Research On Broken Yarn Detection System Of Glass Fiber Winding Machine Based On Machine Vision

The visual broken yarn detection system of filament winding machine is an important application of broken yarn detection technology in the field of composite materials.Compared with the traditional broken yarn detection method,it has the advantages of high accuracy,good flexibility and low cost.It is of great significance in improving the quality of wound gas cylinders.At present,the research on the broken yarn detection system is not mature enough,the system is expensive,vulnerable to noise interference and other problems,and there is a lack of broken yarn position detection technology.Based on the above problems,this paper studies the key problems such as whether the glass fiber is disconnected,the position of glass fiber disconnection and alarm function,designs a set of glass fiber visual yarn breakage detection system,realizes the judgment of glass fiber disconnection and the identification of glass fiber disconnection position,and realizes the alarm function.The main work of this paper is as follows:Firstly,according to the characteristics of glass fiber and the process flow of filament winding machine,a yarn breaking detection scheme based on machine vision is designed.The hardware of the vision system is analyzed and selected,and the specific models of industrial camera,industrial lens and light source are determined.Secondly,the specific process of image processing is formulated.Through Open CV,the image processing methods used in the image processing process are compared and analyzed,and the effects of different parameters on glass fiber image are compared,so as to confirm the important parameters of glass fiber image processing.Through debugging,the number of contour pixels of each glass fiber is obtained,the threshold of the number of contour pixels is confirmed,and a method to judge whether the glass fiber is broken is proposed.According to the change law of the number of contour pixels of broken glass fiber in motion state,the recognition of whether the glass fiber is broken is realized.A "one-to-one" multi classification method based on SVM is proposed to identify the broken position of glass fiber.The contour centroid of each glass fiber is obtained through Open CV,and the contour centroid of each glass fiber is regarded as the training feature of SVM multi classification to realize the recognition of broken position of glass fiber.In addition,the visual broken yarn detection interface is compiled through Qt software,which realizes the real-time display of glass fiber image,the detection of broken yarn number and the prompt of abnormal position,and completes the serial communication between the host computer and 51 single chip microcomputer controller.The voice module,display module and indicator module are programmed by Keil4 software to realize the display functions of voice broadcast,display screen and indicator.Finally,an experimental platform for broken yarn detection of glass fiber winding machine based on machine vision is built to verify the accuracy of SVM multi classification;The host computer realizes the functions of glass fiber image display,yarn breaking detection and serial communication with 51 single chip microcomputer;The lower computer alarm system is verified.Through the comprehensive experiment of visual yarn breaking detection,the yarn breaking detection function of the yarn breaking detection system and the stability of the system are proved.