Computer program, flat knitting machine and information processing methods

Abstract

A computer program, a flat knitting machine, and an information processing procedure are provided that output the position of a knitting link identified as defective and / or information about the defect. The computer program instructs a computer to perform processing to: obtain an image of a knitting link from a flat knitting machine that has a plurality of knitting links arranged in a row; determine, based on the obtained image of the knitting link, whether the knitting link is acceptable or defective; and output the position of the knitting link identified as defective and / or information about the defect.

Description

Technical field

[0001] The present invention relates to a computer program, a flat knitting machine and an information processing method. State of the art

[0002] A flat knitting machine has a large number of knitting needles arranged in rows on at least two needle beds facing each other at the front and back. It creates a knitted fabric by hooking the supplied yarn, while the knitting needles are selectively driven by a carriage moving over the needle beds. Bending or breaking of knitting needles results in knitting defects.

[0003] Patent literature 1 describes a method for optically monitoring the quality of needles of a knitting machine by taking pictures of the needles with a camera. Reference list of patent literature:

[0004] Patent literature 1: Examined Japanese patent application no. H07-33998 Summary: Technical Problems

[0005] The method described in patent literature 1 can detect defects such as major or minor damage to the needles, but neither the position of a defective needle nor information about the defect is provided.

[0006] The present invention refers to these circumstances, and it is an object of the invention to provide a computer program, a flat knitting machine and an information processing method that output the position of a knitting link identified as defective and / or information about the defect. Problem solving

[0007] A computer program according to the present invention causes a computer to perform processing for: obtaining an image of a knitting link of a flat knitting machine having a plurality of knitting links arranged in a row; determining whether the knitting link is acceptable or defective, based on the obtained image of the knitting link; and outputting a position of the knitting link determined to be defective and / or information about the defect.

[0008] A flat knitting machine according to the present invention comprises: a storage unit that receives an image of a knitting element of a flat knitting machine having a plurality of knitting elements arranged in a row; a determination unit that determines, based on the received image of the knitting element, whether the knitting element is acceptable or defective; and an output unit that outputs a position of the knitting element determined to be defective and / or information about the defect.

[0009] An information processing method according to the present invention comprises: obtaining an image of a knitting element of a flat knitting machine having a plurality of knitting elements arranged in a row; determining whether the knitting element is acceptable or defective, based on the obtained image of the knitting element; and outputting a position of the knitting element determined to be defective and / or information about the defect.

[0010] According to the invention, the system determines whether a knitting element is acceptable or defective based on the image obtained of the knitting element. The knitting element is, for example, a knitting needle. However, a loop presser, a transfer pusher, or similar component can also be provided in addition to the knitting needle. The knitting needle comprises a hook, which is the end of the needle, a tongue, a slider, a needle shaft, etc. For the pass / fail determination, the similarity (e.g., a cosine similarity) between the obtained image of the knitting element and the image of a normal knitting element, as well as a machine learning model, can be used. If the knitting element is determined to be defective, the position of the defective knitting element and / or information about the defect are identified and output.Examples of information regarding the defect include the reliability level (probability) of the defect and the cause of the defect. According to the invention, the position of the strand identified as defective and / or information about the defect can be provided.

[0011] The computer program according to the invention causes the computer to perform a processing operation to output a measure for the knitted link identified as defective, based on a database in which defects in knitted links are associated with measures.

[0012] A defect in the knitting link is, for example, one cause of the defect and includes bending or splitting of the knitting link. In the case of a hook, this could involve bending, breaking, snagging (displacing the hook end), and similar issues. The remedy includes adjusting the knitting process (for example, split knitting, changing the stitch size), adjusting the output of a servo motor for driving a needle bed, adjusting the pull-down force of the knitted fabric, etc. According to the invention, a suitable countermeasure for the defective knitting link can be presented to the user.

[0013] The computer program according to the invention causes the computer to perform processing to: obtain progress information including a knitting program and knitting machine setting data used for knitting; determine a factor that exerts a load on the knitting link identified as defective, based on the information about the knitting link identified as defective and on the obtained progress information; and output a determination result.

[0014] The information about the knitting link identified as defective includes, for example, the position of the defective link and the cause of the defect. The knitting program is a set of instructions that specifies how the knitting link (for example, a knitting needle) is to be moved in each knitting stroke, how the yarn guide or carriage is to be moved, and so on. A stroke is a single movement of the carriage during knitting. The knitting machine settings include, for example, settings for a pull-down mechanism that pulls the knitted fabric downwards (for example, which area of ​​the knitted fabric is to be pulled downwards, for how long, and with what force), settings for the knitting speed (carriage movement speed), and settings for a stitch value (for example, the distance the lock lowers the knitting needle 11, which holds the yarn, so that the stitch size matches the target).The history information is obtained for each knitting operation by collecting the knitting program and flat knitting machine setting data used for that particular knitting operation. Specifically, the history information includes a knitting program and flat knitting machine setting data for tracing purposes when a knitted link is identified as defective, in order to identify what types of knitting the knitted link was used for in the past.

[0015] Based on the knitting program, stitch connections can be emulated. The connection of stitches can be expressed, for example, two-dimensionally, with the position of the knitting needle plotted on the horizontal axis and the stroke (time) plotted on the vertical axis, or three-dimensionally by adding the relationship between the front and back needle beds. Based on the knitting program and the knitting machine settings, the condition of all knitting links can be monitored in each stroke. Because it is possible to know how the knitting link will behave in each stroke in relation to the position of the identified defective link, and what the condition of the knitting link is with respect to the knitting process, the factor exerting a load on the knitting link can be determined.

[0016] According to the invention, by linking the defect of the knitting link (a defective part or the position of the knitting link) with the knitting program and the knitting machine setting data, the factor exerting a load on the knitting link can be determined. This prevents the exertion of an excessive load on the knitting link.

[0017] The computer program according to the invention causes the computer to perform processing for: inputting the obtained image of the knitting link, which is determined to be normal, into a first learning model that is trained to restore the image of the knitting link when an image of a knitting link is input, and to output a restored image; and determining whether the knitting link is acceptable or defective by comparing the obtained image of the knitting link with the output restored image.

[0018] An autocoder, for example, can be used as the first training model. This model is generated (trained) such that when an image of a normal knitting link is input, the image is reconstructed. The reconstructed image of the knitting link is then input into the first training model. The input image is compared to the reconstructed image output by the first training model. Preferably, the difference between the input image of the knitting link and the reconstructed image is calculated for this comparison. If the difference is small, the knitting link can be considered normal. If the difference is large, the knitting link can be considered defective.According to the invention, even if it is difficult to collect images of defective knitting links as training data, it is possible to determine whether the knitting link is acceptable or defective by using images of normal knitting links as training data.

[0019] The computer program according to the invention causes the computer to perform processing to determine whether the knitted link is acceptable or defective, based on a similarity between the obtained image of the knitted link and an image of a normal knitted link.

[0020] For example, cosine similarity can be used as a measure of similarity. Cosine similarity is the cosine value of the angle between two vectors and can be calculated by dividing the inner product of the two vectors by their magnitudes. The pixel values ​​of each pixel in the image of the knitted link can be vectorized. The cosine similarity is normalized within the range of -1 to 1. If the cosine similarity is 1 or close to 1, the knitted link can be determined to be normal. Conversely, if the cosine similarity is -1 or close to -1, the knitted link can be determined to be defective. Thus, if the cosine similarity is equal to or less than a predetermined threshold, the knitted link can be determined to be defective.According to the invention, it is not necessary to prepare a large number of images of non-defective and defective products, thereby reducing the effort required for prior preparation while still enabling processing with very accurate differentiation.

[0021] The computer program according to the invention causes the computer to perform processing to input the obtained image of the knitted link into a second learning model, which is trained to output information on whether the knitted link is acceptable or defective when an image of a knitted link is input, and to output information on whether the knitted link is acceptable or defective.

[0022] The information regarding whether the knitted link is acceptable or defective includes, for example, a probability (a degree of confidence) that the knitted link is defective and a probability (a degree of confidence) that the knitted link is normal. For example, if the threshold for determining that the knitted link is normal is 90%, then if the output of the second learning model indicates that the probability that the knitted link is normal is 80% and the probability that the knitted link is defective is 20%, then the knitted link can be determined to be defective. And if the output of the second learning model indicates that the probability that the knitted link is normal is 95% and the probability that the knitted link is defective is 5%, then the knitted link can be determined to be normal.Furthermore, the threshold can be set to different values ​​depending on the type (hook, tongue, slider, needle shaft, etc.) of the knitting needle, loop presser, transfer pusher, etc. According to the present invention, the accuracy of determining whether the knitted link is acceptable or defective can be improved by appropriately setting the threshold.

[0023] The computer program according to the invention causes the computer to perform processing to input the obtained image of the knitted link into a third learning model, which is trained to output a defective part of the knitted link and a cause of the defect when an image of a knitted link is input, and to output a defective part of the knitted link and a cause of the defect.

[0024] The third learning model can be, for example, YOLO (You Only Look Once), SSD (Single Shot MultiBox Detector), and Faster R-CNN. This third learning model extracts features from the input image of the knitted link, indicates the defective part of the link with a frame box, and estimates the probability of the cause of the defect. The cause with the highest probability can then be determined as the cause of the defect.

[0025] According to the invention, the defective part of the knitted link and the cause of the defect can be presented to the user.

[0026] The computer program according to the invention causes the computer to perform processing to output information relating to the defect of the knitting link to a monitor.

[0027] According to the invention, information about the defect of the knitting element (for example, the reliability level (the probability) of the defect and the cause of the defect) can be presented to the user.

[0028] The computer program according to the invention causes the computer to perform processing to output, to a monitor, an image of the knitting link identified as defective and / or information for identifying the position of the knitting link identified as defective.

[0029] According to the invention, an image of the knitting link identified as defective and / or information for identifying the position of the knitting link identified as defective can be presented to the user.

[0030] The computer program according to the invention causes the computer to perform a processing operation to move the knitting link identified as defective to a different height than the height of the knitting link according to a knitting program.

[0031] Because, according to the invention, the height of the knitting link identified as defective is moved to a different height than the normal height, the user can easily identify the position of the defective knitting link.

[0032] The computer program according to the invention causes the computer to perform processing for: outputting the position of the defective knitting link and the cause of the defect to a monitor of a customer's flat knitting machine; receiving information regarding the correctness or incorrectness of the output position of the knitting link and the cause of the defect; and storing, as training data, the received information regarding the correctness or incorrectness of the image of the knitting link identified as defective.

[0033] Because, according to the invention, the position of the defective knitting link and the cause of the defect are presented to the customer, the customer can determine whether the presented information is correct or not and can modify it if necessary. The information, such as the position of the defective knitting link and the cause of the defect, which the customer has determined to be correct or incorrect, can be collected as training data for the third learning model, which is used in the flat knitting machine operated by the customer.

[0034] The computer program according to the invention causes the computer to perform processing to specify the third learning model, which was retrained using the training data, as a model for the customer's flat knitting machine.

[0035] According to the invention, the third learning model, which is tailored to the customer's flat knitting machine, can be used. Advantageous effects of the invention

[0036] According to the invention, the position of the knitted link identified as defective and / or information about the defect can be provided. Brief description of the drawings Fig. Figure 1 is a schematic front view of the overall configuration of a flat knitting machine according to the invention. Fig. Figure 2 is a schematic view showing an example of the configuration of a needle bed and a carriage. Fig. Figure 3 is a schematic view showing an example of the configuration of a locking mechanism provided in a sled. Fig. Figure 4 is a schematic view showing an example of the configuration of a control unit. Fig. 5A is a schematic view showing an example of the configuration of a knitting needle and an image of a defective knitting needle. Fig. 5B is a schematic view showing an example of the configuration of a knitting needle and an image of a defective knitting needle. Fig. 5C is a schematic view showing an example of the configuration of a knitting needle and an image of a defective knitting needle. Fig. 5D is a schematic view showing an example of the configuration of a knitting needle and an image of a defective knitting needle. Fig. Figure 6 is a schematic view showing a first example of determining whether the knitting needle is acceptable or defective. Fig. Figure 7 is a schematic view showing a second example of determining whether the knitting needle is acceptable or defective. Fig. Figure 8 is a schematic view showing a third example of determining whether the knitting needle is acceptable or defective. Fig. Figure 9A is a schematic view showing an example of a procedure for determining the position of a defective knitting needle and the cause of the defect. Fig. Figure 9B is a schematic view showing an example of a procedure for determining the position of a defective knitting needle and the cause of the defect. Fig. Figure 10 is a schematic view showing a first example of a result of determining whether the knitting needle is acceptable or defective. Fig. Figure 11 is a schematic view showing a second example of a result of determining whether the knitting needle is acceptable or defective. Fig. Figure 12 is a schematic view showing an example of information that associates the causes of defects in knitting needles with measures. Fig. Figure 13A is a schematic view showing an example of a division knitting. Fig. Figure 13B is a schematic view showing an example of a division knitting. Fig. Figure 13C is a schematic view showing an example of a division knitting. Fig. 13D is a schematic view showing an example of a division knitting. Fig. Figure 13E is a schematic view showing an example of a division knitting. Fig. Figure 13F is a schematic view showing an example of a division knitting. Fig. Figure 14 is a schematic view showing the emulation results of mesh connections. Fig. Figure 15 is a schematic view showing an example of a factor that exerts a load on a knitting needle. Fig. Figure 16 is a schematic view showing an example of a processing procedure performed by a control unit. Description of embodiments

[0037] The following describes embodiments of the invention. Fig. Figure 1 is a schematic front view of the overall configuration of a flat knitting machine 100 according to the invention. Fig. Figure 2 is a schematic view showing an example of the configuration of a needle bed 10 and a carriage 20. Fig. Figure 3 is a schematic view showing an example of the configuration of a locking mechanism 21 provided in the slide 20. The following are indicated by the arrows O, U, V, H, L and R in Fig. 1 and Fig. The three directions given are defined as up, down, forward, backward, left, and right. Furthermore, some elements may be omitted from the views for the sake of simplicity.

[0038] As in Fig. 1 and Fig. As shown in Figure 2, the flat knitting machine 100 comprises the needle bed 10, the carriage 20, a yarn guide rail 30, a yarn stand 40, a servo motor and a control unit (not shown), etc. The flat knitting machine 100 knits a knitted fabric K.

[0039] The needle beds 10 are arranged such that they face each other at the front and back, with a hook slot S between them. Viewed from the side, the front and back needle beds 10 are arranged in an inverted V-shape and inclined upwards towards the front-back center (the side where the needle beds 10 face each other) (see Fig. 2) Each needle bed 10 is equipped with a large number of knitting needles 11 arranged along the lengthwise (left-right direction) of the needle bed 10. The front and back needle beds 10 can move relative to each other in the left-right direction as they transfer stitches between them. In this description, the knitting needle 11 includes a hook, which is the end of the needle, a tongue, a slider, a needle shaft, etc., which are described below. Furthermore, in Fig. 1 and Fig. 2 Two needle beds 10 are arranged such that they face each other at the front and back, although the configuration is not limited to this. Two needle beds 10 can also be arranged at the top and bottom, one each on the front and back sides, so that the flat knitting machine 100 comprises a total of four needle beds 10. In this embodiment, the knitting element is, for example, the knitting needle 11. In addition to the knitting needle 11, a loop presser, a transfer pusher, etc., can also be provided. The loop presser is a link driven by the carriage 20 for pressing down the yarn extending between stitches, which is used to prevent the knitted fabric or the knitting yarn from lifting. The transfer pusher is used to receive stitches from the knitting needles 11 in the needle bed 10, move the stitches to the left and right relative to the needle bed 10, and then transfer the stitches back to the knitting needles 11.During the knitting process, defects can occur in the loop presser, the transfer pusher, etc., and also in the knitting needles 11. For example, if, in the case of the loop presser, the needle bed on the opposite side moves by a smaller offset (distance of movement left / right) than the target due to increased yarn tension, there is a possibility that the loop presser will collide with the knitting needle or similar object on the opposite side and be damaged if the loop presser is advanced further. Similarly, in the case of the transfer pusher, as with bending of knitting needle 11, there is a possibility that the transfer pusher will bend if knitting needle 10 is moved relatively under high yarn tension. Knitting needle 11 is described here as an example of a knitting link, but the knitting link is not limited to knitting needle 11.

[0040] The carriages 20 are arranged in a front / rear pair such that they face the front and rear needle beds 10 from above. The carriage 20 can be moved back and forth along the longitudinal direction of the needle bed 10 by a servo motor (not shown).

[0041] Within the front and rear needle beds 10, the front needle bed 10 is also referred to as a front needle bed 10F, and the rear needle bed 10 is also referred to as a rear needle bed 10B. Similarly, within the front and rear carriages 20, the front carriage 20 is also referred to as a front carriage 20F, and the rear carriage 20 is also referred to as a rear carriage 20B.

[0042] A thread guide 31, which feeds a knitting thread Y, is held on the thread guide rail 30 in such a way that it can be moved.

[0043] As in Fig. As shown in Figure 2, a camera 22 and an optical element 23 (for example, a prism or a lens) are provided in the carriage 20. The camera 22 captures an image of the knitting needle 11 when the knitting needle 11 is moved to a predetermined position by the locking mechanism 21 and no yarn is held by the knitting needle 11. The optical element 23 guides light from the knitting needle 11 to the camera 22 by refracting or reflecting the light. A lighting unit 24 is also provided at a corresponding position on the carriage 20. The lighting unit 24 comprises, for example, a corresponding number of LED elements, with the LED elements used for each needle bed 10 being divided so that the lighting conditions (exposure time, current of the LED elements, etc.) can be optimized.

[0044] As in Fig. As shown in Figure 3, the front carriage 20F is provided with three locking mechanisms 21 for moving the knitting needles 11 backwards and forwards. In particular, a first transfer locking mechanism 21A, a knitting locking mechanism 21B and a second transfer locking mechanism 21C are arranged along the direction of movement (left-right direction) of the carriage 20.

[0045] Each locking mechanism 21 can move the knitting needle 11 backwards and forwards by guiding the foot of the knitting needle 11, which is selected based on a knitting program, along a backwards and forwards extending movement path L. This allows for the formation of stitches using the knitting yarn Y and the transfer of stitches. The foot is a link that is actuated by the locking mechanism 21 when the knitting needle 11 is advanced into the hook slot S.

[0046] If the sled 20 moves as indicated by the arrow in Fig. When the carriage 20 moves to the right, the knitting lock mechanism 21B acts as a leading system for forming stitches. Furthermore, in this case, the first transfer lock mechanism 21A acts as a following system for transferring the stitches formed by the knitting lock mechanism 21B. Additionally, in this case, the second transfer lock mechanism 21C does not transfer the stitches. The same applies when the carriage 20 moves to the left.

[0047] The operation of the lock mechanism 21 described above is to be understood as an example, whereby, for example, three lock mechanisms 21 can also be used to sequentially perform a re-hanging, a mesh formation and a re-hanging.

[0048] The front slide 20F was described above with reference to Fig. 3 described, but the rear slide 20B is also operated in the same way.

[0049] Fig. Figure 4 is a schematic view showing an example of the configuration of a control unit 50. The control unit 50 comprises a control section 51, which controls the entire control unit 50, a communication section 52, a memory 53, an interface section 54, a destination section 55, and a storage section 56. The control unit 50 is connected to a database 71, a monitor 72, and a servo motor 73.

[0050] Control section 51 is configured using a required number of CPUs, MPUs, GPUs, etc. Alternatively, control section 51 can also be configured using a combination of a DSP, an FPGA, etc.

[0051] The communication section 52 includes a communication module and has a function for communicating with an external facility (not shown).

[0052] Memory 53 can be configured using semiconductor memory such as SRAM, DRAM, ROM, or flash memory.

[0053] The interface section 54 has a function as an interface with the monitor 72 and the servo motor 73.

[0054] The control section 51 can control the operation of the servo motor 73 via the interface section 54. The control section 51 can move the carriage 20 as desired by controlling the operation of the servo motor 73. Furthermore, the control section 51 can detect the position of the carriage 20 based on the number of rotations of the servo motor 73. Additionally, the control section 51 can control the operation of the locking mechanism 21, the camera 22, and the lighting unit 24 via the interface section 54.

[0055] By controlling the operation of the servo motor 73, the camera 22, and the light unit 24, the control section 51 can maintain the image of the knitting needle 11 by capturing an image of the knitting needle 11 between the time a knitting project is completed and the time the next one is started, immediately before the knitting machine is switched off after knitting the required number of garments, or during a temporary stop for oiling or cleaning. The control section 51 thus functions as a maintenance section that preserves an image of the knitting needle 11. If the knitting process exerts a heavy load on the knitting needle 11, the predetermined number can be set lower. Conversely, if the knitting process exerts a light load on the knitting needle 11, the predetermined number can be set higher.When the image of the knitting needle 11 is taken, the knitting needle 11 is preferably moved to the advanced state by the closing mechanism 21 and no knitting thread is held by the knitting needle 11.

[0056] Only the image of a part that could be defective can be extracted from the image captured by camera 22, and the extracted image can be output to control unit 50. In this way, a reduction in the accuracy of determining whether knitting needle 11 is acceptable or defective, which could be caused by the influence of thread remnants (such as lint or fiber dust), can be suppressed.

[0057] The memory section 56 can be configured, for example, by a hard disk or semiconductor memory and can store a computer program (a program product) 60, a knitting program 61, knitting machine setting data 62, a first learning model 63, a second learning model 64, a third learning model 65, a template image 66, and required information. The first learning model 63, the second learning model 64, and the third learning model 65 comprise a pre-training model, a post-training model, and a post-retraining model.

[0058] Computer program 60 is loaded into memory 53 and executed by control section 51. Control section 51 can perform processing defined by computer program 60. The processing by control section 51 is also processing by computer program 60.

[0059] The computer program 60 can be downloaded from an external device via the communication section 52 and stored in the memory section 56. Alternatively, the computer program 60 recorded on a recording medium M (for example, an optically readable, disc-shaped storage medium such as a CD-ROM) can be read by a recording medium reader (not shown) and stored in the memory section 56. The computer program 60 can be loaded for execution on a single computer or on a multitude of computers located in one place or distributed across multiple locations and interconnected by a communication network.

[0060] The determination section 55, the knitting program 61, the knitting machine setting data 62, the first learning model 63, the second learning model 64, the third learning model 65, and the template image 66 are described further below. Instead of storing the first learning model 63, the second learning model 64, and the third learning model 65 in the control unit 50, the first learning model 63, the second learning model 64, and the third learning model 65 can also be stored on a cloud server. Processing can then be carried out on the server using the first learning model 63, the second learning model 64, and the third learning model 65, and the processing results can be sent to the control unit 50.

[0061] The following describes the images of knitting needle 11 and the defective knitting needle 11.

[0062] Fig. Figures 5A to 5D are schematic views showing examples of the configuration of knitting needle 11 and the image of the defective knitting needle 11. Fig. Figure 5A shows the configuration of an end part of a tongue needle 1 as a knitting needle 11. A hook 2 is provided at the end of the tongue needle 1, and the hook 2 is opened and closed by a tongue 3. The tongue 3 pivots around a shaft 4. The tongue 3 can pivot between a state in which the hook 2 is closed and a state in which the hook 2 is fully open. When the tongue needle 1, which holds a needle loop of a stitch in the hook 2, pivots to the right Fig. When needle 5A moves, a loop of the stitch is pressed downwards by a movable plate or similar device. And when the old loop is pressed downwards by a puller or similar device, the knitting yarn of the needle loop moves relatively to the left in relation to the tongue needle 1. As the tongue needle 1 moves further to the right, the knitting yarn of the needle loop opens the tongue 3 and moves to the side of the needle shaft 5. On the side of the needle shaft 5, there is a shoulder 6 with a large step. Thus, even if the tongue needle 1 moves further to the right during the transfer, the knitting yarn of the needle loop remains on the shoulder 6 and does not move to the side of the base 7 of the tongue needle 1.

[0063] Fig. Figure 5B shows the main part of a sliding needle 12, similar to a knitting needle 11. The sliding needle 12 comprises a needle body 13 and a slider 16. The slider 16 comprises two blades 14 and a base 15. The needle body 13 has a hook 13a at the end of the needle shaft, and a slider groove 13c is provided in the needle shaft. At least a lower part of the blade 14 is received in the slider groove 13c of the needle body 13 in a state where the two blades 14 overlap each other in the lateral direction. A peg 14a is provided at the front end of the blade 14. The blade 14 and the base 15, in combination, function as the slider 16. Fig. Figure 5B shows a state in which the opening of the hook 13a is closed by the pin 14a. When the needle body 13 moves forward relative to the slider 16, the opening of the hook 13a is opened.

[0064] In this description, the knitting needle 11 includes the hooks 2 and 13a, the tongue 3, the slider 16 (especially the blade 14), the needle shaft 5, etc.

[0065] Fig. Figure 5C shows an example image of a defective hook. Causes of hook defects include, for example, breakage (damage) of the hook, bending of the hook, and hanging (or reverse hanging) of the hook. A broken hook is a condition in which the end of the hook is missing. A bent hook is a condition in which part of the hook is curved or bent. A hanging hook is a condition in which the end of the hook is positioned closer to the needle stem than on a non-defective product. And reverse hanging is a condition in which the end of the hook is positioned on the opposite side from the needle stem compared to a non-defective product.

[0066] Fig. 5D shows an example image of a defective slider. In Fig. 5D is missing part of one of two sliders. The image of the defective knitting needle 11 is not on the one in Fig. The 5 examples shown are limited.

[0067] The following describes a procedure for determining whether knitting needle 11 is acceptable or defective.

[0068] Fig. Figure 6 is a schematic view showing a first example of determining whether knitting needle 11 is acceptable or defective. Control section 51 can determine whether the obtained knitting needle 11 is acceptable or defective based on the similarity between the obtained image of knitting needle 11 (also called the "captured image") and the image of the normal knitting needle 11 (also called the "template image"). As shown in Fig. As shown in Figure 6, the sizes (resolutions) of the captured image and the template image are assumed to be N points × M points, where the pixel value of a pixel (i, j) in the captured image is denoted as a i,j is specified and the pixel value of a pixel (i, j) in the template image is called b i,j The captured image can be an image containing the entire knitting needle 11. However, areas outside the pass / fail target area can also be removed from the captured image of knitting needle 11 beforehand.

[0069] For example, cosine similarity can be used as a measure of similarity. Cosine similarity is the cosine value of the angle between two vectors and can be calculated by dividing the inner product of the two vectors by their magnitudes. The pixel values ​​of each pixel in the image can be vectorized. The cosine similarity is normalized in the range of -1 to 1. If the cosine similarity is 1 or close to 1, knitting needle 11 can be determined to be normal. Conversely, if the cosine similarity is -1 or close to -1, knitting needle 11 can be determined to be defective. Thus, if the cosine similarity is equal to or less than a predetermined threshold, knitting needle 11 can be determined to be defective.According to this embodiment, it is not necessary to prepare a large number of images of non-defective and defective products, thus reducing the effort required for preliminary preparation while still enabling processing with very accurate differentiation. Furthermore, the similarity is not limited to cosine similarity, and other methods such as Euclidean distance can also be used.

[0070] Instead of determining whether knitting needle 11 is normal (a non-defective product) or defective, the degree of deterioration of knitting needle 11 can also be determined as a pass / fail criterion. The degree of deterioration can be classified into several levels, such as "significantly deteriorated," "somewhat deteriorated," "hardly deteriorated," and "not deteriorated." The level can be determined corresponding to the value of the cosine similarity in the range of -1 to 1.

[0071] An ID can be assigned to each of the recorded images of all knitting needles 11 that are targets, corresponding to the position of knitting needle 11, whereby the position of the defective knitting needle 11 can be identified based on the ID of the recorded image determined to be defective.

[0072] The template image 66 can be pre-stored in memory section 56 for each type (hook, tongue, slider, needle shaft, etc.) of the knitting needle 11. Furthermore, the operating time of the flat knitting machine 100 can be divided into multiple sections, and the template image 66 can be prepared for each section. Because the condition of the knitting needles 11 gradually deteriorates as the operating time of the flat knitting machine 100 progresses, the accuracy of the pass / fail determination can be improved using the template image 66, which represents the deteriorating condition.

[0073] Furthermore, a large number of template images 66 can be prepared corresponding to different usage times of the knitting needle 11. The usage time since the start of use of a new knitting needle 11 on the flat knitting machine 11, or since the start of use of the knitting needle 11 after a replacement, is stored for each knitting needle 11. Once the determination of whether the knitting needle 11 is acceptable or defective has been carried out, the control section 51 can identify the template image 66 corresponding to the usage time from the large number of template images 66 based on the usage time of the knitting needle 11 at the time of the determination and perform a pass / fail determination using the identified template image 66.

[0074] Fig. Figure 7 is a schematic view showing a second example of determining whether knitting needle 11 is acceptable or defective. When an image of knitting needle 11, which is determined to be normal, is input, the control section 51 can input the received image of knitting needle 11 into the first learning model 63, which has been trained to recover the image of knitting needle 11, and output a recovered image. It can then determine whether knitting needle 11 is acceptable or defective by comparing the received image of knitting needle 11 with the recovered image output from the first learning model 63. Thus, the control section 51 functions as a determination unit that, using the first learning model 63, determines whether knitting needle 11 is acceptable or defective.

[0075] For example, an autocoder can be used as the first learning model 63, comprising an encoder 631 and a decoder 632. Encoder 631 extracts a feature vector from the input image, and decoder 632 restores (reconstructs) the original image based on the extracted feature vector. Furthermore, the first learning model 63 is not limited to an autocoder; other models, such as Conditional GAN, can also be used.

[0076] The first learning model 63 is therefore generated (trained) so that when an image of the normal knitting needle 11 is entered, the first learning model 63 can recreate the image. As in Fig. As shown in Figure 7, the captured image of knitting needle 11 is entered into the first learning model 63. The entered image of knitting needle 11 is compared with the recovered image output from the first learning model 63. This comparison can be performed by calculating the difference between the entered image of knitting needle 11 and the recovered image. The difference can be obtained by calculating the difference in pixel value for each corresponding pixel between the image of knitting needle 11 and the recovered image, and summing the calculated differences for each pixel. If the difference is small, knitting needle 11 can be determined to be normal. If, on the other hand, the difference is large, knitting needle 11 can be determined to be defective.Therefore, if it is difficult in this embodiment to collect the images of the defective knitting needles 11 as training data, it can be determined whether the knitting needle 11 is acceptable or defective by using the images of the normal knitting needles 11 as training data. Furthermore, the difference can be calculated as follows: The difference in the pixel value for each corresponding pixel between the image of knitting needle 11 and the recovered image can be compared to a predetermined threshold, and the number of pixels with a difference equal to or greater than the threshold can be calculated as the difference.

[0077] The image of knitting needle 11 entered into the first learning model 63 can be an image containing the entire knitting needle 11, or it can be a split image obtained by dividing the image into parts (hooks, sliders, tongues, etc.) of the knitting needle 11. Furthermore, areas outside the pass / fail determination target can be removed from the captured image of knitting needle 11 beforehand. If an image containing the entire knitting needle 11 is used, a pass / fail determination can be performed for each area by calculating the difference for each from the multitude of areas in the entire image, and it can be determined which part of the knitting needle 11 is abnormal by identifying the area corresponding to a defective part.In the case of split images, the difference between each split image and a restored image can be calculated in accordance with the split image in order to perform a pass / fail determination for each split image.

[0078] Furthermore, instead of determining whether knitting needle 11 is normal (a non-defective product) or defective, the degree of deterioration of knitting needle 11 can be determined as a pass / fail assessment. The degree of deterioration can be classified into several levels, such as "significantly deteriorated," "somewhat deteriorated," "hardly deteriorated," and "not deteriorated." The level can be determined according to the value of the difference. This allows, for example, a thermal map to show which parts of knitting needle 11 are deteriorated and which parts are not.

[0079] The first learning model 63 can be generated (trained) as follows. Images of standard knitting needles 11 are collected as training data. In this case, the images of the standard knitting needles 11 can be collected as training data by changing the image capture conditions, the distance to the knitting needles 11, their positions, etc. Based on the collected training data, the images of the standard knitting needles 11 are entered into the first learning model 63, and the parameters of the first learning model 63 are adjusted so that the recovered images output by the first learning model 63 closely resemble the input images of the knitting needles 11.

[0080] Fig. Figure 8 is a schematic view showing a third example of determining whether knitting needle 11 is acceptable or defective. When an image of knitting needle 11 is input, control section 51 can input the received image (captured image) of knitting needle 11 into the second learning model 64, which has been trained to output information about whether knitting needle 11 is acceptable or defective, and output this information. Control section 51 thus functions as an output unit that outputs information about whether knitting needle 11 is acceptable or defective using the second learning model 64. The image of knitting needle 11 input into the second learning model 64 can be an image containing the entire knitting needle 11, or it can be a split image obtained by dividing the image into parts (hook, slider, tongue, etc.) of knitting needle 11.Furthermore, areas outside the good / bad determination target can be removed beforehand from the recorded image of knitting needle 11.

[0081] For example, a Convolutional Neural Network (CNN) can be used as the second learning model 64. The information regarding whether knitting needle 11 is acceptable or defective includes, for example, a probability (a degree of confidence) that knitting needle 11 is defective and a probability (a degree of confidence) that knitting needle 11 is normal. If, for example, the threshold for determining that knitting needle 11 is normal is assumed to be 90%, and the output of the second learning model 64 indicates that the probability that knitting needle 11 is normal is 80% and the probability that knitting needle 11 is defective is 20%, then knitting needle 11 can be determined to be defective.And if the output of the second learning model 64 indicates that the probability that knitting needle 11 is normal is 95% and the probability that knitting needle 11 is defective is 5%, then knitting needle 11 can be determined to be normal. Furthermore, the threshold can be set to different values ​​depending on the type (hook, tongue, slider, needle shaft, etc.) of knitting needle 11. In this embodiment, the accuracy of determining whether knitting needle 11 is acceptable or defective can be improved by setting the threshold accordingly.

[0082] Furthermore, instead of determining whether knitting needle 11 is normal (a non-defective product) or defective, the degree of deterioration of knitting needle 11 can be determined as a pass / fail assessment. The degree of deterioration can be classified into several levels, such as "significantly deteriorated," "somewhat deteriorated," "hardly deteriorated," and "not deteriorated." The level can be determined according to the probability of being defective.

[0083] The second learning model 64 can be generated (trained) as follows. Images of normal knitting needles 11, images of defective knitting needles 11, "normal" labels associated with the images of normal knitting needles 11, and "defective" labels associated with the images of defective knitting needles 11 are collected as training data. Based on the collected training data, the images of normal knitting needles 11 are entered into the second learning model 64, and the parameters of the second learning model 64 are adjusted so that the label output from the second learning model 64 approximates the "normal" label. Furthermore, based on the collected training data, the images of defective knitting needles 11 are entered into the second learning model 64, and the parameters of the second learning model 64 are adjusted so that the label output from the second learning model 64 approximates the "defective" label.Furthermore, using the training data in which the images of the defective knitting needles 11 are associated with "defect" labels indicating the causes of the defects (for example, bending, hanging or breaking the hook, bending or splitting the slider and bending or splitting the tongue) in the knitting needles 11, the second learning model 64 can be configured to output the cause of the defect in the knitting needles 11 when the image of the knitting needle 11 is input.

[0084] The following describes a procedure for determining whether a defective part of knitting needle 11 is defective and the cause of the defect.

[0085] Fig. 9A and Fig. Figure 9B shows schematic views illustrating an example of a procedure for determining a defective part of the defective knitting needle 11 and the cause of the defect. When an image of the knitting needle 11 is input, the control section 51 can input the received image of the knitting needle 11 into the third learning model 65, which has been trained to output the defective part of the knitting needle 11 and the cause of the defect, and output the defective part of the knitting needle 11 and the cause of the defect. That is, the control section 51 functions as an output unit that outputs the defective part of the knitting needle 11 and the cause of the defect using the third learning model 65. The image of the knitting needle 11 input into the third learning model 65 can be an image containing the entire knitting needle 11 or a split image obtained by dividing the image into parts (hook, slider, tongue, etc.) of the knitting needle 11.Furthermore, areas outside the good / bad determination target can be removed beforehand from the recorded image of knitting needle 11.

[0086] The third learning model 65 can be, for example, YOLO (You Only Look Once), SSD (Single Shot MultiBox Detector), and Faster R-CNN. As in Fig. As shown in Figure 9A, the third learning model 65 comprises an input layer 651, a convolution / pooling layer 652, and an output layer 653. The third learning model 65 extracts features from the input image (captured image) of the knitting needle 11. As shown in Fig. As shown in Figure 9B, the third learning model 65, for example, identifies the defective part of knitting needle 11 using a frame box and estimates the probability of the cause of the defect. The cause with the highest probability can be determined as the cause of the defect (for example, bending, snagging, or breaking of the hook; bending or splintering of the slider; and bending or splintering of the tongue). Furthermore, the third learning model 65 is not limited to YOLO.

[0087] According to the configuration described above, the defective part of the defective knitting needle 11 and the cause of the defect can be presented to the user.

[0088] The third learning model 65 can be generated (trained) as follows. Training data, in which images of the defective knitting needles 11, defect positions of the defective knitting needles 11, and "cause" labels indicating the causes of the defects are associated with each other, are collected beforehand. Based on the collected training data, the images of the defective knitting needles 11 are entered into the third learning model 65, and the parameters of the third learning model 65 are adjusted so that the defective parts and the labels output by the third learning model 65 approximate the defective parts and the "cause" labels of the defective knitting needle 11 in the training data.In addition, training data containing the images of the normal knitting needles 11 are collected, the images of the normal knitting needles 11 are entered into the third learning model 65, and the parameters of the third learning model 65 are adjusted so that the third learning model 65 does not output defective parts.

[0089] The following describes a procedure for presenting the result of a determination as to whether knitting needle 11 is acceptable or defective to the user.

[0090] Fig. Figure 10 is a schematic view showing a first example of the result of a determination of whether knitting needle 11 is acceptable or defective. A pass / fail determination result screen content 200 can be displayed on monitor 72. Between the completion of one knitting project and the start of the next, immediately before switching off the knitting machine after knitting the required number of garments, or during a temporary stop of machine operation for oiling or cleaning, the taking of images of knitting needles 11, the pass / fail determination, etc., are performed automatically, and the pass / fail determination result screen content 200 is displayed on monitor 72. Control section 51 thus functions as an output unit that outputs the position of the knitting needle 11 identified as defective and / or information about the defect.The Good / Bad Determination Result screen (200) displays data collection results. These results include, for example, the total number of defective knitting needles (11) identified as defective within all knitting needle beds (11), the number of defective needles per bed (front top, front bottom, back top, back bottom) in a model containing four needle beds, the positions of the defective needles, and the degree of deterioration of the defective needles. The position of the defective needle for each bed is displayed when a bed selection tab (204) is selected. In the example of... Fig. When the front upper bed is selected (level 10), the number of the defective needle (needle number) is displayed. The degree of deterioration of the defective needle can be classified, for example, into four levels from level 1 to level 4, where level 1 is "not deteriorated," level 2 is "hardly deteriorated," level 3 is "somewhat deteriorated," and level 4 is "significantly deteriorated." Level 4 is a deterioration level that requires replacement. Level 3 can be a deterioration level that should be taken into consideration. Additionally, although not shown, the reliability level (the probability) of identifying each defective needle can be displayed. For example, the probability of the defect for the XXth knitting needle is 00%. Regarding the location of the defective needle, the information can include details about which bed the defective needle belongs to, in addition to the information about the defective needle number.

[0091] In the pass / fail result screen (200), a deterioration status field (206) is displayed. This field contains a selection box for choosing a defective needle, the degree of deterioration of the selected defective needle at the time of a previous test, and the degree of deterioration of the selected defective needle at the time of the current test. This allows the user to identify the degree of deterioration each time a pass / fail test is performed.

[0092] The pass / fail result screen content 200 can display an image of a normal needle (a non-defective product) and an image of one identified as defective, allowing for comparison. Additionally, an action field 207 is displayed within the pass / fail result screen content 200. For example, the user might be prompted to replace a defective knitting needle 11. Furthermore, action fields 207 can display actions for the identified knitting needle 11 based on database 71, which associates the cause of the defect in knitting needle 11 with corresponding actions.

[0093] The threshold is a threshold for pass / fail determination, where a threshold can be set for each of the hook, the slide, and the tongue. If the in Fig. The cosine similarity shown in Figure 6 is used to determine whether knitting needle 11 is acceptable or defective. The pass / fail determination is made based on whether the cosine similarity is equal to or greater than the threshold value. And if this is in Fig. The first learning model 63 shown in Figure 7 is used to determine whether knitting needle 11 is acceptable or defective. The pass / fail determination is made based on whether the difference is equal to or greater than the threshold. And if this is in Fig. When the second learning model 64 shown in Figure 8 is used to determine whether knitting needle 11 is acceptable or defective, the pass / fail determination is made according to whether the probability is equal to or greater than the threshold. A "Reassess" symbol 202 is a symbol for confirming the entered threshold.

[0094] By pressing a "Start" symbol 201, it can be manually determined whether knitting needle 11 is acceptable or defective. A "Mark" symbol 203 is used to perform dummy knitting and to change the height of a needle identified as defective. Dummy knitting is knitting in which knitting needles 11 are raised and lowered without lifting the yarn from the yarn guide. During the dummy knitting process, images of knitting needles 11 are taken to determine whether they are acceptable or defective. An image of a needle identified as defective can be displayed so that the user can determine whether the needle is acceptable or defective. The user can review the image and press a confirmation symbol 205 for the needle identified as defective (needle number).After checking all questionable needle patterns, the "Mark" symbol 203 can be activated to move the carriage and shift the defective knitting needle 11 to a different height than its normal height. The configuration is not limited to manually changing the height of the defective knitting needle 11 to a different height than its normal height based on the "Mark" symbol 203; a configuration can also be used in which the height of the defective knitting needle 11 is automatically changed to a different height than its normal height.

[0095] As described above, the control section 51 can output information about defects in the knitting needle 11 to the monitor 72. In this way, information about defects (for example, the reliability level (the probability) of the defect and the cause of the defect) in the knitting needle 11 can be presented to the user.

[0096] Furthermore, the control section 51 can output the image of the defective knitting needle 11 and / or the information for identifying the position of the defective knitting needle 11 to the monitor 72. In this way, the image of the defective knitting needle 11 and / or the information for identifying the position of the defective knitting needle 11 can be presented to the user.

[0097] Furthermore, the control section 51 can move the knitting needle 11 identified as defective to a different height than the height specified in the knitting program. This allows the user to easily identify which knitting needle 11 is defective among the knitting needles 11 arranged in a row on the needle bed 10.

[0098] Fig. Figure 11 is a schematic view showing a second example of the result of determining whether knitting needle 11 is acceptable or defective. The capture results in a Good / Bad Determination Result screen content 210 include the position of a defective needle for each bed (front top, front bottom, back top, back bottom), the cause of the defect, etc. In the example of Fig. In step 11, the front, upper bed is selected, and the number of the defective needle (needle number) and the cause of the defect are displayed. For example, it can be seen that the cause of the defect in the XXth needle is a bent hook, and the cause of the defect in the XOth needle is a splintered slider. The user can view the image of the defective needle by pressing selection icon 215 for the knitting needle 11 identified as defective.

[0099] The pass / fail result screen content 210 can be used to process the results of the processing of the in Fig. 9 of the third learning model shown, 65 will be displayed.

[0100] In the pass / fail result screen content 210, the user (the customer using the flat knitting machine 100) can enter whether the detection result of the control unit 50 is correct or not. As in Fig. As shown in 11, the user can correct the recording result by entering or selecting in the True / False field whether the position of the defective knitting needle 11 is correct or not and whether the cause of the defect is correct or not, and by pressing a “Correction” symbol 211.

[0101] As described above, the control section 51 can output the position of the defective knitting needle 11 and the cause of the defect to the monitor 72 of the customer's flat knitting machine, receiving information on the correctness or incorrectness of the output position of the knitting needle 11 and the cause of the defect, and storing the received information on the correctness or incorrectness in association with the image of the knitting needle 11 identified as defective as training data in the memory unit 56 or the database 71.

[0102] Because the position of the defective knitting needle 11 and the cause of the defect are presented to the customer in this way, the customer can determine whether the presented information is correct or not and can modify it if necessary. The information, such as the position of the defective knitting needle 11 and the cause of the defect, which the customer has determined to be correct or incorrect, can be collected as training data for the third learning model 65, which is used in the flat knitting machine operated by the customer.

[0103] Control section 51 can specify the third learning model 65, which has been retrained using the customer-specific training data collected as described above, as a model for the customer's flat knitting machine. The specified third learning model 65 can replace the third learning model 65 already installed in the customer's flat knitting machine. Therefore, even if the third learning model 65 is the same, the customer-specific third learning model 65, tailored to the customer's flat knitting machine, can be used.

[0104] Database 71 stores information that associates the causes of defects in knitting needles 11 with measures.

[0105] Fig. Figure 12 is a schematic view showing an example of information where the cause of a defect in knitting needle 11 is associated with a measure. As in Fig. As shown in Figure 12, a basic method for preventing a hook from bending is to use a dividing knitting technique, although measures such as changing the stitch size (yarn length) can also be employed. For example, knitting with a larger stitch size can prevent a hook from bending due to a certain degree of pulling.

[0106] Fig. Figures 13A to 13F are schematic views showing an example of dividing knitting. Dividing knitting is described with reference to Fig. 13 described. Fig. Figures 13A to 13C show an example without a division knitting, and Fig. Figures 13D to 13F show an example using a dividing knitting technique. As in Fig. 13A and Fig. As shown in diagram 13B, and especially at loop numbers #2 and #3, #2 and #3 are directly connected to each other with a one-needle spacing. As shown in Fig. As shown in Figure 13C, in the cable pattern knitting process where the positions of pairs (#1, #2) and (#3, #4) are exchanged, #2 and #3 are pulled diagonally across a three-needle gap. Therefore, a hook may bend.

[0107] Fig. Figures 13D to 13F show a case of a division knitting in which a stroke of a division knitting is inserted. As in Fig. 13D and Fig. As shown in 13E, and especially at loop numbers #2, #3, and #5, #2 and #5 are directly connected to each other via a diagonal three-needle spacing, and #3 is directly connected to #2A, which is formed in the knitting stroke immediately before #2 and #4. As shown in Fig. As shown in Figure 13F, during the knitting process, the front and back needle beds move relative to each other by the size of three needles, causing #2 and #3 to be pulled diagonally away from each other. However, because #2A is directly connected to #3, the load on the needle is reduced, thus preventing or suppressing hook bending.

[0108] As again in Fig. As shown in Figure 12, the output of the servomotor 73 for driving the needle bed can be adjusted to prevent the hook from sticking. Hook sticking can be attributed to a situation where, under extreme load on the knitting needle 11, the position of the needle bed during the offset (the relative movement of the needle beds) can deviate from the target due to the pulling of the knitted fabric. This can cause the hook to collide with the yarn guide, which was stopped during the needle advance, or with the knitting needle 11 on the opposite side when transferring stitches. To prevent this, the output of the servomotor 73 is adjusted so that it cannot be overcome by lateral pulling (an external force) caused by the knitted fabric.

[0109] If the slider or tongue is damaged, the pulling force on the knitting can be adjusted. Specifically, if the mechanism for pulling the knitting down is not properly adjusted, excessive stress is exerted on knitting needle 11, causing damage to the slider or tongue. The information relating to the cause of the defect in knitting needle 11 and this measure is for illustrative purposes only and is not exhaustive. Fig. The 12 examples shown are limited.

[0110] As described above, control section 51 can output a measure for the knitting needle 11 identified as defective, based on database 71, in which defects (causes of defects) of knitting needle 11 are associated with measures. The identified knitting needle 11 and the measure can be output (displayed) on monitor 72. In this way, an appropriate measure for the defective knitting needle 11 can be presented to the user.

[0111] Then, conditions that could exert a load on the knitting needle identified as defective can be extracted and presented to the user, leaving the decision on an appropriate action (process) to the user. The following explanations refer to this.

[0112] The information regarding the defective knitting needle 11 includes, for example, the position of the defective knitting needle 11 and the cause of the defect. The knitting program 61 contains commands that specify how the knitting needle 11 should be moved in each knitting stroke, how the yarn guide or carriage should be moved, etc. A stroke is understood to be a single movement of the carriage during knitting. The knitting machine setting data 62 includes, for example, settings for a pull-down device that pulls down the knitted fabric (for example, which area of ​​the knitted fabric should be pulled down, for how long, and with what force), settings for the knitting speed (carriage movement speed), and settings for a stitch value (for example, the distance by which the lock lowers the knitting needle 11 holding the knitting yarn so that the stitch size corresponds to the target).The progress information can be stored in database 71 and is collected for each knitting by gathering the knitting program and the flat knitting machine setting data used during knitting.

[0113] Fig. Figure 14 is a schematic view showing the emulation result of mesh connections. The history information described above can be used for the emulation. Fig. The horizontal axis (number 14) indicates the needle number (position) of knitting needle 11, and the vertical axis indicates the stroke (time). Stroke 1 indicates the bottom side (lower edge) of the garment, with the stroke numbers increasing towards the top side (upper edge). The in Fig. The emulation result shown in Figure 14 is an emulation result of stitch connections based on the knitting program 61, whereby the state of all knitting needles 11 can be monitored in each stroke based on the knitting program 61 and the knitting machine setting data 62. In particular, the magnitude of a tensile force (load) exerted on the knitting needle 11 can be determined based on the number of threads held by the knitting needle 11, the needle spacing between stitches, the size of the stitch, the movement of the knitting needle 11, etc. A stitch in which the tensile force exerted on the knitting needle 11 exceeds a predetermined threshold can be previously identified as a knitting operation that exerts a load on the needle.

[0114] A program that can reverse the determination process described above is being prepared. If, as in Fig. Figure 14 shows that a needle with a specific number is identified as defective. By entering the defective needle's number into the program described above, the knitting process that exerts a load on the needle in each stroke can be identified in accordance with the defective needle's number. In other words, for each stroke, the knitting state can be determined in accordance with the number of the defective knitting needle 11, such as how needle 11 moves, the number of threads held by needle 11, the needle spacing between stitches, and the stitch size. Therefore, the factor exerting a load on knitting needle 11 can be identified. In the example of Fig. 14. It can be identified that the knitting exerts a load on the needles in two strokes.

[0115] Fig. Figure 15 is a schematic view showing an example of a factor that exerts a load on knitting needle 11. Examples of factors that exert a load on knitting needle 11 include offsetting, stitch size, multiple hooks, special knitting techniques, the coefficient of friction of the yarn, and the pull-down force of the knitted fabric.

[0116] Regarding the displacement, the thread extending between the front and back is pulled due to the relative movement of the needle beds. Regarding the stitch size, when forming smaller stitches, the sliding resistance increases because the needle passes through the smaller loop of the stitch as it advances. Regarding the pulling force on the knitted fabric, an inappropriate pulling force places excessive stress on the knitting needles 11. The factors that exert stress on knitting needles 11 are examples and are not specific to the work described in the text. Fig. The 15 examples shown are limited.

[0117] The determination section 55 (control section 51) can receive progress information, including the knitting program and the knitting machine setting data used for knitting; determine a factor that exerts a load on the knitting needle 11 identified as defective, based on the information about the knitting needle 11 identified as defective and on the received progress information; and output the determination result to the monitor 72. The information about the knitting needle 11 identified as defective includes, for example, the position of knitting needle 11 (needle number).

[0118] In particular, control section 51 specifies the correspondence between the knitting needles 11 and the knitting patterns that exert a load on the knitting needles 11 in each stroke of the knitting, based on the obtained history information. For the knitting needle 11 identified as defective, control section 51 can identify the knitting that exerts a load on the defective knitting needle 11 using the specified correspondence and determine the factor exerting a load on the knitting needle 11 based on the identified knitting.

[0119] By linking the defect of knitting needle 11 (the position of knitting needle 11) with the knitting program 61 or the knitting machine setting data 62, the factor exerting a load on knitting needle 11 can be determined. By presenting the result of this determination to the user, the user can perform the necessary processing to prevent an overload from being exerted on knitting needle 11.

[0120] Fig. Figure 16 is a schematic view showing an example of a processing procedure carried out by the control unit 50. The control section 51 receives an image of the knitting needle 11 (S11) and determines whether the knitting needle 11 is acceptable or defective, based on the received image (S12). The pass / fail determination can be performed using the Fig. The procedures shown in sections 6 to 8 are carried out. Control section 51 determines whether a defective knitting needle 11 is present or not (S13). If no defective knitting needle 11 is present (NO in S13), control section 51 executes step S22, which is described below.

[0121] If a defective knitting needle 11 is present (YES in S13), the control section determines the position of the defective knitting needle 11 and the cause of the defect (S14). For example, the following can be used to process step S14: Fig. The 9 procedures shown are used. Control section 51 specifies a measure for the defective knitting needle 11 based on database 71, in which defects in the knitting needle 11 are associated with measures (S15).

[0122] Control section 51 determines the factor exerting a load on the defective knitting needle 11, based on the knitting program and the knitting machine setting data (S16). For example, the processing of step S16 can be based on the following: Fig. 14 and Fig. The procedures shown in Figure 15 are used. Control section 51 outputs an image of the defective knitting needle 11, its position, the cause of the defect and a measure for the defect (S17) and outputs the factor that exerts a load on the knitting needle 11 (S18).

[0123] Control section 51 determines whether a requirement for the determination result is received (S19). If a requirement is received (YES in S19), control section 51 receives the requirement (S20) and stores the modified determination result in memory section 56 (S21). If no requirement is received (NO in S19), control section 51 performs the processing described above in step S22.

[0124] Control section 51 determines whether processing should be terminated or not (S22), and if processing should not be terminated (NO in S22), continues knitting (S23) and determines whether it is time to photograph the knitting needles 11 or not (S24). The time for taking the photograph could, for example, be the time at which a predetermined number of garment products have been produced, but it is not limited to this.

[0125] If it is not time to capture an image (NO in S24), control section 51 continues processing step S24. If it is time to capture an image (YES in S24), control section 51 captures an image of knitting needle 11 and continues processing from step S11. If processing is to be terminated (YES in S22), control section 51 terminates processing.

[0126] As described above, the control section 51 receives the image of the knitting needle 11 of the flat knitting machine 100, which has a plurality of knitting needles 11. Based on the received image of the knitting needle 11, it determines whether the knitting needle 11 is acceptable or defective and outputs at least the position of the knitting needle 11 described as defective and / or information about the defect. The information about the defect includes, for example, the reliability level (probability) of the defect and the cause of the defect. According to this embodiment, the position of the knitting needle 11 identified as defective and / or information about the defect can be provided.

[0127] In this embodiment, for example, a method for indicating the defective knitting needle 11 with a laser pointer or a method for marking the defective knitting needle 11 with a corresponding color in an identifiable manner can be used as a means for notifying the position of the knitting needle 11 identified as defective.

[0128] The present embodiment can also be used to determine whether a knitting link other than knitting link 11 is acceptable or defective. Examples of the knitting link include links such as loop pressers or transfer pushers, which are arranged in several rows on the needle bed and are configured to be movable forwards and backwards with respect to the hook slot. List of reference symbols 10 Needle bed 11 knitting needles 20 sleds 21 Locking mechanism 22 Camera 23 optical element 24 lighting units 50 control unit 51 Control section 52 Communication section 53 storage 54 Interface section 55 Determination section 56 Storage section 60 computer programs 61 Knitting program 62 Knitting machine setting data 63 first learning model 64 second learning model 65 third learning model 66 stencil image 71 database 72" Monitor 73 Servomotor 100 flat knitting machines QUOTES INCLUDED IN THE DESCRIPTION

[0000] This list of documents cited by the applicant was automatically generated and is included solely for the reader's convenience. The list is not part of the German patent or utility model application. The DPMA accepts no liability for any errors or omissions. Cited patent literature

[0000] JP 07-33998

[0004]

Claims

[1] A computer program that causes a computer to perform a processing operation to: Obtaining an image of a knitting link of a flat knitting machine, which has a multitude of knitting links arranged in a row, Determine whether the knitted link is acceptable or defective, based on the obtained image of the knitted link, and Output of the position of the identified defective link and / or information about the defect. [2] Computer program according to claim 1, wherein the computer is caused to perform a processing operation to output a measure for the knitting link identified as defective, based on a database in which defects in knitting links are associated with measures. [3] Computer program according to claim 1 or 2, wherein the computer is caused to perform a processing operation to: Obtaining progress information, including a knitting program and knitting machine setting data used for knitting, Determining a factor that exerts a load on the identified defective rope link, based on information about the identified defective rope link and on the obtained progress information, and Output of a determination result. [4] Computer program according to any one of claims 1 to 3, wherein the computer is caused to perform a processing operation to: Inputting the obtained image of the knitting link determined to be normal into an initial learning model that is trained to restore the image of the knitting link when an image of a knitting link is input, and to output a restored image, and Determine whether the knitting link is acceptable or defective by comparing the received image of the knitting link with the output recovered image. [5] Computer program according to any one of claims 1 to 4, wherein the computer is caused to perform processing to determine whether the knitted link is acceptable or defective, based on a similarity between the obtained image of the knitted link and an image of a normal knitted link. [6] Computer program according to any one of claims 1 to 5, wherein the computer is caused to perform a processing operation to input the obtained image of the knitting link into a second learning model that is trained to output information on whether the knitting link is acceptable or defective when an image of a knitting link is input, and to output information on whether the knitting link is acceptable or defective. [7] Computer program according to any one of claims 1 to 6, wherein the computer is caused to perform processing to input the obtained image of the knitted link into a third learning model that is trained to output a defective part of the knitted link and a cause of the defect when an image of a knitted link is input, and to output a defective part of the knitted link and a cause of the defect. [8] Computer program according to any one of claims 1 to 7, wherein the computer is caused to perform processing to output information about the defect of the knitting link to a monitor. [9] Computer program according to any one of claims 1 to 8, wherein the computer is caused to perform processing to output to a monitor an image of the knitting link identified as defective and / or information for identifying the position of the knitting link identified as defective. [10] Computer program according to any one of claims 1 to 9, wherein the computer is caused to perform a processing operation to move the knitting link identified as defective to a height other than a height of the knitting link in accordance with a knitting program. [11] Computer program according to claim 7, wherein the computer is caused to perform a processing operation to: Outputting the position of the defective knitting link and the cause of the defect to a monitor of a customer's flat knitting machine, Receiving information regarding the accuracy or inaccuracy of the reported position of the knitting link and the reported cause of the defect, and Store, as training data, the received information regarding the correctness or incorrectness and the image of the knitted link identified as defective. [12] Computer program according to claim 11, wherein the computer is caused to perform a processing operation to specify the third learning model, which has been retrained using the training data, as a model for the customer's flat knitting machine. [13] A flat knitting machine, which includes a preservation unit that maintains an image of a knitting link of a flat knitting machine which has a plurality of knitting links arranged in a row, a determination unit that, based on the obtained image of the knitting link, determines whether the knitting link is acceptable or defective, and an output unit that outputs the position of the knitting link identified as defective and / or information about the defect. [14] An information processing procedure that includes: Obtaining an image of a knitting link of a flat knitting machine, which has a multitude of knitting links arranged in a row, Determine whether the knitted link is acceptable or defective, based on the obtained image of the knitted link, and Output of the position of the identified defective link and / or information about the defect.

Images