| In the automated production line of circular weft knitting robot,it has been possible to use the barrel changer to take full drums from the bracket,and to place full drums and empty drums from the bracket instead of manual full drums feeding and empty drums feeding from the bracket.However,in this process,due to the influence of the yarn gravity,vibration,frame installation and other factors,the position of the yarn will change,which will affect the mechanical claws to take the empty and full drums away.To solve the above problems,in order to ensure the accurate capture and placement of the yarn by the winder robot,this paper uses machine vision to complete the design of the automatic winder positioning system of the frame.The contents of the system are as follows:(1)For the process of automatic rewinding of the rewinding robot,the overall scheme of the automatic rewinding positioning system of the frame is designed.Firstly,the coordinate initialization of the pole on the overall frame is completed by the pole positioning,which provides the initial coordinates for the winder position and the subsequent winder position of the winder robot.Then,by identifying and positioning the spout of the yarn,the yarn can be grabbed during the process of taking out the yarn to achieve the function of automatic changing the spool.In this process,the monocular camera is used as the image acquisition module,and complete positioning function through the principle of small hole imaging.(2)To solve the problems of rusty material,inconsistent surface and difficult identification of the yarn pole on the frame,the yarn pole is identified and positioned by the combination of markers and traditional image processing.First,uniform surface features of the yarn pole are installed by the pole cap.Then,image preprocessing is completed by graying the image of the pole,OTSU threshold segmentation,median filter,morphological opening operation to effectively remove the image background interference.Next,the outline of the yarn pole is obtained by Canny operator,and then the yarn pole is fitted by least squares method.The image information needed in the positioning process of the single camera for the yarn pole is extracted by fitting,and the positioning of the yarn pole is finally completed.(3)To solve the problem that traditional images cannot accurately identify the yarn in complex environment,the yarn is identified by deep learning.First,Yolov5 is selected as the basic model,and then adds the Ghost module to replace some of the modules,which reduces the amount of model parameters and improves the speed of model detection.Then,CBAM module is introduced to improve the detection accuracy,and an experimental test is carried out.The test effectiveness verify that the model can correctly identify the mouth of yarn,and the accuracy is as high as 99.2%.Compared with the original model,the parameters are compressed by 47.2%,and the FPS increased by 5 frames/s.Finally,the yarn is identified by yolov5,and the yarn is positioned using the location information of the yarn at the center of the image.(4)To meet the requirements of the automatic spool changing positioning system of the yarn frame,the software platform of the positioning system is designed,including the functions of camera control,parameter adjustment,image acquisition,etc.At the same time,the yarn pole recognition and positioning algorithm and the yarn recognition and positioning algorithm are fused into the upper computer software to achieve the recognition and positioning of the collected yarn pole and yarn images,and the detection results will be displayed in real time.In the end,by collecting images of yarn rods and tubes and comparing them with actual data,this positioning method meets the requirements of the automatic bobbin changing positioning system. |
PDF Full Text Request