A non-sensitive operation quality tracing method for ultra-fine yarn

Abstract

The present invention provides a method for traceability of the quality of ultra-fine yarn without feeling operation, which relates to the technical field of intelligent management. The method includes: in the wire drawing area, automatically identifying and numbering the bobbin and cake on the wire drawing machine through a camera and a machine vision system; the variety produced by each wire drawing machine is bound to the machine vision system, and a visible bar code is printed on the cake as the unique identifier of the cake, corresponding to the number identified by the machine vision system; after wire drawing is completed, the cake is sent to the twisting area through a transfer channel; cameras are installed in the transfer channel, and machine vision technology is used to identify and record the bar code information on the passing cake, so as to ensure that the tracking of the cake during the transfer process is not interrupted. The present invention can accurately monitor each link in the production process, realize the rapid positioning and solution of problems, and thus effectively improve production efficiency and product quality.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent management, and particularly to a method for traceability of the quality of ultra-fine yarn without feeling operation. Background Art

[0002] Drawing, twisting and packaging are the core links in yarn production, each having an irreplaceable role. As the starting point of the production process, the quality of the drawing link lays the foundation for the subsequent processes and directly affects the smoothness of the entire production process. The twisting link plays a decisive role in the internal structure and strength of the yarn and is the key to ensuring that the final performance of the product meets the standards. And the packaging link, as the interface where the product directly contacts the consumer, not only concerns the aesthetic appearance of the product, but also affects the customer's first impression of the product and the purchase decision.

[0003] Although traditional methods already include certain traceability means, these means often have many defects. For example, relying on paper records or simple spreadsheets for tracking is not only inefficient, but also the data is prone to errors or loss. When problems occur in these key links, some traditional traceability methods cannot quickly and accurately lock the source of the problem. If it is found that the quality of a batch of yarn does not meet the standard, the enterprise may need to invest a lot of manpower and time to review and check each production link one by one, which obviously cannot meet the dual pursuit of efficiency and quality in modern production. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method for traceability of the quality of ultra-fine yarn without feeling operation, which can accurately monitor each link in the production process, realize the rapid positioning and solution of problems, and thus effectively improve production efficiency and product quality.

[0005] To solve the above technical problems, the technical solution of the present invention is as follows:

[0006] In the first aspect, a method for traceability of the quality of ultra-fine yarn without feeling operation, the method includes:

[0007] In the drawing area, the bobbins and cakes on the drawing machine are automatically identified and numbered through cameras and machine vision systems; the variety produced by each drawing machine is bound to the machine vision system, and a visible bar code is printed on the cake as the unique identifier of the cake, corresponding to the number identified by the machine vision system;

[0008] After the drawing is completed, the cakes are sent to the twisting area through the transfer channel; cameras are installed in the transfer channel, and machine vision technology is used to identify and record the bar code information on the passing cakes, so as to ensure that the tracking of the cakes during the transfer process is not interrupted;

[0009] In the twisting area, each twisting machine is equipped with a unique machine number, which is displayed in the form of a QR code; when the silk cake reaches the twisting area, a scanning device is used to scan the QR code of the twisting machine and the barcode of the silk cake, and the information is bound;

[0010] After twisting, the yarn is transported to the packaging area for inspection and packing; through the manipulator in the packaging area, the tube yarn with a QR code is identified and grabbed; the QR code scanning device in the packaging area scans and confirms the tube yarn to be packed, so that each step of the operation is traceable and the whole process of yarn production is managed.

[0011] Furthermore, the silk bobbins and silk cakes on the wire drawing machine are automatically identified and numbered through a camera and a machine vision system, including:

[0012] Install the camera at a preset position on the wire drawing machine to capture the images of the silk bobbins and silk cakes on the wire drawing machine in real time through the camera;

[0013] Preprocess the collected images, perform image segmentation, and separate the silk bobbins and silk cakes from the background;

[0014] Obtain a frame of color image from the camera, and calculate the gray value for each pixel in the color image;

[0015] Generate a grayscale image according to the gray value, and calculate the new value of each pixel through convolution operation with a Gaussian function to obtain an image after Gaussian filtering;

[0016] Use the horizontal and vertical direction kernels of the Sobel operator to perform convolution operations on the image after Gaussian filtering respectively, and obtain the gradient components of each pixel point in the horizontal and vertical directions;

[0017] For each pixel point, calculate the gradient amplitude according to the gradient components of each pixel point in the horizontal and vertical directions to obtain a gradient image containing the gradient amplitude and direction of each pixel point;

[0018] Traverse the gradient image of the gradient amplitude and direction of each pixel point, and identify and extract the continuous edge contours in the image;

[0019] For each extracted contour, calculate the number of pixel points on the contour, the radius of curvature or curvature angle of each point on the contour, the overall direction of the contour or the direction of each segment;

[0020] According to the number of pixel points on the contour, the radius of curvature or curvature angle of each point on the contour, the overall direction of the contour or the direction of each segment, screen out the edge features of the silk bobbins and silk cakes;

[0021] According to the edge features of the silk bobbins and silk cakes, identify the silk bobbins and silk cakes in the image, and determine the position and attitude of the silk bobbins and silk cakes;

[0022] Assign a unique number to each identified bobbin and cake of yarn, and associate the number with the images, positions, and times of the bobbins and cakes of yarn.

[0023] Furthermore, generate a grayscale image based on the grayscale values, and calculate the new value of each pixel through convolution operation using the Gaussian function to obtain the image after Gaussian filtering, including:

[0024] Calculate the grayscale values of all pixels to create a new image, where the value of each pixel is the grayscale value, and the new image is the grayscale image;

[0025] Define a Gaussian kernel, which is a two-dimensional matrix whose values are calculated according to the Gaussian function;

[0026] Starting from the upper left corner of the image, select the first pixel, multiply the first pixel and its surrounding pixels by the values at the corresponding positions of the Gaussian kernel; align the center of the Gaussian kernel with the selected pixel, and perform the following operations:

[0027] Multiply the upper left corner value of the Gaussian kernel by the pixel value at the corresponding position in the image;

[0028] Multiply the upper middle value of the Gaussian kernel by the pixel value at the corresponding position in the image until all the values of the Gaussian kernel are multiplied by the pixel values at the corresponding positions in the image, and add up all the products. The result is the new pixel value at the corresponding position in the filtered image;

[0029] Move the Gaussian kernel one pixel position to the right until the new values of all pixels in the image are calculated, and create a new image using the calculated new pixel values. The new image is the image after Gaussian filtering.

[0030] Furthermore, perform convolution operations on the image after Gaussian filtering using the horizontal and vertical direction kernels of the Sobel operator respectively to obtain the gradient components of each pixel point in the horizontal and vertical directions, including:

[0031] Starting from the upper left corner of the image, select a pixel point as the current processing point; cover the horizontal kernel of Sobel on the current pixel point and its surrounding pixels, and perform convolution operation:

[0032] Multiply the upper left corner element of the horizontal kernel by the value of the corresponding image pixel; multiply the upper middle element of the horizontal kernel by the value of the corresponding image pixel, and repeat this process until all the elements of the horizontal kernel are multiplied by the corresponding image pixels; add up all the products. The result is the gradient component of the current pixel point in the horizontal direction;

[0033] Perform convolution operation on the same pixel point and its surrounding pixels using the vertical kernel of Sobel:

[0034] Multiply the upper left element of the vertical kernel by the value of the corresponding image pixel; multiply the upper middle element of the vertical kernel by the value of the corresponding image pixel, and repeat this process until all elements of the vertical kernel are multiplied by the corresponding image pixels. Add up all the products, and the result is the gradient component of the current pixel point in the vertical direction;

[0035] Move the currently processed point one pixel to the right until the gradient components of all pixels in the image are calculated in both the horizontal and vertical directions;

[0036] Generate a gradient component image based on the gradient components of all pixels in the image in both the horizontal and vertical directions.

[0037] Furthermore, install the camera at a preset position on the wire drawing machine to capture images of the bobbin and cake on the wire drawing machine in real time, including:

[0038] Determine the set of all potential position points where the camera can be installed. The position points are fixed points around the wire drawing machine; define the attributes of each position point, including coordinates and viewing angle range;

[0039] Randomly select a part of all potential position points as the initial positions of the ants; assign a memory table to each ant to record the positions it has visited and the corresponding pheromone concentrations;

[0040] Determine the initial concentration and evaporation rate of the pheromone. Each ant selects the next position to move to based on the pheromone concentration and heuristic information around its current position;

[0041] After a certain number of iterations, determine the final position point based on the memory tables and path evaluation results of all ants;

[0042] Install the camera according to the final position point and capture images of the bobbin and cake on the wire drawing machine in real time through the camera.

[0043] Furthermore, the variety produced by each wire drawing machine is bound to the machine vision system. A visible barcode is printed on the cake as the unique identifier of the cake, corresponding to the number recognized by the machine vision system, including:

[0044] Assign a unique number to the variety produced by each wire drawing machine;

[0045] Print the assigned number on the cake in the form of a barcode as the unique identifier of the cake;

[0046] Establish a database or lookup table to store the correspondence between the variety and the number;

[0047] Preprocess the acquired images using image processing techniques, scan and analyze the preprocessed images, and extract the serial number information on the cake;

[0048] According to the extracted serial number information, find the corresponding variety information and bind the variety information and the serial number information.

[0049] Furthermore, assign a unique serial number to each variety produced by each wire drawing machine, including:

[0050] Determine the basic rules of the serial number, including the length of the serial number and the character set used;

[0051] Obtain the variety information produced by all wire drawing machines, including variety names, specifications, and characteristics;

[0052] Represent the serial number of each variety as a gene code. According to the defined gene code, randomly generate an initial population, and each individual in the population represents a serial number assignment scheme;

[0053] Determine the fitness function used to evaluate the quality of each individual;

[0054] According to the fitness function, select the corresponding individuals from the current population to enter the next generation, perform crossover operations on the selected individuals to generate new individuals, and perform mutation operations on the newly generated individuals until the preset number of iterations is reached to obtain the final serial number assignment scheme;

[0055] According to the final serial number assignment scheme, assign a unique serial number to each variety produced by each wire drawing machine.

[0056] In a second aspect, an ultrafine yarn non-sensing operation quality traceability system is applied to the method described above, including:

[0057] An encoding module for automatically identifying and numbering the bobbins and cakes on the wire drawing machine through a camera and a machine vision system in the wire drawing area; each variety produced by each wire drawing machine is bound to the machine vision system, and a visible bar code is printed on the cake as the unique identifier of the cake, corresponding to the number identified by the machine vision system;

[0058] A tracking module for, after wire drawing is completed, sending the cakes to the twisting area through a transfer channel; cameras are installed in the transfer channel, and machine vision technology is used to identify and record the bar code information on the passing cakes so that the tracking of the cakes during the transfer process is not interrupted;

[0059] A scanning module for, in the twisting area, each twisting machine is equipped with a unique machine number and is displayed in the form of a two-dimensional code; when the cake reaches the twisting area, a scanning device is used to scan the two-dimensional code of the twisting machine and the bar code of the cake and bind the information;

[0060] A processing module is used to complete the twisting of yarn, and then the yarn is transported to the packaging area for inspection and packing. Through the manipulator in the packaging area, it can identify and grasp the tube yarn with a QR code. The QR code scanning device in the packaging area scans and confirms the tube yarn to be packed, so that each operation can be traced, and the whole process of yarn production can be managed.

[0061] In a third aspect, a computing device includes:

[0062] One or more processors;

[0063] A storage device for storing one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the method described above.

[0064] In a fourth aspect, a computer-readable storage medium stores a program, and when the program is executed by a processor, the method described above is implemented.

[0065] The above solution of the present invention has at least the following beneficial effects.

[0066] By integrating machine vision, barcode and QR code technologies, the present invention realizes the comprehensive monitoring and data recording of each link in the yarn production process. This greatly enhances the transparency of the production process, makes each operation traceable, and helps enterprises to discover and solve problems in a timely manner.

[0067] The automated identification and recording system reduces human operation errors and delays, and improves production efficiency. At the same time, by accurately tracking each link, problems in production can be quickly located and processed, thus effectively improving product quality. The automated traceability system reduces the need for manual inspection and recording, thereby reducing labor costs. At the same time, by reducing waste and rework in the production process, the operating costs are further reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 is a schematic flow chart of a method for quality traceability of ultra-fine yarn non-sensing operation provided by an embodiment of the present invention.

[0069] Figure 2 is a schematic diagram of a system for quality traceability of ultra-fine yarn non-sensing operation provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0070] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be fully communicated to those skilled in the art.

[0071] As Figure 1 shown, an embodiment of the present invention provides a method for traceability of the quality of ultra-fine yarn without-sense operation. The method includes the following steps:

[0072] Step 11, in the wire drawing area, automatically identify and number the spools and cakes on the wire drawing machine through a camera and a machine vision system; the variety produced by each wire drawing machine is bound to the machine vision system, and a visible barcode is printed on the cake as the unique identifier of the cake, corresponding to the number identified by the machine vision system;

[0073] Step 12, after wire drawing is completed, the cakes are sent to the twisting area through a transfer channel; cameras are installed in the transfer channel, and machine vision technology is used to identify and record the barcode information on the passing cakes, so that the tracking of the cakes during the transfer process is not interrupted;

[0074] Step 13, in the twisting area, each twisting machine is equipped with a unique machine number and is displayed in the form of a two-dimensional code; when the cake arrives at the twisting area, a scanning device is used to scan the two-dimensional code of the twisting machine and the barcode of the cake, and the information is bound;

[0075] Step 14, after twisting is completed, the yarn is transferred to the packaging area for inspection and packing; through the manipulator in the packaging area, identify and grab the tube yarn with a two-dimensional code; the two-dimensional code scanning device in the packaging area scans and confirms the tube yarn to be packed, so that each operation is traceable and the whole process of yarn production is managed.

[0076] In the embodiment of the present invention, in Step 11, automatically identifying and numbering the spools and cakes on the wire drawing machine through a camera and a machine vision system improves the accuracy and efficiency of identification and avoids human errors. Binding the variety produced by each wire drawing machine to the machine vision system ensures the consistency and traceability of production data. The visible barcode printed on the cake as the unique identifier facilitates accurate tracking and management of the cake in subsequent links.

[0077] In Step 12, installing cameras in the transfer channel and using machine vision technology to identify and record the barcode information on the cakes realizes continuous tracking of the cakes during the transfer process, avoids loss or interruption of tracking information, and this non-sense tracking method reduces manual intervention and improves transfer efficiency and accuracy.

[0078] Step 13: Each twister is equipped with a unique machine number, which is displayed in the form of a QR code, facilitating the quick identification and recording of information. A scanning device is used to scan the QR code of the twister and the bar code of the yarn cake, realizing the automatic binding of information and ensuring the integrity and consistency of production data. This binding method helps to accurately trace and locate problems in the production process subsequently.

[0079] Step 14: The manipulator in the packaging area identifies and grabs the tube yarn with a QR code, improving the degree of automation of packaging and reducing the complexity and errors of manual operations. The QR code scanning device in the packaging area scans and confirms the tube yarn to be packed, ensuring the accuracy and traceability of each operation. The whole process of yarn production is managed in an intelligent way, improving the efficiency and level of production management, helping the enterprise to discover and solve problems in a timely manner, and enhancing product quality and customer satisfaction.

[0080] In a preferred embodiment of the present invention, in the above step 11, the automatic identification and numbering of the yarn cones and yarn cakes on the wire drawing machine by the camera and the machine vision system may include:

[0081] Step 111: The camera is installed at a preset position on the wire drawing machine to capture the images of the yarn cones and yarn cakes on the wire drawing machine in real time through the camera.

[0082] Step 112: Preprocess the captured image and perform image segmentation to separate the spool and the bobbin from the background. Specifically, it includes: obtaining a frame of the image to be processed from the real-time video stream captured by the camera; if the captured image is a color image (such as in RGB format), convert it to a color space more suitable for image segmentation, such as the HSV (hue, saturation, value) color space, as the HSV color space is more conducive to image segmentation based on color information; setting one or more thresholds according to the characteristics of the spool and the bobbin, such as color and brightness, where these thresholds are used to distinguish the spool, the bobbin, and the background; applying threshold operations to convert the image into a binary image. In the binary image, pixels above the threshold are set to one value (such as white), and pixels below the threshold are set to another value (such as black). In this way, the spool and the bobbin are clearly distinguished from the background; perform morphological operations on the binary image, such as erosion, dilation, opening operation, or closing operation. These operations can remove noise, smooth edges, or fill holes, thereby improving the effect of image segmentation; use contour detection algorithms (such as Canny edge detection or the FindContours function) to identify the continuous edges in the binary image, where these edges correspond to the contours of the spool and the bobbin; filter out the contours representing the spool and the bobbin according to the size, shape, or other characteristics of the contours. For example, the minimum and maximum values of the contour area can be set to exclude overly small or large contours; generate a mask based on the filtered contours. The mask is a binary image with the same size as the original image, where the positions of the spool and the bobbin are marked as white, and the remaining positions are black; perform a pixel-by-pixel multiplication operation using the mask and the original image to extract the images of the spool and the bobbin, thus achieving image segmentation;

[0083] Step 113: Obtain a frame of color image from the camera and calculate the gray value for each pixel in the color image. Specifically, it includes: obtaining a frame of color image from the camera device, which is accomplished by calling the camera API or using an image processing library (such as OpenCV); using a double loop to traverse each pixel of the image. The outer loop traverses the rows (height) of the image, and the inner loop traverses the columns (width) of the image; for each pixel, calculate the gray value according to its RGB values. Grayscale conversion is the process of converting a color image into a grayscale image, where the gray value represents the brightness information of the pixel. For example, the weighted average method is adopted, that is, different weights are assigned to the R, G, and B components according to the sensitivity of the human eye to different colors, and then weighted averaging is performed; assign the calculated gray value to the pixel, which means creating a new single-channel image (grayscale image) or modifying the original image in place to convert it into a grayscale image;

[0084] Step 114: Generate a grayscale image based on the gray values, and calculate the new value of each pixel through convolution operation using the Gaussian function to obtain the image after Gaussian filtering;

[0085] Step 115: Respectively use the horizontal and vertical directions of the Sobel operator to perform a convolution operation on the image after Gaussian filtering to obtain the gradient components of each pixel point in the horizontal and vertical directions;

[0086] Step 116: For each pixel point, calculate the gradient magnitude based on the gradient components of each pixel point in the horizontal and vertical directions to obtain a gradient image containing the gradient magnitude and direction of each pixel point, specifically including: creating a new image with the same size as the original grayscale image to store the gradient magnitude and direction of each pixel point. This new image is a two-channel image, where one channel is used to store the gradient magnitude and the other channel is used to store the gradient direction; using a double loop to traverse each pixel of the grayscale image. The outer loop traverses the rows (height) of the image, and the inner loop traverses the columns (width) of the image; for the currently traversed pixel point, calculate the gradient magnitude and direction based on its gradient components in the horizontal and vertical directions. The horizontal and vertical gradient components are obtained by performing a convolution operation on the image with the Sobel operator. Calculate the gradient magnitude and gradient direction based on the horizontal and vertical gradient components;

[0087] Step 117: Traverse the gradient image of each pixel point's gradient magnitude and direction, identify and extract the continuous edge contours in the image, specifically including: preparing an empty contour list to store the identified edge contours; setting a gradient magnitude threshold to determine whether a pixel point belongs to an edge; using a double loop to traverse each pixel point of the gradient image and check its gradient magnitude. If the gradient magnitude of the current pixel point is greater than the set threshold, then this point is considered an edge point. For the identified edge points, use an edge tracing algorithm (such as the hysteresis threshold method and edge tracing in Canny edge detection) to connect adjacent edge points to form continuous edge contours. A queue or stack data structure can be used to assist the tracing process, continuously adding adjacent edge points to the contour until no more adjacent edge points can be found. After each tracing is completed, add the formed contour to the contour list. After the traversal is completed, the obtained contour list contains all the identified continuous edge contours in the image;

[0088] Step 118: For each extracted contour, calculate the number of pixel points on the contour, the radius of curvature or the curvature angle of each point on the contour, and the overall direction of the contour or the direction of each segment. Specifically, it includes: For each contour extracted in Step 117, calculate its features; traverse each point on the contour, accumulate the number of points to obtain the length of the contour; for each point on the contour, calculate the curvature based on its positional relationship with adjacent points. The three-point method (the current point and its adjacent points before and after) can be used to estimate the radius of curvature or the curvature angle of the point. The radius of curvature can be obtained by calculating the radius of the circle formed by the three points, and the curvature angle can be the angle between adjacent line segments. Calculate the overall direction of the contour or the direction of each segment. The principal direction of the contour can be obtained by calculating the covariance matrix of the coordinates of all points on the contour; for the direction of each segment, the contour can be segmented into several segments, and the direction of the line connecting the starting point and the ending point of each segment can be used as the direction of the segment.

[0089] Step 119: According to the number of pixel points on the contour, the radius of curvature or the curvature angle of each point on the contour, and the overall direction of the contour or the direction of each segment, screen out the edge features of the spool and the bobbin; based on the edge features of the spool and the bobbin, identify the spool and the bobbin in the image, and determine the positions and postures of the spool and the bobbin; assign a unique number to each identified spool and bobbin, and associate the number with the image, position, and time of the spool and the bobbin. Specifically, it includes:

[0090] According to prior knowledge or experimental data, set appropriate thresholds for the edge features of the spool and the bobbin. These features include the number of pixel points on the contour, the radius of curvature or the curvature angle, the overall direction or the direction of each segment, etc. Traverse all the contour features calculated in Step 118. For each contour, check whether its feature values meet the threshold conditions of the spool or the bobbin; for example, the spool may have a large number of pixel points and a specific curvature range, while the bobbin may have a different set of features. Classify the contours that meet the spool feature thresholds as spool candidates, and classify the contours that meet the bobbin feature thresholds as bobbin candidates.

[0091] For each candidate contour of the spool and the bobbin, use a shape matching algorithm (such as contour-based matching, template matching, etc.) to further verify its identity. The matching can be performed by comparing the similarity between the candidate contour and the preset spool or bobbin shape template. For the successfully matched spool and bobbin contours, calculate their centroids or geometric centers as the positions of the spool or the bobbin, and determine the postures of the spool or the bobbin according to the overall direction or the principal axis direction of the contour.

[0092] Assign a globally unique number to each identified spool and cake. This can be achieved through an incrementing counter, UUID, or other unique identifier generation methods; associate the assigned number with the image data of the spool or cake, the calculated position and pose information, and the current timestamp, and a data structure (such as an object, dictionary, or database record) can be created to store this information.

[0093] In the embodiment of the present invention, through steps such as image segmentation, grayscale processing, Gaussian filtering, and Sobel operator gradient calculation, the images of the spool and cake can be more accurately separated from the complex background, reducing the cases of misidentification and missed identification, thereby improving the accuracy of identification. Gaussian filtering can effectively remove the noise in the image, and the Sobel operator can highlight the edge information in the image. These processing steps enhance the anti-interference ability of the system in a complex environment and ensure the stability of identification. By optimizing the image processing algorithm, such as using efficient convolution operations and contour extraction methods, the speed of image processing can be accelerated, thereby realizing the rapid identification and numbering of the spool and cake and meeting the requirements of real-time production. Automatic identification and numbering reduce the need for manual intervention, realize the automated management of the spool and cake, reduce labor costs, and improve production efficiency and management level at the same time. Associating the identified spool and cake with a unique number ensures the consistency and traceability of production data. Accurate identification and numbering make the production process smoother, reduce production interruptions or rework caused by misidentification, further optimize the production process, and improve the overall production efficiency.

[0094] In a preferred embodiment of the present invention, step 111, installing the camera at a preset position on the wire drawing machine to capture the images of the spool and cake on the wire drawing machine in real time by the camera, includes:

[0095] Step 1111: Determine the set of all potential location points where the camera can be installed. The location points are fixed points around the wire drawing machine. Define the attributes of each location point, including coordinates and viewing range. Specifically, it includes the following steps: Through visual inspection, initially identify all possible fixed installation points around the wire drawing machine, such as walls, brackets, crossbeams, etc. Exclude those locations that are obviously not suitable for installation, such as high-temperature areas, near moving parts, or places where the space is narrow and difficult to operate. Use professional measuring tools (such as laser rangefinders, three-dimensional coordinate measuring instruments) to accurately measure the coordinates of each potential location point. Record the three-dimensional coordinates (X, Y, Z) of each location point, ensuring that the coordinate system is consistent with the working space of the wire drawing machine. Place a temporary marker or a simulated camera at each potential location point; use a camera, a mobile phone camera, or other imaging devices to take photos or videos of the wire spools and cakes on the wire drawing machine at each marked location. Analyze these image data to evaluate the viewing range that each location point can provide. Pay special attention to the presence of obstacles, reflections, or blind spots in the field of view. According to the evaluation results, define a viewing range attribute for each location point, which can be a specific angle value, a percentage of the field of view coverage, or a descriptive label (such as "full view", "partially blocked", etc.). Create a database or spreadsheet to store the relevant information of each potential location point;

[0096] Step 1112: Randomly select a part of all potential location points as the initial positions of the ants; allocate a memory table for each ant to record the locations it has visited and the corresponding pheromone concentrations. Specifically, it includes: From the set of potential location points determined in Step 1111, use a random number generator to randomly select a part of the location points as the initial positions of the ants. Create a memory table for each ant. This table will be used to store the location points visited by the ant during the search process and the corresponding pheromone concentrations. The memory table can be implemented using data structures such as arrays, lists, or hash tables for quick lookup and update; At initialization, the memory table can be empty, or a preset initial pheromone concentration value (usually a small positive number) can be set for each location point; Associate the initialized memory table with the corresponding ant, which can be achieved by maintaining references to the ant objects and their respective memory tables in the data structure of the algorithm. After completing the above initialization steps, the ants are ready to start their search process;

[0097] Step 1113: Determine the initial concentration and evaporation rate of pheromone. Each ant selects the next position to move to based on the pheromone concentration and heuristic information around its current position. Specifically, it includes: Determine an initial pheromone concentration value, which will serve as the starting concentration of pheromone on all paths. Set an evaporation rate of pheromone. The evaporation rate determines the speed at which pheromone decreases over time, simulating the natural dissipation process of pheromone in the real world. For each ant, based on its current position, retrieve the pheromone concentration and heuristic information (such as the straight-line distance to the target point, field of view quality, etc.) of the reachable position points around it. Combine the pheromone concentration and heuristic information, and use a probability formula to calculate the probability of the ant moving to each reachable position point. The position points with higher pheromone concentration and better heuristic information will have a higher probability of being selected. According to the calculated probability, use a random selection mechanism (such as roulette wheel selection) to determine the position point that the ant will move to next. Among them, the calculation formula of the probability is:

[0098] ;

[0099] Among them, represents the probability that the ant moves from position to position . The higher this probability value, the greater the likelihood that the ant selects this position; represents the pheromone concentration from position to position . The pheromone concentration usually represents the quality of the path. The higher the concentration, the more popular this path is. represents the heuristic information of position , which is the reciprocal of the distance to the target point. The closer to the target, the higher the heuristic information; represents the pheromone concentration from position to position ; represents the heuristic information of position , which is the reciprocal of the distance to the target point. represents the weight of pheromone; represents the weight of heuristic information. represents the set of all reachable positions of the current ant position ; represents the current position or node where the ant is located. The ant evaluates the pheromone concentration and heuristic information of the reachable positions around this position. represents the target position or node that the ant considers moving to. This position is one of the reachable positions from the current position . The ant calculates the probability of moving to . A general index representing all reachable positions, used to represent all positions reachable from the current position All reachable positions.

[0100] Update the memory table of the ants, recording the newly visited position points and the pheromone concentrations at these position points. After all ants complete a move, according to the set evaporation rate, reduce the pheromone concentrations on all paths. For the paths visited by ants, a certain amount of pheromone can be increased to simulate the behavior of ants releasing pheromones. The increased amount can be adjusted according to the quality of the path (such as length, field of view, etc.);

[0101] Step 1114, after a certain number of iteration times, according to the memory tables of all ants and the path evaluation results, determine the final position point, specifically including: Repeat Step 1113 to let the ant colony perform multiple iterative searches among potential position points. Each iteration will update the pheromone concentration and the memory table of the ants; Set a maximum number of iteration times as the termination condition of the algorithm. This number should be sufficient for the ant colony to explore most of the potential position points and find a better solution. After reaching the maximum number of iteration times, stop the search process. At this time, the memory table of each ant records a complete search path and its corresponding pheromone concentration. According to the preset evaluation criteria (such as the total length of the path), evaluate the paths of all ants. Select the path with the best evaluation result, and the end position of this path is the final position point for installing the camera. If there are multiple similar excellent paths, other decision-making methods (such as voting, weighted average, etc.) can be further used to determine the final position;

[0102] Step 1115, install the camera according to the final position point, and capture the images of the bobbin and cake on the wire drawing machine in real time through the camera, specifically including: According to the final position point determined in Step 1114, prepare the corresponding installation materials and tools (such as brackets, screws, cameras, etc.); Install the camera bracket at the final position point and ensure its stability and reliability. Fix the camera on the bracket and adjust its direction and angle to ensure that the images of the bobbin and cake on the wire drawing machine can be clearly captured; Connect the power supply line and data transmission line of the camera, and perform necessary debugging and testing to ensure that the camera works properly; Connect the camera to the monitoring system and start capturing and transmitting image data in real time; According to the actual monitoring effect, adjust the parameters of the camera (such as focal length, exposure time, etc.) to optimize the image quality; Set the alarm and notification functions of the monitoring system so that abnormal situations can be responded to and processed in a timely manner.

[0103] In the embodiments of the present invention, by comprehensively considering the coordinates and viewing range of potential location points, and by iteratively searching for the optimal solution through algorithms, it can be ensured that the selected location can provide the best visual coverage of the wire spools and cakes on the wire drawing machine. This helps to reduce the visual blind spots and improve the comprehensiveness and accuracy of monitoring. A suitable installation location can ensure that the camera captures clear and stable images, reducing the degradation of image quality caused by poor angles or occlusions. High-quality images are crucial for subsequent image processing, analysis, and recognition. By intelligently selecting the installation location through algorithms, unnecessary trial and error and manual adjustments can be avoided, thus saving installation time and costs. In addition, a reasonable location selection can also reduce the possibility of readjusting or replacing the camera in the future due to poor viewing angles. The location selection method based on the heuristic optimization algorithm can find the optimal solution among multiple potential locations, which means that even when some locations become unavailable due to environmental factors (such as light changes, equipment failures, etc.), the system can still quickly adjust to other suitable locations to maintain the continuity of monitoring. The entire location selection process is automatically completed by the algorithm, reducing the need for human intervention and improving the automation level. This helps to achieve the automated monitoring and management of the wire drawing machine production process and improve production efficiency.

[0104] In a preferred embodiment of the present invention, step 114, generating a grayscale image according to the grayscale values, and calculating the new value of each pixel through convolution operation by the Gaussian function to obtain the image after Gaussian filtering, may include:

[0105] Step 1141, calculating the grayscale values of all pixels to create a new image, where the value of each pixel is the grayscale value. Here, the new image is a grayscale image, which specifically includes: having a color image as the input. This image is usually composed of three color channels: red, green, and blue (RGB). Each channel corresponds to a two-dimensional matrix, and each value in the matrix represents the brightness of that color channel. For each pixel in the color image, its grayscale value needs to be calculated. The grayscale value is calculated by adding the values of the RGB three channels according to a certain ratio (such as linear weighting); after calculating the grayscale values of all pixels, these values will be used to create a new two-dimensional matrix, that is, the grayscale image. In this new image, the value of each pixel is its grayscale value, and the range is usually 0 to 255 (for 8-bit images);

[0106] Step 1142, defining a Gaussian kernel. The Gaussian kernel is a two-dimensional matrix, and its values are calculated according to the Gaussian function, which specifically includes: determining the size of the Gaussian kernel, which is a matrix of odd number multiplied by odd number (such as 3×3, 5×5, etc.) to ensure there is a center point. Each value of the Gaussian kernel is calculated according to the two-dimensional Gaussian function. The calculated Gaussian function values are filled into the two-dimensional matrix to form the Gaussian kernel. The values of the Gaussian kernel will be normalized so that the sum of all values is 1 to ensure that the brightness of the filtered image remains unchanged;

[0107] Step 1143, starting from the upper left corner of the image, select the first pixel, and multiply the first pixel and the pixels around it by the values at the corresponding positions of the Gaussian kernel; align the center of the Gaussian kernel with the selected pixel, and perform the following operations, specifically including: starting from the upper left corner of the input image (here it is a grayscale image), select the first pixel as the currently processed pixel; align the center of the Gaussian kernel with the selected pixel. If the Gaussian kernel is larger than the image, boundary processing (such as padding, mirroring, etc.) needs to be performed on the image; multiply each value in the Gaussian kernel by the pixel value at the corresponding position in the image, and then add up all the products. This process is the convolution operation, which is actually a weighted sum of the selected pixel and the pixels around it. The result of the convolution operation is the new pixel value at the corresponding position in the filtered image, and this value will replace the pixel value at the corresponding position in the original image. After processing the current pixel, move the Gaussian kernel one pixel position to the right (or down as needed), and then repeat the above convolution operation until all pixels in the image are processed:

[0108] Step 1144, multiply the upper left corner value of the Gaussian kernel by the pixel value at the corresponding position in the image; multiply the upper middle value of the Gaussian kernel by the pixel value at the corresponding position in the image until all values of the Gaussian kernel are multiplied by the pixel values at the corresponding positions in the image, and add up all the products. The result obtained is the new pixel value at the corresponding position in the filtered image, specifically including: determine the pixel position in the image corresponding to the upper left corner of the current position of the Gaussian kernel; starting from the upper left corner of the Gaussian kernel, multiply the value (weight) of the Gaussian kernel at this position by the pixel value at the corresponding position in the image; then, multiply the value at the upper middle position of the Gaussian kernel by the pixel value at the corresponding position in the image, where the value at the upper middle position refers to the value at the middle position of the upper half of the Gaussian kernel; and so on, from left to right, from top to bottom, multiply each value in the Gaussian kernel by the pixel value at the corresponding position in the image one by one; add up all the products obtained, and this sum is the new pixel value at the corresponding position in the filtered image corresponding to the center of the current Gaussian kernel. The new pixel value reflects the weighted average of the position and the pixels around it in the original image, and the weight is determined by the Gaussian kernel;

[0109] Step 1145, move the Gaussian kernel one pixel position to the right until the new values of all pixels in the image are calculated, and create a new image using the calculated new pixel values. The new image is the image after Gaussian filtering, specifically including:

[0110] After completing the Gaussian filtering calculation at the current position, move the Gaussian kernel one pixel to the right; if the Gaussian kernel has reached the end of the image row, move it to the starting position of the next row; at the new position of the Gaussian kernel, repeat the convolution calculation process in step 1144 to obtain the filtered pixel value at the new position; by continuously moving the Gaussian kernel and repeating the convolution calculation until each pixel position in the image has been processed; use all the calculated new pixel values to create a new two-dimensional matrix, that is, the filtered image, in this new image, the pixel value at each position is the value after Gaussian filtering.

[0111] In the embodiment of the present invention, through the convolution operation, it can average the high-frequency noise (such as salt-and-pepper noise, Gaussian noise, etc.) in the image, thereby reducing the impact of noise on the image quality. Gaussian filtering can better retain the edges and details of the image while smoothing the image. This is because the Gaussian function has the characteristics of isotropy and gradually decreasing smoothness, so that the filter can assign different weights according to the distance from the central pixel when processing pixels, thus more effectively protecting the important features of the image. By defining the Gaussian kernel and calculating its value at one time, this kernel can be reused throughout the filtering process, thereby improving the calculation efficiency. In addition, since Gaussian filtering is a linear filtering method, its calculation process is relatively simple and easy to parallelize, further accelerating the processing speed. Grayscale images are the most commonly used image format in the field of image processing. It simplifies the representation form of the image and reduces the data dimension, enabling many algorithms to process and analyze the image more efficiently.

[0112] In a preferred embodiment of the present invention, in step 115 above, the horizontal and vertical direction kernels of the Sobel operator are respectively used to perform convolution operations on the image after Gaussian filtering to obtain the gradient components of each pixel point in the horizontal and vertical directions, including:

[0113] Step 1151, starting from the upper left corner of the image, select a pixel point as the current processing point; cover the horizontal kernel of Sobel on the current pixel point and its surrounding pixels, and perform the convolution operation: multiply the upper left element of the horizontal kernel by the corresponding image pixel value; multiply the upper middle element of the horizontal kernel by the corresponding image pixel value, and repeat this process until all elements of the horizontal kernel are multiplied by the corresponding image pixels; add up all the products, and the result is the gradient component of the current pixel point in the horizontal direction, specifically including the following steps:

[0114] Starting from the upper left corner of the image, select the first pixel as the current processing point. Determine the size of the Sobel horizontal kernel, usually 3×3. Ensure that there are enough pixels around the current processing point for the kernel to cover (i.e., avoid pixels outside the boundary). Align the center of the Sobel horizontal kernel with the current processing point. Ensure that the rest of the kernel covers the surrounding pixels of the current processing point. Perform the convolution operation:

[0115] Starting from the upper left element of the Sobel horizontal kernel, multiply its value by the value of the corresponding image pixel; then, multiply the value of the upper middle element of the horizontal kernel by the image pixel it covers, and continue this process, from left to right, from top to bottom, multiplying each element in the kernel by the corresponding image pixel one by one. Sum to obtain the gradient component:

[0116] Add up all the products obtained. This sum is the gradient component of the current pixel point in the horizontal direction. Record or store the calculated gradient component, move the current processing point one pixel to the right; if the end of the image row is reached, move the processing point to the starting position of the next row. Repeat the steps until the gradient components of all pixels in the image in the horizontal direction are calculated.

[0117] Step 1152, perform the convolution operation on the same pixel point and its surrounding pixels using the Sobel vertical kernel: multiply the upper left element of the vertical kernel by the value of the corresponding image pixel; multiply the upper middle element of the vertical kernel by the corresponding image pixel value, and repeat this process until all elements of the vertical kernel are multiplied by the corresponding image pixels, add up all the products, and the result is the gradient component of the current pixel point in the vertical direction; move the current processing point one pixel to the right until the gradient components of all pixels in the image in the horizontal and vertical directions are calculated, specifically including the following steps:

[0118] Continue to use the current processing point selected in step 1151 as the starting point; ensure that there are enough pixels around the current processing point for the kernel to cover. Perform the convolution operation:

[0119] Align the center of the Sobel vertical kernel with the current processing point; starting from the upper left element of the Sobel vertical kernel, multiply its value by the value of the corresponding image pixel; then, multiply the value of the upper middle element of the vertical kernel by the image pixel it covers, and continue this process, multiplying each element in the kernel by the image pixel value it covers. Calculate the vertical gradient component:

[0120] Add up all the products obtained. This sum is the gradient component of the current pixel point in the vertical direction. Record or store the calculated vertical gradient component, move the current processing point one pixel to the right, and if the processing point reaches the end of the image row, move it to the starting position of the next row. Repeat the steps until the gradient components of all pixels in the image in the vertical direction are calculated.

[0121] Step 1153: Generate a gradient component image based on the gradient components of all pixels in the image in the horizontal and vertical directions, which specifically includes the following steps:

[0122] Steps 1151 and 1152, that is, the gradient components of all pixels in the image in the horizontal and vertical directions have been calculated, and these gradient components are stored in numerical form. Each pixel point corresponds to a pair of horizontal gradient components (Gx) and vertical gradient components (Gy). Create a gradient component image:

[0123] Initialize a new image with the same size as the original image to store and display the gradient components. This new image can be a grayscale image, where the brightness or color of each pixel represents the gradient intensity or direction at that position. Calculate the gradient magnitude:

[0124] For each pixel point, the horizontal gradient component (Gx) and vertical gradient component (Gy) can be used to calculate the gradient magnitude. Assign to the gradient component image:

[0125] If the gradient magnitude is calculated, the gradient magnitude value of each pixel point can be assigned to the corresponding position in the new image. Another method is to use the horizontal gradient component (Gx) and vertical gradient component (Gy) as the two channels of the new image (for example, the red and green channels), so that the direction and intensity of the gradient can be represented simultaneously by color. In this case, the new image may be a color image, where the hue of the color represents the direction of the gradient, and the brightness or saturation of the color represents the intensity of the gradient.

[0126] Since the numerical range of the gradient components may be very large, it may be necessary to normalize or scale them to better highlight the edge information when displaying. Normalization can map the gradient component values to the range of 0 to 255 (for 8-bit grayscale images), or map them to a color space suitable for display (for color images). Finally, an image processing library or software can be used to display the generated gradient component image.

[0127] In an embodiment of the present invention, the Sobel operator is an effective edge detection algorithm that detects edges by calculating the first or second derivative of the image grayscale. In step 115, by applying the Sobel operator to the Gaussian-filtered image, the edge information of the image can be detected more accurately. Gaussian filtering can smooth the image and remove noise, and the Sobel operator further highlights the edge features on this basis. Steps 1151 and 1152 calculate the gradient components in the horizontal and vertical directions respectively. By first performing Gaussian filtering on the image and then applying the Sobel operator, the robustness of edge detection can be improved. Gaussian filtering can reduce noise interference in the image, making the Sobel operator more stable during processing and less susceptible to noise. The convolution operation in step 115 is performed on the entire image by means of a sliding window. This processing method can utilize the optimized algorithms and parallel computing capabilities in the computer vision library, thereby improving the processing speed and achieving efficient edge detection.

[0128] In a preferred embodiment of the present invention, in the above step 11, the variety produced by each drawing machine is bound to the machine vision system. A visible bar code is printed on the silk cake as the unique identifier of the silk cake, corresponding to the number recognized by the machine vision system. It may further include:

[0129] Step 1190, assign a unique number to the variety produced by each drawing machine;

[0130] Step 1191, print the assigned number in the form of a bar code on the silk cake as the unique identifier of the silk cake, specifically including: ensuring that each variety produced by the drawing machine has been assigned a unique number. This number can be generated sequentially or according to a certain specific rule. The key is to ensure its uniqueness. Use professional bar code generation software or library to convert the assigned number into a bar code image. The type of bar code (such as EAN-13, Code128, etc.) should be selected according to the actual application requirements and printing conditions. Prepare the printer for printing the bar code and the corresponding printing materials (such as label paper, carbon ribbon, etc.). Ensure that the printer can clearly print the bar code for subsequent scanning and recognition. At an appropriate link in the silk cake production process, print the generated bar code on the silk cake or stick it on the label of the silk cake. The printing position should be convenient for subsequent scanning operations and will not be damaged due to the handling or storage of the silk cake;

[0131] Step 1192, establish a database or lookup table to store the correspondence between varieties and numbers, specifically including: Select a suitable database system (such as MySQL, SQL Server, etc.) according to requirements, and design a database table containing variety information and corresponding numbers. The table should at least include fields such as variety name (or description), number, etc. Enter the information of each variety and its corresponding number into the database. Ensure the accuracy and integrity of the data. To improve query efficiency, an index can be set for the number field. In this way, when looking up variety information according to the number later, the relevant data can be located more quickly. Establish a mechanism for regularly backing up the database and prepare a corresponding data recovery plan to prevent data loss or damage;

[0132] Step 1193, use image processing technology to preprocess the collected images, scan and analyze the preprocessed images, and extract the number information on the cake, specifically including: Use devices such as cameras or scanners to collect images containing the cake barcode. Preprocess the collected images, including operations such as denoising, binarization, and contrast enhancement, to improve the recognizability of the barcode. Locate the barcode area in the image through image processing algorithms (such as edge detection, morphological transformation, etc.). Use a barcode recognition library or algorithm to scan and analyze the located barcode and extract the number information therein. Ensure the accuracy of the parsing result;

[0133] Step 1194, according to the extracted number information, look up the corresponding variety information and bind the variety information and number information, specifically including: Extract the number information on the cake from the previous step. Use the extracted number information as a query condition to look up the corresponding variety information in the database. Ensure the accuracy and efficiency of the query operation. Bind the queried variety information with the extracted number information. This can be achieved by creating data structures (such as dictionaries, objects, etc.) in memory, or directly storing the binding result back to the database or file.

[0134] In the embodiments of the present invention, by automatically identifying the barcodes on the silk cakes through a machine vision system, the variety information of the silk cakes can be quickly and accurately obtained, avoiding the errors and delays in traditional manual identification and recording, thereby improving production efficiency. The barcode on each silk cake serves as its unique identifier, enabling the information of each link in the production process to be recorded and traced. This helps to quickly locate the cause in case of problems and take timely measures to improve the product quality control level. The database or lookup table stores the correspondence between varieties and numbers, making inventory management more convenient and efficient. By querying the database, the inventory status of silk cakes of various varieties can be quickly understood. The entire process realizes the automated collection, storage, and processing of data, enhancing the informatization level of the enterprise. The introduction of automation and informatization reduces labor costs, avoids losses and wastes caused by human factors, and thus reduces the overall operating costs of the enterprise.

[0135] In a preferred embodiment of the present invention, step 1190, which assigns a unique number to each variety produced by the drawing machines, includes:

[0136] Step 11901, determining the basic rules of the number, including the length of the number and the character set used. Specifically, it includes setting a suitable length for the number. For example, if the number of varieties to be represented is in the thousands to tens of thousands range, a moderately long number, such as 6 - 8 digits, can be selected. According to actual needs, select a suitable character set. Common character sets include numbers (0 - 9), uppercase letters (A - Z), lowercase letters (a - z), or their combinations. If readability and input convenience are to be considered, a preference may be given to using pure numbers or a combination of numbers and uppercase letters;

[0137] Step 11902, obtaining the variety information produced by all drawing machines, including variety names, specifications, and characteristics. Specifically, it includes determining the data source storing the variety information of the drawing machines; according to the type of the data source, using appropriate query statements or API interfaces to extract the variety information produced by all drawing machines, including keyword fields such as variety names, specifications, and characteristics. Clean and organize the extracted data to ensure the accuracy and integrity of the data. This includes removing duplicate records, handling missing values, correcting incorrect data, etc. Store the cleaned and organized data in a data structure convenient for subsequent processing, such as a list, data frame, or database table;

[0138] Step 11903: Represent the number of each variety as a gene code. According to the defined gene code, randomly generate an initial population, where each individual in the population represents a number assignment scheme, specifically including: Define a suitable gene coding scheme according to the numbering rules determined in Step 11901. For example, if the number length is 8 digits and only numerical characters are used, the code of each gene can be a number between 0 and 9. Set a suitable initial population size, which will determine the search space and convergence speed of the algorithm; Use a random number generator to randomly generate an initial population according to the defined gene coding scheme. Each individual (i.e., each number assignment scheme) should meet the requirements of the number length and character set. Represent the generated initial population as a suitable data structure, such as a list, array, or matrix, and store it in memory for subsequent processing. Ensure that each individual (number assignment scheme) is associated with its corresponding variety information;

[0139] Step 11904: Determine the fitness function used to evaluate the quality of each individual. Among them, the specific calculation formula corresponding to the fitness function is:

[0140] ;

[0141] Among them, represents the number of number conflicts, indicating how many numbers are repeated in the individual ; represents the number of numbers with lengths not meeting expectations, indicating how many numbers in the individual have lengths not equal to the expected value;

[0142] Step 11905: According to the fitness function, select the corresponding individuals from the current population to enter the next generation, perform crossover operations on the selected individuals to generate new individuals, and perform mutation operations on the newly generated individuals until the preset number of iterations is reached to obtain the final number assignment scheme, specifically including:

[0143] Using the roulette wheel selection mechanism, select a part of the individuals from the current population to enter the next generation according to the individual fitness values calculated by the fitness function. During the selection process, individuals with higher fitness should have a greater probability of being selected. For the selected individuals, randomly pair them and perform crossover operations to generate new individuals; the crossover operations can adopt methods such as single-point crossover, multi-point crossover, or uniform crossover. Select the crossover points in the gene sequences of the paired individuals according to the defined gene encoding scheme and exchange some gene segments. Perform mutation operations on the newly generated individuals with a certain mutation probability to increase the diversity of the population. The mutation operations can include random replacement, insertion, or deletion of gene loci to ensure that the mutated individuals still meet the basic rules of the numbering. Repeat the above selection, crossover, and mutation operations to form a new generation of the population; evaluate the fitness of the new generation of the population and record the best individual (i.e., the numbering allocation scheme with the highest fitness); determine whether the preset number of iterations is reached or other termination conditions are met (such as fitness threshold, reduction of population diversity, etc.). If satisfied, stop the iteration; otherwise, continue the evolution of the next generation; after the iteration ends, output the best individual as the final numbering allocation scheme;

[0144] Step 11906, according to the final numbering allocation scheme, assign a unique number to each variety produced by the drawing machines, specifically including: obtaining the final numbering allocation scheme from the output of Step 11905; according to the final numbering allocation scheme, match each number with the corresponding variety produced by the drawing machines; ensure that each variety is assigned one and only one unique number.

[0145] In the embodiments of the present invention, by determining the basic rules of the numbering, such as the length and character set, and using the gene encoding method, it can be ensured that the numbers assigned to each variety are unique, avoiding the problems of numbering conflicts or duplications. At the same time, this method makes the numbering system more standardized and systematic, facilitating management and maintenance. The gene encoding method allows various factors to be considered during the numbering allocation process, such as variety names, specifications, and characteristics. This flexibility enables the system to easily adapt to the addition or change of different varieties while maintaining the consistency and scalability of the numbering system. By evaluating the numbering allocation scheme through the fitness function, better individuals can be selected to enter the next generation, thereby gradually optimizing the numbering allocation scheme. This optimization process not only improves the efficiency of numbering allocation but also helps to reasonably allocate resources, such as ensuring that important or frequently occurring varieties have more easily recognizable and memorable numbers. The automated numbering allocation method based on gene encoding reduces human participation and intervention, thereby reducing errors and delays caused by human factors. This improves the efficiency and accuracy of the production process while reducing the burden on the staff.

[0146] In another preferred embodiment of the present invention, in step 12 above, after wire drawing is completed, the wire cake is sent to the twisting area through a transfer channel; a camera is installed in the transfer channel, and machine vision technology is used to identify and record the barcode information on the passing wire cake, so that the tracking of the wire cake during transfer is not interrupted, which may include:

[0147] Processing of the wire cake after wire drawing: Once the wire drawing process is completed, the wire cake will be removed from the wire drawing machine; workers or automated equipment will place the wire cake on the transfer channel and prepare to send it to the twisting area; the transfer channel is designed as an automated conveying system with a conveyor belt or rollers to ensure the smooth and continuous movement of the wire cake; high-resolution cameras are installed at key positions in the transfer channel, such as the entrance, exit, and turning points; an integrated machine vision system that can capture images of the wire cake on the conveyor belt in real time; the machine vision system automatically identifies the barcode on the wire cake and records relevant information, such as the wire cake number, production time, etc.; the identified data is recorded and transmitted to the central management system in real time to ensure that the tracking information of the wire cake during transfer is not interrupted.

[0148] In step 13, in the twisting area, each twisting machine is equipped with a unique machine number and is displayed in the form of a two-dimensional code; when the wire cake arrives at the twisting area, a scanning device is used to scan the two-dimensional code of the twisting machine and the barcode of the wire cake, and the information is bound, which may include:

[0149] Generation of the twisting machine number and two-dimensional code: Assign a unique machine number to each twisting machine, and use a two-dimensional code generation tool to convert these numbers into two-dimensional code format; paste the generated two-dimensional code labels at prominent positions on each twisting machine for easy scanning and identification; after the wire cake arrives at the twisting area through the transfer channel, the worker prepares to place it on the corresponding twisting machine; the scanning device held by the worker or fixed in position is ready to scan the two-dimensional code and barcode; the worker first scans the two-dimensional code of the twisting machine, and then scans the barcode of the wire cake, and the system binds these two pieces of information together to record which wire cake is being processed on which twisting machine; the bound information is recorded and updated to the central management system in real time to ensure the transparency and traceability of the production process.

[0150] Specific implementation process of step 14:

[0151] After the twisting process is completed, the operator removes the yarn (already in the form of a tube yarn at this time) from the twisting machine and places it on a special transfer cart or conveyor belt. The transfer cart or conveyor belt transports the yarn smoothly to the packaging area to ensure that the yarn is not damaged during transfer.

[0152] In the packaging area, high-precision robotic arms are configured. These robotic arms are equipped with vision recognition systems and can accurately identify the QR codes on the cheese. The robotic arms scan the incoming cheeses through the vision system, quickly locate and identify the cheeses with QR codes. After successful identification, the robotic arms precisely grasp the cheeses according to the preset programs and algorithms and place them at the designated inspection or packing positions.

[0153] The grasped cheeses first go through the quality inspection process. The inspectors check the key indicators such as the quality, appearance, and twist of the yarn. The inspection results are recorded in the information management system and associated with the QR code information of the cheeses to ensure traceability in the follow-up. Before packing, the QR code scanning device in the packaging area scans the QR code on the cheese again to confirm the identity information of the cheese and the previous production process records. The scanning device is connected to the enterprise's information management system and uploads the scanning data in real time to ensure that every operation is accurately recorded and traced.

[0154] After confirmation, the cheeses are sent to the packing machine for automatic or semi-automatic packing operations, including bagging, sealing, labeling, etc. Labels or QR codes containing production information, quality inspection results, etc. are also attached to the packed yarn products for use in subsequent logistics and sales links. All the data generated during the whole process (including transfer records, inspection results, packing information, etc.) are integrated into the enterprise's information management system. Through the system, the entire production process of any yarn product can be queried and traced at any time.

[0155] As Figure 2 shown, an embodiment of the present invention also provides an ultra-fine yarn non-sensing operation quality traceability system 20, including:

[0156] An encoding module 21, used in the wire drawing area, to automatically identify and number the bobbins and cheeses on the wire drawing machine through a camera and a machine vision system; the variety produced by each wire drawing machine is bound to the machine vision system, and a visible bar code is printed on the cheese as the unique identifier of the cheese, corresponding to the number identified by the machine vision system;

[0157] A tracking module 22, used after wire drawing, the cheeses are sent to the twisting area through a transfer channel; cameras are installed in the transfer channel, and machine vision technology is used to identify and record the bar code information on the passing cheeses to ensure uninterrupted tracking of the cheeses during the transfer process;

[0158] A scanning module 23, used in the twisting area, each twisting machine is equipped with a unique machine number and is displayed in the form of a QR code; when the cheese reaches the twisting area, a scanning device is used to scan the QR code of the twisting machine and the bar code of the cheese to bind the information;

[0159] The processing module 24 is used to complete the twisting of yarn, and then the yarn is transported to the packaging area for inspection and packing. Through the manipulator in the packaging area, the tube yarn with a QR code is identified and grasped. The QR code scanning device in the packaging area scans and confirms the tube yarn to be packed, so that each step of the operation can be traced, and the whole process of yarn production can be managed.

[0160] It should be noted that this system corresponds to the above method, and all implementation manners in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0161] An embodiment of the present invention further provides a computing device, including: a processor and a memory storing a computer program. When the computer program is run by the processor, it executes the method as described above. All implementation manners in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0162] An embodiment of the present invention further provides a computer-readable storage medium storing instructions. When the instructions are run on a computer, the computer is caused to execute the method as described above. All implementation manners in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

Claims

1. A method for tracing the quality of ultra-fine yarn without feeling operation, characterized in that: The method comprises the following steps: In the wire drawing area, the wire tubes and wire cakes on the wire drawing machine are automatically identified and numbered through the camera and the machine vision system, including: capturing the images of the wire tubes and wire cakes on the wire drawing machine in real time through the camera; preprocessing the collected images, performing image segmentation, and separating the wire tubes and wire cakes from the background; obtaining a frame of color image from the camera, and calculating the grayscale value of each pixel in the color image; generating a grayscale image according to the grayscale value, and calculating the new value of each pixel through the convolution operation of the Gaussian function to obtain the image after Gaussian filtering; using the Sobel operator in the horizontal and vertical directions to perform convolution operations on the Gaussian filtered image to obtain each pixel The gradient components of the point in the horizontal and vertical directions; for each pixel point, the gradient amplitude is calculated according to the gradient components of each pixel point in the horizontal and vertical directions to obtain a gradient image containing the gradient amplitude and direction of each pixel point; the gradient image of the gradient amplitude and direction of each pixel point is traversed to identify and extract the continuous edge contour in the image; for each extracted contour, the number of pixels on the contour, the curvature radius or curvature angle of each point on the contour, and the overall direction of the contour or the direction of each segment are calculated; the variety produced by each wire drawing machine is bound to the machine vision system, and a visible barcode is printed on the wire cake as the unique identification of the wire cake, which corresponds to the number recognized by the machine vision system; After the wire drawing is completed, the silk cake is sent to the twisting area through the transfer channel. A camera is installed in the transfer channel to use machine vision technology to identify and record the barcode information on the silk cake passing by, so that the tracking of the silk cake is not interrupted during the transfer process. In the twisting area, each twisting machine is equipped with a unique machine number, which is displayed in the form of a QR code. When the silk cake arrives at the twisting area, a scanning device is used to scan the QR code of the twisting machine and the barcode of the silk cake to bind the information. After twisting, the yarn is transported to the packaging area for inspection and packaging. The robotic arm in the packaging area identifies and grabs the bobbin with a QR code. The QR code scanning device in the packaging area scans and confirms the bobbin to be packaged, so that each step of the operation can be traced and the entire yarn production process can be managed in a digital way.

2. The ultra-fine yarn non-sensing operation quality tracing method according to claim 1 is characterized in that: The camera is installed on the preset wire drawing machine; Automatic identification and numbering of wire tubes and wire cakes on the wire drawing machine through cameras and machine vision systems, including: The edge features of the silk tube and silk cake are screened out according to the number of pixel points on the contour, the curvature radius or curvature angle of each point on the contour, and the overall direction of the contour or the direction of each segment; According to the edge features of the silk tube and the silk cake, the silk tube and the silk cake in the image are identified, and the positions and postures of the silk tube and the silk cake are determined; Each identified bobbin and cake is assigned a unique number that is associated with the image, location and time of the bobbin and cake.

3. The ultra-fine yarn non-sensing operation quality tracing method according to claim 2 is characterized in that: Generate a grayscale image based on the grayscale value, and calculate the new value of each pixel through the convolution operation of the Gaussian function to obtain the image after Gaussian filtering, including: Calculate the grayscale values ​​of all pixels to create a new image, in which the value of each pixel is a grayscale value, wherein the new image is a grayscale image; Define the Gaussian kernel, which is a two-dimensional matrix whose values ​​are calculated according to the Gaussian function; Starting from the upper left corner of the image, select the first pixel, multiply the first pixel and its surrounding pixels by the value of the corresponding position of the Gaussian kernel; align the center of the Gaussian kernel to the selected pixel, and do the following: Multiply the upper left corner value of the Gaussian kernel by the pixel value at the corresponding position in the image; Multiply the upper median value of the Gaussian kernel by the pixel value at the corresponding position in the image until all values ​​of the Gaussian kernel are multiplied by the pixel value at the corresponding position in the image, add up all the products, and the result is the new pixel value at the corresponding position of the filtered image; Move the Gaussian kernel one pixel to the right until the new values ​​of all pixels in the image are calculated, and use the calculated new pixel values ​​to create a new image. The new image is the image after Gaussian filtering.

4. The ultra-fine yarn non-sensing operation quality tracing method according to claim 3 is characterized in that: The horizontal and vertical kernels of the Sobel operator are used to perform convolution operations on the Gaussian filtered image to obtain the horizontal and vertical gradient components of each pixel, including: Starting from the upper left corner of the image, select a pixel as the current processing point; cover the Sobel horizontal kernel on the current pixel and its surrounding pixels, and perform a convolution operation: Multiply the upper left corner element of the horizontal kernel by the value of the corresponding image pixel; multiply the upper middle element of the horizontal kernel by the value of the corresponding image pixel, and repeat this process until all elements of the horizontal kernel are multiplied by the corresponding image pixels; add all the products to get the gradient component of the current pixel in the horizontal direction; Use Sobel's vertical kernel to perform convolution on the same pixel and its surrounding pixels: Multiply the upper left corner element of the vertical kernel by the value of the corresponding image pixel; multiply the upper middle element of the vertical kernel by the value of the corresponding image pixel, and repeat this process until all elements of the vertical kernel are multiplied by the corresponding image pixels. Add all the products together to get the gradient component of the current pixel in the vertical direction. Move the current processing point one pixel to the right until the gradient components of all pixels in the image in the horizontal and vertical directions are calculated; A gradient component image is generated based on the gradient components of all pixels in the image in the horizontal and vertical directions.

5. The ultra-fine yarn non-sensing operation quality tracing method according to claim 4 is characterized in that: The camera is installed at a preset position on the wire drawing machine to capture the images of the wire tube and wire cake on the wire drawing machine in real time through the camera, including: Determine the set of all potential locations where the camera can be installed. The location points are fixed points around the wire drawing machine. Define the properties of each location point, including coordinates and viewing angle range. Randomly select a portion of all potential locations as the ant's initial location; assign a memory table to each ant to record the locations it has visited and the corresponding pheromone concentrations; Determine the initial concentration and volatilization rate of pheromones, and each ant chooses the next location to move to based on the pheromone concentration and heuristic information around its current location; After a certain number of iterations, the final position point is determined based on the memory tables and path evaluation results of all ants; The camera is installed according to the final position point, and the image of the wire tube and wire cake on the wire drawing machine is captured in real time by the camera.

6. The ultra-fine yarn non-sensing operation quality tracing method according to claim 5 is characterized in that: Each wire drawing machine produces a variety of products that are bound to the machine vision system. A visible barcode is printed on the wire cake as the unique identifier of the wire cake, which corresponds to the number recognized by the machine vision system, including: Assign a unique number to each type of wire drawing machine; Print the assigned number on the silk cake in the form of a barcode as a unique identifier of the silk cake; Establish a database or lookup table to store the corresponding relationship between varieties and numbers; The collected images are preprocessed by using image processing technology, and the preprocessed images are scanned and analyzed to extract the number information on the silk cakes; According to the extracted number information, the corresponding variety information is searched, and the variety information and number information are bound.

7. The ultra-fine yarn non-sensing operation quality tracing method according to claim 6 is characterized in that: Each wire drawing machine is assigned a unique number, including: Determine basic rules for numbering, including number length and character set to be used; Obtain information on all varieties produced by wire drawing machines, including variety name, specifications and characteristics; The number of each variety is represented as a gene code. According to the defined gene code, an initial population is randomly generated. Each individual in the population represents a number allocation scheme. Determine the fitness function used to evaluate the quality of each individual; According to the fitness function, the corresponding individuals are selected from the current population to enter the next generation, the selected individuals are crossover operated to generate new individuals, and the newly generated individuals are mutated until the preset number of iterations is reached to obtain the final number allocation scheme; According to the final numbering allocation plan, a unique number is assigned to each variety produced by the wire drawing machine.

8. A superfine yarn non-sensing operation quality tracing system, characterized in that: Applied to the method according to any one of claims 1 to 6, comprising: The encoding module is used to automatically identify and number the wire tubes and wire cakes on the wire drawing machine in the wire drawing area through the camera and machine vision system; the variety produced by each wire drawing machine is bound to the machine vision system, and a visible barcode is printed on the wire cake as the unique identification of the wire cake, which corresponds to the number recognized by the machine vision system; Tracking module: After the wire drawing is completed, the silk cake is sent to the twisting area through the transfer channel; a camera is installed in the transfer channel, which uses machine vision technology to identify and record the barcode information on the passing silk cake, so that the tracking of the silk cake is not interrupted during the transfer process; The scanning module is used in the twisting area. Each twisting machine is equipped with a unique machine number and displayed in the form of a QR code. When the silk cake arrives at the twisting area, the scanning device is used to scan the QR code of the twisting machine and the barcode of the silk cake to bind the information. The processing module is used to transport the yarn to the packaging area for inspection and packaging after the twisting is completed; the robot in the packaging area identifies and grabs the bobbin with a QR code; the QR code scanning device in the packaging area scans and confirms the bobbin to be packaged, so that each step of the operation can be traced and the whole process of yarn production can be managed in a digital way.

9. A computing device, characterized in that include: one or more processors; A storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, the one or more processors implement the method as claimed in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a program, which, when executed by a processor, implements the method according to any one of claims 1 to 7.

Images