grayscale conversion; min(I gray ) and max(I gray ) are the minimum and maximum gray values ​​of the image, respectively; I norm The normalized pixel value ranges from [0, 1].

8. The intelligent floating yarn detection system for a computerized flat knitting machine according to claim 7, characterized in that, The optimization criteria for the CNN-LSTM model include: calculating the squared difference between the recognition thresholds of different parameters for floating yarn images and non-floating yarn images under the same conditions; optimizing the model performance by minimizing this squared difference, thereby determining the most ideal threshold parameters for different yarn structures.

9. The intelligent floating yarn detection system for a computerized flat knitting machine according to claim 1, characterized in that, The communication module (501) adopts the CONN-TH_6P interface, supports 24V power supply and CAN communication protocol, and is used for quick access to the flat knitting machine operating system; The camera recognition module (101) includes a power conversion module, an image processing module 401, and a communication processing module. The communication processing module uses an STM32F103 and an SN65HVD chip to achieve CAN communication. The power conversion module uses an MP1584 voltage conversion chip to perform step-down and voltage regulation.

10. The intelligent floating yarn detection system for a computerized flat knitting machine according to claim 1, characterized in that, It includes a flat knitting machine end operation window, which is used to realize data transmission, parameter adjustment, template data viewing and alarm prompt functions. The flat knitting machine end operation window is integrated into the Logica system and supports three working modes: detection mode, recording mode and comparison mode.

Images