Design Of Real-time Detection System For Three-dimensional Braided Fabric

Braided fabric has good mechanical properties after forming,so it has a lot of applications in China's aerospace field.At the same time,the mechanical properties of the braided fabric are affected by many factors,among which the braiding angle and yarn skipping are the key factors.However,in the actual working process of the braided machine,the detection of the braiding angle and yarn skipping on the braided fabric is still in the manual detection stage,which leads to the phenomenon of wrong detection and missing detection,and the efficiency is also low.To solve the problem that is without good automation degree in the process of three-dimensional braided fabric,a real-time detection system of three-dimensional braided fabric based on machine vision is designed.The detection system consists of three parts: image acquisition,real-time detection software and PLC control.First of all,according to the requirements of the enterprise,the real-time detection system of three-dimensional braided fabric is designed as a whole,and the hardware selection and software function modules are designed according to the requirements.Hardware selection includes industrial camera,lens,light source,image acquisition card,PLC,motor,driver and other hardware selection;The software function module includes hardware control module,image acquisition module,image preprocessing module,braided angle detection module,yarn skipping detection module,data storage and transmission module,PLC control and result display module.Among them,image preprocessing,braided angle detection and yarn skipping detection module are the three most important software function modules of the detection system.In the image preprocessing module,image enhancement,noise reduction,binarization,morphological processing and edge extraction are carried out.In the braiding angle detection module,the straight line extraction part of the braiding angle detection algorithm is optimized,and the accuracy and real-time of the braiding angle detection are verified;In the yarn skipping detection module,two different detection schemes are designed: Based on machine learning and Hough transform.The accuracy and effectiveness of the two systems are tested and compared in real time,and the application scenarios,advantages and disadvantages of the two systems are summarized.Finally,the detection scheme based on machine learning is determined.Then,the data transmission between the software and PLC motion control part is established,and the configuration interface is designed for man-machine operation and real-time display of test results.Finally,the hardware platform of the whole system is built,and the system is tested by many groups of experiments.The main content of the test is to test the braids with different braiding angles for a long time.The test results show that the detection of braiding angle and yarn skipping has the required detection accuracy and speed,and can establish a connection with PLC,so that the start and stop of the braiding machine motor can be controlled in real time when the fault is detected.The whole real-time detection system of three-dimensional braids basically meets the requirements of enterprises for automatic detection of braids.