Textile automatic production line control system and method based on internet of things

Abstract

The application discloses a textile automatic production line control system and method based on the Internet of Things, and relates to the field of textile machinery visual detection.The method comprises the following steps: S1: calculating the cloth reflectivity and the illumination interference factor based on the collected original data, generating the brightness color temperature adjustment coefficient, adjusting the brightness and color temperature of the light source array, and obtaining the optimized cloth image; S2: based on the optimized cloth image, calculating the edge density tension function and the structure consistency factor, and extracting the cloth surface defect information; S3: combining the cloth surface defect information and the obtained tension sensor data, calculating the tension adjustment coefficient, and generating the adjustment voltage and the adjustment frequency.Through dynamically adjusting the light source brightness and color temperature, combining multi-dimensional feature analysis and closed-loop control, high-stability image acquisition, accurate defect identification and adaptive weaving control are realized, and the cloth imaging quality, consistency and production line quality stability are effectively improved.

Description

Technical Field

[0001] This invention relates to the field of visual inspection in textile machinery, specifically to a control system and method for automated textile production lines based on the Internet of Things. Background Technology

[0002] With the continuous improvement of intelligence and automation in the textile industry, traditional weaving methods that rely on manual quality inspection and manual parameter adjustment can no longer meet the production demands of high efficiency, high precision, and high consistency. Against this backdrop, automated production lines that integrate machine vision, image processing, and equipment feedback control are gradually becoming a development trend.

[0003] Current textile quality inspection systems mostly focus on the static identification of surface defects, and generally suffer from insufficient detection accuracy, high response delay, and inability to achieve closed-loop adjustment of loom parameters. In addition, factors such as changes in ambient light, the diversity of fabric materials, and tension fluctuations during weaving can all interfere with image acquisition and defect judgment, affecting system stability and identification accuracy.

[0004] Therefore, there is an urgent need for an automated control method that integrates multi-source sensing acquisition, image enhancement, defect identification and equipment regulation to achieve dynamic perception and intelligent response of the weaving process, thereby improving the overall weaving quality and the level of system intelligence. Summary of the Invention

[0005] In view of the shortcomings of the prior art described above, the purpose of this invention is to provide an Internet of Things-based automated production line control system and method for textiles to solve the above-mentioned technical problems.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a control method for automated textile production lines based on the Internet of Things, comprising:

[0007] S1: Calculate the fabric reflectivity and illuminance interference factor based on the collected raw data, generate brightness and color temperature adjustment coefficients, adjust the brightness and color temperature of the light source array, and obtain an optimized fabric image;

[0008] S2: Based on the optimized fabric image, calculate the edge density tension function and structural consistency factor to extract fabric surface defect information;

[0009] S3: Combining information on fabric surface defects and acquired tension sensor data, calculate the tension adjustment coefficient and generate the adjustment voltage and adjustment frequency.

[0010] The present invention is further configured such that S1 includes:

[0011] The raw data includes: during the deployment phase, dark field calibration values ​​and reference whiteboard reflectance values ​​are collected using spectral sensors and ambient light sensors; during system operation, the raw intensity of the fabric band and the ambient illuminance are collected in real time, and raw fabric images are collected simultaneously using a high-resolution industrial camera.

[0012] A surface scattering correction sequence is preset based on the production fabric classification. The sequence includes: fabric material and surface scattering correction factor.

[0013] The present invention is further configured to calculate the fabric reflectance for eliminating fabric spectral distortion based on the original intensity of the band, the dark field calibration value, and the reflectance value of the reference white plate, combined with the scattering correction factor.

[0014] The illuminance interference factor is calculated based on the preset maximum illuminance value and the real-time collected on-site illuminance.

[0015] By combining the fabric reflectivity and illuminance interference factor, a brightness and color temperature adjustment coefficient is generated.

[0016] The present invention is further configured to generate a brightness adjustment value based on the obtained original brightness value of the light source array and the brightness color temperature adjustment coefficient;

[0017] Based on the standard color temperature value, a color temperature adjustment value is generated by combining the brightness color temperature adjustment coefficient;

[0018] The brightness and color temperature are adjusted by controlling the light source array via PWM based on the brightness and color temperature adjustment values, and the high-resolution industrial camera is controlled to acquire and optimize the fabric image.

[0019] The present invention is further configured such that S2 includes:

[0020] The optimized fabric image is converted to a grayscale image and then divided into image blocks according to a preset size.

[0021] Based on the calculation of the second derivative of Laplacian based on image patches, an edge density tension function is constructed to enhance the local edge response.

[0022] Based on the edge density tension function and the brightness color temperature adjustment coefficient, combined with the maximum edge density value in historical records, the structural consistency perturbation factor is calculated.

[0023] The present invention is further configured to determine abnormal regions based on the spatial distribution of structural consistency perturbation factors in the image, and output fabric surface defect information, wherein the fabric surface defect information includes: defect center coordinates, defect category, defect level score, and defect level;

[0024] The coordinates of the defect center are obtained by extracting the position of the centroid of the tension weight in the image through the spatial distribution of the edge density tension function.

[0025] The present invention is further configured to extract abnormal pixel points based on the edge density tension function and using a dynamic thresholding method to generate a binary mask image;

[0026] Connectivity analysis is performed based on binary mask images to extract continuously clustered high-tension regions as candidate defect regions.

[0027] Based on the defect candidate region, the gray-level variance gradient intensity of pixels within the region is statistically analyzed using the Laplacian operator.

[0028] Based on the optimized fabric image, the image is converted from RGB to Lab color space using standard color space conversion, and the color offset value of each pixel value in the defect candidate region is calculated by combining the color mean in the color space.

[0029] The local texture direction of the defect candidate region is extracted based on the Gabor filter directional response, and the global texture direction of the fabric image is optimized. The texture tilt is obtained based on the difference between the local texture direction and the global texture direction.

[0030] Defect categories are obtained by threshold segmentation based on grayscale variance gradient intensity, color offset value, and texture tilt.

[0031] The present invention is further configured such that, based on the defect candidate region, the area of ​​the defect region is obtained by counting the number of all pixels in the candidate region and combining the preset area of ​​a single pixel.

[0032] The texture direction loss ratio is obtained by combining the defect area, the global texture direction, and the texture direction of each pixel in the defect area extracted based on the Gabor filter direction response.

[0033] The defect level score is calculated based on the structural consistency perturbation factor, defect area, and texture direction loss ratio, and the defect level is classified according to the defect level score.

[0034] The present invention is further configured such that S3 includes:

[0035] Based on the defect level score in the fabric surface defect information and the real-time collected current loom tension status, calculate the tension adjustment coefficient required to adjust the tension roller control command;

[0036] Based on the tension adjustment coefficient and preset motor control parameters, the control voltage of the tension roller servo motor is generated;

[0037] The motor control frequency of the Wemi stepper motor is generated based on the tension adjustment coefficient and the basic stepping frequency, combined with the defect level score.

[0038] This invention also provides an Internet of Things-based automated textile production line control system, the system comprising:

[0039] Illumination adjustment module: Calculates fabric reflectivity and illuminance interference factor based on the collected raw data, generates brightness and color temperature adjustment coefficients, adjusts the brightness and color temperature of the light source array, and obtains an optimized fabric image;

[0040] Defect identification module: Based on optimized fabric images, calculates edge density tension function and structural consistency factor to extract fabric surface defect information;

[0041] Feedback adjustment module: Combining information on fabric surface defects and acquired tension sensor data, it calculates the tension adjustment coefficient and generates the adjustment voltage and adjustment frequency.

[0042] This invention provides an IoT-based automated textile production line control system and method. The method comprises: S1: calculating fabric reflectivity and illuminance interference factor based on collected raw data, generating brightness and color temperature adjustment coefficients, adjusting the brightness and color temperature of the light source array, and obtaining an optimized fabric image; S2: calculating the edge density tension function and structural consistency factor based on the optimized fabric image, and extracting fabric surface defect information; S3: combining the fabric surface defect information and acquired tension sensor data to calculate the tension adjustment coefficient, generate adjustment voltage and adjustment frequency. The resulting benefits include:

[0043] Improve image acquisition stability and imaging quality: By introducing a visible light spectrometer and an ambient light sensor, combined with a fabric surface scattering correction factor, the brightness and color temperature of the light source are dynamically adjusted, significantly reducing the impact of ambient light interference on fabric image acquisition and improving imaging consistency and subsequent recognition accuracy.

[0044] Enhanced defect identification sensitivity and robustness: By utilizing the edge density tension function and structural consistency factor, combined with multi-dimensional features, multi-scale analysis of fabric images is performed to effectively identify subtle, hidden, or structural defects, thereby improving detection coverage and accuracy.

[0045] Achieving closed-loop linkage between defect information and weaving control: Based on defect level and real-time tension data, dynamically generating tension adjustment coefficients and weft density control commands, driving tension servo motors and stepper motors to adaptively adjust loom parameters, controlling defect expansion and weaving error accumulation from the source, and improving fabric consistency and production line quality stability.

[0046] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:

[0048] Figure 1 A flowchart illustrating an exemplary embodiment of the present invention of an IoT-based automated textile production line control method;

[0049] Figure 2 This is a schematic diagram illustrating the structure of an IoT-based automated textile production line control system, which is an exemplary embodiment of the present invention. Detailed Implementation

[0050] The embodiments of the present invention will be described below with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention and not for limiting the scope of protection of the present invention.

[0051] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Therefore, the drawings only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0052] In the following description, numerous details are explored to provide a more thorough explanation of embodiments of the invention. However, it will be apparent to those skilled in the art that embodiments of the invention may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring embodiments of the invention.

[0053] Example 1:

[0054] IoT-based control methods for automated textile production lines, such as Figure 1 As shown, it includes:

[0055] S1: Calculate the fabric reflectivity and illuminance interference factor based on the collected raw data, generate brightness and color temperature adjustment coefficients, adjust the brightness and color temperature of the light source array, and obtain an optimized fabric image;

[0056] S2: Based on the optimized fabric image, calculate the edge density tension function and structural consistency factor to extract fabric surface defect information;

[0057] S3: Combining information on fabric surface defects and acquired tension sensor data, calculate the tension adjustment coefficient and generate the adjustment voltage and adjustment frequency.

[0058] The present invention is further configured such that S1 includes:

[0059] The raw data includes: during the deployment phase, dark field calibration values ​​and reference whiteboard reflectance values ​​are collected using spectral sensors and ambient light sensors; during system operation, the raw intensity of the fabric band and the ambient illuminance are collected in real time, and raw fabric images are collected simultaneously using a high-resolution industrial camera.

[0060] A surface scattering correction sequence is pre-set based on the fabric classification. This sequence includes the fabric material and the surface scattering correction factor. Specifically, the dark-field calibration value acquired by the spectral sensor refers to the background signal received by the sensor in the absence of light, used to eliminate noise from the spectral sensor under no-signal conditions. The reference white board reflectance value refers to the reflectance value measured using a standard white reference board, typically with a reflectance close to 100%. This serves as a known standard for calibrating the sensor and calculating the fabric surface reflectance, ensuring the accuracy of the reflectance measurement results. The original intensity of the fabric band refers to the intensity of reflected light from the fabric measured by the spectral sensor within a specific wavelength range. This specific wavelength range is set at 400nm, 450nm, 500nm, 550nm, 600nm, 650nm, and 700nm, respectively measuring the reflectance characteristics of the fabric surface at different spectral bands to provide basic data for subsequent fabric reflectance calculations. The ambient illuminance refers to the current ambient light intensity, measured by an ambient light sensor. Image data acquired by a high-resolution industrial camera includes details and structural features of the fabric surface. Fabric material refers to the type of fabric, such as cotton or nylon. Surface scattering correction factors are a set of preset factors used to correct the light reflection characteristics of different fabric surfaces; for example, 0.85 for cotton and 0.72 for nylon. Implementation steps: During deployment, a spectral sensor is used to collect dark-field calibration values ​​in a dark environment without fabric. An ambient light sensor is used to collect the background light intensity of the surrounding environment to obtain ambient illuminance. A reference white board is used to calibrate the spectral sensor, and the reflectance value of the reference white board is collected. During system operation: The spectral sensor collects reflectance intensity data of the fabric in different wavelength ranges. Simultaneously, an industrial camera captures high-resolution images of the fabric, recording detailed information about the fabric surface.

[0061] The present invention is further configured to calculate the fabric reflectance for eliminating fabric spectral distortion based on the original intensity of the band, the dark field calibration value, and the reflectance value of the reference white plate, combined with the scattering correction factor.

[0062] The illuminance interference factor is calculated based on the preset maximum illuminance value and the real-time collected on-site illuminance.

[0063] By combining fabric reflectivity and illuminance interference factor, a brightness and color temperature adjustment coefficient is generated. Specifically, fabric reflectivity represents the ability of a fabric surface to reflect light of different wavelengths. In visual inspection, fabric reflectivity is an important indicator, affecting the brightness, color, and defect recognition accuracy in subsequent processing. The fabric reflectivity calculation logic is as follows: ,in, Fabric reflectivity represents the intensity of reflection when light of a specific wavelength strikes the surface of the fabric. The original intensity of the band; This is the dark field calibration value; Use the whiteboard's reflectance value as a reference. The scattering correction factor corrects for errors in reflectivity calculation caused by varying surface light scattering characteristics in different fabric materials. The illumination interference factor corrects for the impact of ambient light intensity changes on fabric image acquisition. Changes in ambient light can cause fluctuations in image brightness, affecting image quality and consequently the accuracy of subsequent image processing and defect identification. The illumination interference factor calculation logic is as follows: ,in, Illuminance interference factor; This is the maximum illuminance value, which is preset in the early stages of deployment; the current system defaults to 2000 Lux. The ambient illuminance represents the actual light intensity of the environment surrounding the fabric surface. The brightness and color temperature adjustment coefficient is a light source control factor that integrates image reflectivity characteristics and ambient illuminance disturbances. It is used to address image brightness and color distortion caused by differences in fabric material and variations in ambient illuminance, ensuring that images captured by industrial cameras have consistency, comparability, and high recognizability. The calculation logic for the brightness and color temperature adjustment coefficient is as follows: ,in, This refers to the brightness and color temperature adjustment coefficient. Illuminance interference factor; The average reflectance is the average reflectance obtained by taking the visual weighted average of the reflectance of each band. The calculation formula is: , To count the number of bands, such as the wavelengths of 400nm, 450nm, 500nm, 550nm, 600nm, 650nm, and 700nm calculated in this scheme, n is 7; For the fabric reflectivity, and To standardize the parameters, i here is an index, indicating that the fabric reflectivity can be retrieved from any band in the 400nm-700nm band. The weighting coefficients are set with reference to the CIE1931V(λ) curve. The center wavelength of band 400nm is purple with a weighting coefficient of 0.01; the center wavelength of band 450nm is blue with a weighting coefficient of 0.05; the center wavelength of band 500nm is cyan with a weighting coefficient of 0.25; the center wavelength of band 550nm is green with a weighting coefficient of 1.00 (this band is the most sensitive, hence the highest weighting coefficient); the center wavelength of band 600nm is yellow with a weighting coefficient of 0.63; the center wavelength of band 650nm is red with a weighting coefficient of 0.21; and the center wavelength of band 700nm is dark red with a weighting coefficient of 0.01.

[0064] The present invention is further configured to generate a brightness adjustment value based on the obtained original brightness value of the light source array and the brightness color temperature adjustment coefficient;

[0065] Based on the standard color temperature value, a color temperature adjustment value is generated by combining the brightness color temperature adjustment coefficient;

[0066] Based on brightness and color temperature adjustment values, the brightness and color temperature of the light source array are adjusted via PWM control, which in turn controls a high-resolution industrial camera to acquire and optimize fabric images. Specifically, the actual output PWM brightness control signal is obtained by multiplying the brightness and color temperature adjustment coefficients with the original brightness value of the light source array read from the LED control system. This signal serves as the brightness control input for the light source array, adjusting the light source brightness. 6500K is set as the central reference because it corresponds to the standard daylight color temperature. The color temperature adjustment value is calculated, where, This is the color temperature adjustment value; Centered on the benchmark; This refers to the brightness and color temperature adjustment coefficients. The brightness and color temperature adjustment values ​​are mapped to the PWM drive controller. The brightness adjustment value controls the total luminous flux of the LED, and the color temperature adjustment value controls the R / G / B channel ratio or the mixing ratio of dual-color LEDs. Finally, the industrial camera is controlled to acquire images under new lighting conditions, generating an "optimized fabric image" for subsequent analysis.

[0067] The present invention is further configured such that S2 includes:

[0068] The optimized fabric image is converted to a grayscale image and then divided into image blocks according to a preset size.

[0069] Based on the calculation of the second derivative of Laplacian based on image patches, an edge density tension function is constructed to enhance the local edge response.

[0070] Based on the edge density tension function and brightness color temperature adjustment coefficient, combined with the maximum edge density value in historical records, the structural consistency perturbation factor is calculated. Specifically, converting the optimized fabric image into a grayscale image is a fundamental step in image preprocessing. Existing image processing technologies have various mature and standardized implementation methods, such as standard grayscale conversion algorithms based on human eye perception weighting, such as the ITU-R BT.601 weighted method; mainstream implementation tools include OpenCV, PIL, MATLAB, and TensorFlow. These methods can be flexibly selected according to application requirements and can all effectively achieve the conversion of color images to grayscale images. This invention does not limit this. The entire image is divided into multiple image blocks according to a preset size for fine-grained structure analysis. The preset size here can be adaptively adjusted according to the detection efficiency requirements. The default is 16*16. If you want to improve the detection accuracy, you can adjust it to 8*8. If you want to adjust the detection speed, you can adjust it to 32*32. For each divided image block, the Laplacian operator is applied to calculate the second derivative image to obtain the local edge response. The edge density tension function is calculated in combination with the local edge response. The edge density tension function is a function representing the complexity of local edges, used to measure the change in edge density within an image patch, and is used for subsequent defect detection; the calculation logic of the edge density tension function is as follows: ,in, It is the edge density tension function; , These are the side lengths of the image blocks, i.e., the preset dimensions; For the Laplace operator; For image blocks; The second derivative of the image gradient of the image patch is used to detect edge / texture abrupt changes in the image; The power factor is used to amplify edge response and enhance the tension representation of minor defects. The structural consistency perturbation factor describes the "abnormality" of image patches, i.e., the degree of deviation from the normal state of the system; the calculation logic of the structural consistency perturbation factor is as follows: ,in, The structural consistency perturbation factor; It is the edge density tension function; The maximum edge density value is the numerical value of the maximum edge density tension function stored in the system execution record; This refers to the brightness and color temperature adjustment coefficient. It is a sensitivity control factor used to control the degree of response to highly abnormal regions.

[0071] The present invention is further configured to determine abnormal regions based on the spatial distribution of structural consistency perturbation factors in the image, and output fabric surface defect information, wherein the fabric surface defect information includes: defect center coordinates, defect category, defect level score, and defect level;

[0072] The defect center coordinates are obtained by extracting the position of the centroid of tension weights in the image through the spatial distribution of the edge density tension function. Specifically, the "edge density weighted centroid method" is a weighted centroid estimation method based on the distribution of image edge features. Instead of directly finding the location of the maximum or mean value, it calculates the geometric centroid in space by weighting the "edge intensity" of each pixel or image block, thus obtaining the defect center coordinates. The defect center coordinate calculation logic is as follows: , ,in, The x-coordinate of the defect center; The y-coordinate is the coordinate of the defect center. Let be the edge density tension function of the image patch, where i and j represent the row and column indices of the image patch, i.e., the x and y coordinates. This indicates weighted horizontal coordinates. Indicates weighted vertical coordinates; This is the edge density tension function for the entire optimized fabric image.

[0073] The present invention is further configured to extract abnormal pixel points based on the edge density tension function and using a dynamic thresholding method to generate a binary mask image;

[0074] Connectivity analysis is performed based on binary mask images to extract continuously clustered high-tension regions as candidate defect regions.

[0075] Based on the defect candidate region, the gray-level variance gradient intensity of pixels within the region is statistically analyzed using the Laplacian operator.

[0076] Based on the optimized fabric image, the image is converted from RGB to Lab color space using standard color space conversion, and the color offset value of each pixel value in the defect candidate region is calculated by combining the color mean in the color space.

[0077] The local texture direction of the defect candidate region is extracted based on the Gabor filter directional response, and the global texture direction of the fabric image is optimized. The texture tilt is obtained based on the difference between the local texture direction and the global texture direction.

[0078] Defect categories are obtained using a threshold segmentation method based on grayscale variance gradient intensity, color offset value, and texture tilt. Specifically, the binary mask image calculates the edge density tension function at each pixel location to obtain the edge density tension value. All pixels exceeding a preset threshold are identified as abnormal pixels and included in the binary mask image. The preset threshold is obtained by calculating the sum of the mean and standard deviation of the edge density tension function of the optimized fabric image. The standard deviation is obtained by calculating the difference between the edge density tension function of the standard fabric template and the calculated edge density tension function. Defect candidate regions are obtained by performing connected component analysis on the pixels marked as abnormal in the binary mask image. To prevent false positives due to noise, a minimum area can be set; if the area is less than 30 pixels, it is considered noise and not included in the defect candidate region. Grayscale variance gradient intensity is used to measure the severity of structural changes in the image, i.e., the degree of local grayscale changes. It is mainly used to detect typical steep-edge or broken-edge defects such as holes, tears, and weft breaks. The calculation logic for grayscale variance gradient intensity is as follows: ,in, The grayscale variance gradient intensity; This represents the total number of pixels in the defect candidate region; This is a candidate region for defects. The Laplace operator represents a pixel. The second derivative at a certain point reflects the gradient of grayscale changes; 1.2 is an exponential factor used to improve the sensitivity of high-frequency gradient response and avoid over-smoothing. In optimizing the fabric image, a region representing the fabric's "normal color performance" is selected as a reference color region. This region should have no obvious defects, uniform color, and consistent texture. It can be manually calibrated or automatically determined by historical system data. Subsequently, the system will statistically analyze all pixels within this region, calculating the average values ​​of the red, green, and blue color channels as the standard color of the fabric. Then, the entire image is converted from the red-green-blue color space to a color space more consistent with human visual perception, namely the CIELAB color space, also known as the Lab color space. This space consists of three channels: the first channel represents brightness, and the latter two channels represent the two directions of color. The color space changes, such as from green to red, or from blue to yellow. During the conversion process, the red, green, and blue values ​​are typically first converted to an intermediate linear color space, and then mapped to the final Lab space. This conversion process can be automated using OpenCV, MATLAB, or other image processing libraries. For each pixel in a defective region, the system calculates the difference between its Lab value and the average value of the reference region to determine the degree of difference between that pixel and the standard color. Then, all difference values ​​are aggregated or averaged to represent the overall color shift between the defective region and the normal region, resulting in a color shift value. The larger the shift value, the more significant the color difference and the more severe the deviation. Texture tilt is an indicator that measures the difference between the texture direction of a defective region and the overall fabric texture direction. It is mainly used to detect structural defects in fabric images caused by mechanical or technological problems, such as abnormal texture direction, misaligned weaves, and yarn twisting. By applying a set of multi-directional Gabor filters to the defect candidate region, such as sampling every 15 or 30° from 0° to 180°, the filter responses in different directions are obtained. The direction with the largest response value is the local texture direction. Similarly, a multi-directional Gabor filter is used, but applied to the entire optimized cloth image to find the strongest overall direction as the global texture direction. The texture tilt is obtained by calculating the absolute value of the difference between the global texture direction and the local texture direction.Defects are categorized into holes, stains, weft skew, and mixed types. The defect type determination is based on three core feature parameters: grayscale variance gradient intensity, texture tilt, and color shift value. Based on the combination of these three indicators, the defect type in the fabric image can be determined according to the following rules: 1. When the grayscale variance gradient intensity is higher than 15, the texture tilt is lower than 5 degrees, and the color shift value is lower than 10, the area is considered a "hole." This is because holes are usually accompanied by strong abrupt changes in grayscale structure, but do not necessarily cause significant changes in texture direction or color; 2. If the color shift value is significantly higher than the specified value... If the value is greater than 20, and the grayscale variance gradient intensity is weak (less than 5), the area is classified as a "stain" because this type of defect shows obvious color difference changes but no obvious structural abrupt changes or texture abnormalities; 3. When the texture tilt is large (more than 10 degrees), it is classified as "weft skew," meaning that the fabric texture direction has shifted or twisted, usually related to abnormal weaving tension or direction; If no single combination of the above conditions is met, or the distribution of the values ​​of each indicator is relatively complex, the defect is classified as "mixed type," and its type needs to be further determined by combining other features or expert systems; The three judgment rules are matched in sequence, and only if the previous rule is not met can the judgment of the next rule be carried out.

[0079] The present invention is further configured such that, based on the defect candidate region, the area of ​​the defect region is obtained by counting the number of all pixels in the candidate region and combining the preset area of ​​a single pixel.

[0080] The texture direction loss ratio is obtained by combining the defect area, the global texture direction, and the texture direction of each pixel in the defect area extracted based on the Gabor filter direction response.

[0081] A defect level score is calculated based on the structural consistency perturbation factor, defect area, and texture direction loss ratio. Defect levels are then classified according to this score. Specifically, firstly, in the defect candidate region, the number of pixels marked as defects is counted. Each pixel corresponds to a certain physical area; for example, each pixel in a camera covers 0.01 mm². The actual physical area of ​​the defect region is then calculated by multiplying the number of pixels by the area of ​​a single pixel. For each pixel in the defect region, a Gabor filter is used to extract its local texture direction. These pixel directions are compared with the global main texture direction extracted from the entire image, and their average direction difference is calculated. This direction difference is then normalized and converted into a consistency coefficient. A weighted average is used to calculate the texture direction consistency level within the entire defect region, resulting in a texture loss ratio between 0 and 1. The closer the value is to 1, the greater the difference between the local and global texture directions. Finally, the structural consistency perturbation factor, defect area, and texture direction loss ratio are multiplied by weights, with the structural perturbation factor accounting for 40%, the area for 30%, and the texture loss ratio for 30%. The combined score of these three indicators yields the final defect level score. Finally, the defect levels are categorized into intervals based on this defect rating: a score below 2 indicates a very minor defect; a score between 2 and 5 indicates a minor defect; a score between 5 and 10 indicates a moderate defect; a score between 10 and 18 indicates a severe defect; and a score greater than or equal to 18 indicates a very severe defect. These defect levels can be saved in the system's execution log or detection log for administrators to view and quantify defect information in real time.

[0082] The present invention is further configured such that S3 includes:

[0083] Based on the defect level score in the fabric surface defect information and the real-time collected current loom tension status, calculate the tension adjustment coefficient required to adjust the tension roller control command;

[0084] Based on the tension adjustment coefficient and preset motor control parameters, the control voltage of the tension roller servo motor is generated;

[0085] Based on the tension adjustment coefficient and the base stepper frequency, the motor control frequency of the weft density stepper motor is generated in conjunction with the defect level score. Specifically, the tension adjustment coefficient is a dimensionless coefficient used to dynamically adjust the tension roller control command. It is dynamically generated according to the severity of the current fabric defect and the actual tension state of the loom, controlling the magnitude of the tension adjustment. It is used for fine-tuning the loom tension system, ensuring that when structural defects occur, tension impact is reduced through automatic adjustment to prevent further damage to the textile structure and improve fabric quality. Tension adjustment coefficient calculation logic: ,in, This is the tension adjustment coefficient; For the amplitude limiting function, the preceding... This is the calculated tension adjustment coefficient. The 0.7 in the middle is the minimum limit, and the 1.3 at the end is the maximum limit. Rate the defect level; The loom tension status is determined by the real-time tension of the loom, which is collected in real time by a tension sensor. Maximum tension is the maximum safe tension used to prevent fabric damage, preset to different maximum tension values ​​for different fabric materials. Servo motor control voltage controls the voltage value of the tension roller servo motor; higher voltage results in stronger driving force, and vice versa. The control voltage to be applied to the tension control servo motor is obtained by multiplying the tension adjustment coefficient by the motor's gain parameter. This voltage controls the motor to adjust the rotational speed and direction of the tension roller, thereby indirectly regulating the fabric tension. Weft density stepper motor control frequency is a frequency parameter used to control the operating speed of the weft density stepper motor. Weft density refers to the number of weft threads per unit length; higher frequency results in greater weft density, while lower frequency results in relatively lower fabric density. Excessive weft density in defective areas can lead to structural tension misalignment; therefore, reducing the weft density frequency can mitigate the impact of defects on the structure and improve the stability of the finished product. The system then uses the defect level score as the main variable to attenuate the basic weft density frequency. The higher the score, the lower the frequency, thereby reducing the local weft density of the fabric and preventing problems such as weave misalignment and yarn breakage caused by the superposition of high weft density and defect tension. In this way, even if there are obvious defects in the fabric, the process risk can be buffered by automatically controlling the weft input speed.

[0086] Example 2:

[0087] Please see Figure 2 This exemplary IoT-based automated textile production line control system includes:

[0088] Illumination adjustment module: Calculates fabric reflectivity and illuminance interference factor based on the collected raw data, generates brightness and color temperature adjustment coefficients, adjusts the brightness and color temperature of the light source array, and obtains an optimized fabric image;

[0089] Defect identification module: Based on optimized fabric images, calculates edge density tension function and structural consistency factor to extract fabric surface defect information;

[0090] Feedback adjustment module: Combining information on fabric surface defects and acquired tension sensor data, it calculates the tension adjustment coefficient and generates the adjustment voltage and adjustment frequency.

[0091] It should be noted that the IoT-based automated textile production line control system and the IoT-based automated textile production line control method provided in the above embodiments belong to the same concept. The specific methods by which each module and unit performs its operations have been described in detail in the method embodiments and will not be repeated here. In practical applications, the IoT-based automated textile production line control system provided in the above embodiments can be configured to distribute the above functions among different functional modules as needed, that is, the internal structure of the system can be divided into different functional modules to complete all or part of the functions described above. This is not a limitation here.

[0092] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A control method for automated textile production lines based on the Internet of Things, characterized in that, include: S1: Based on the collected raw data, calculate the fabric reflectance and illuminance interference factor, generate brightness and color temperature adjustment coefficients, adjust the brightness and color temperature of the light source array, and obtain an optimized fabric image. The raw data includes: during deployment, collecting dark field calibration values ​​and reference whiteboard reflectance values ​​using spectral and ambient light sensors; during system operation, collecting real-time raw fabric intensity and ambient illuminance, and simultaneously acquiring raw fabric images using a high-resolution industrial camera; pre-setting a surface scattering correction sequence based on the fabric classification, the sequence including: fabric material and surface scattering correction factor; calculating the fabric reflectance to eliminate spectral distortion based on the raw intensity, dark field calibration value, and reference whiteboard reflectance, combined with the scattering correction factor; calculating the illuminance interference factor based on the preset maximum illuminance value and real-time collected ambient illuminance; and generating brightness and color temperature adjustment coefficients by combining the fabric reflectance and illuminance interference factor. S2: Based on the optimized fabric image, calculate the edge density tension function and structural consistency factor, and extract fabric surface defect information, including: converting the optimized fabric image into a grayscale image and dividing the image into blocks according to a preset size; calculating the second derivative of Laplacian based on the image blocks to construct an edge density tension function to enhance local edge response; and calculating the structural consistency perturbation factor based on the edge density tension function and the brightness color temperature adjustment coefficient, combined with the maximum edge density value in historical records. S3: Combining fabric surface defect information and acquired tension sensor data, calculate the tension adjustment coefficient, generate the adjustment voltage and adjustment frequency, including: calculating the tension adjustment coefficient required to adjust the tension roller control command based on the defect level score in the fabric surface defect information and the real-time acquired current loom tension status; generating the control voltage of the tension roller servo motor based on the tension adjustment coefficient and preset motor control parameters; and generating the motor control frequency of the weft density stepper motor based on the tension adjustment coefficient and the basic stepper frequency, combined with the defect level score.

2. The control method for an automated textile production line based on the Internet of Things according to claim 1, characterized in that, S1 includes: A brightness adjustment value is generated based on the original brightness value of the light source array and the brightness color temperature adjustment coefficient; Based on the standard color temperature value, a color temperature adjustment value is generated by combining the brightness color temperature adjustment coefficient; The brightness and color temperature are adjusted by controlling the light source array via PWM based on the brightness and color temperature adjustment values, and the high-resolution industrial camera is controlled to acquire and optimize the fabric image.

3. The control method for an automated textile production line based on the Internet of Things according to claim 1, characterized in that, S2 includes: Based on the spatial distribution of the structural consistency perturbation factor in the image, abnormal areas are identified, and fabric surface defect information is output. The fabric surface defect information includes: defect center coordinates, defect category, defect level score, and defect level. The coordinates of the defect center are obtained by extracting the position of the centroid of the tension weight in the image through the spatial distribution of the edge density tension function.

4. The IoT-based automated textile production line control method according to claim 3, characterized in that, Based on the edge density tension function, anomaly pixels are extracted using a dynamic thresholding method to generate a binary mask image; Connectivity analysis is performed based on binary mask images to extract continuously clustered high-tension regions as candidate defect regions. Based on the defect candidate region, the gray-level variance gradient intensity of pixels within the region is statistically analyzed using the Laplacian operator. Based on the optimized fabric image, the image is converted from RGB to Lab color space using standard color space conversion, and the color offset value of each pixel value in the defect candidate region is calculated by combining the color mean in the color space. The local texture direction of the defect candidate region is extracted based on the Gabor filter directional response, and the global texture direction of the fabric image is optimized. The texture tilt is obtained based on the difference between the local texture direction and the global texture direction. Defect categories are obtained by threshold segmentation based on grayscale variance gradient intensity, color offset value, and texture tilt.

5. The IoT-based automated textile production line control method according to claim 4, characterized in that, Based on the defect candidate region, the area of ​​the defect region is obtained by counting the number of all pixels in the candidate region and combining it with the preset area of ​​a single pixel. The texture direction loss ratio is obtained by combining the defect area, the global texture direction, and the texture direction of each pixel in the defect area extracted based on the Gabor filter direction response. The defect level score is calculated based on the structural consistency perturbation factor, defect area, and texture direction loss ratio, and the defect level is classified according to the defect level score.

6. An IoT-based automated textile production line control system, used to implement the IoT-based automated textile production line control method according to any one of claims 1-5, characterized in that, include: Illumination adjustment module: Calculates fabric reflectivity and illuminance interference factor based on the collected raw data, generates brightness and color temperature adjustment coefficients, adjusts the brightness and color temperature of the light source array, and obtains an optimized fabric image; Defect identification module: Based on optimized fabric images, calculates edge density tension function and structural consistency factor to extract fabric surface defect information; Feedback adjustment module: Combining information on fabric surface defects and acquired tension sensor data, it calculates the tension adjustment coefficient and generates the adjustment voltage and adjustment frequency.

Images