Research On Online Machine Vision Technology For Relay Contact And Riveting Point

As a kind of electrical component that controls the circuit to be on and off,the relay has a wide range of applications in control systems and power systems.The contact pieces act as a core component of the relay to control the on-off,and the quality of the contact will directly affect the performance of the entire relay.In this thesis,in view of the low efficiency and poor detection stability of the current manual detection of relay contacts,we propose a relay contact online detection system based on machine vision under the funding of the project "Relay Detection System Based On Machine Vision" entrusted by Zhejiang Meishu Electric Technology Co.,Ltd.We use the contact after the riveting as the specific inspection object..We have conducted in-depth research from two aspects of dimensional measurement and defect detection..The design and development of the whole system is completed,and the techniques of dimension measurement and surface defect detection are carried out.The thesis mainly includes the following contents:(1)On the basis of analyzing the measured object and various detected contents and detected requirements,we designed the overall scheme of the online visual inspection system for relay riveting contacts.The image acquisition module and lighting module of the system are carried out.And the functional module design of the software of the visual inspection system is completed.(2)We introduced the key dimension measurement method of relay riveting contacts.A fast extraction method of ROI region based on local feature matching is proposed.Then,a homography mapping measurement method is proposed,which is verified by experiments to have higher measurement accuracy.(3)Then,we analyzes the characteristics and types of surface defects of contacts and riveting points,we proposed corresponding detection methods for different surface defects of contacts and riveting points.Based on adaptive thresholding segmentation,connected domain calculation and geometric feature-based defect detection,we realized the defect detection on the surface of riveted point.Based on the comparative analysis of defect detection algorithms based on GLCM,LBP and HOG features,we proposed a kind of defect detection algorithm based on extended neighborhood LBP features and it is proved by experiments to have better recognition effect;Aiming at the insufficiency of KNN algorithm,we proposed a HashKNN algorithm based on hash table.It is verified by experiments that it has a faster calculation speed than KNN when the data set is larger.(5)Finally,we introduced the development of the visual inspection system for the relay contact rivet quality.The external I/O realized the two-way stable communication between the visual inspection system and the automation equipment,and we further improved the system performance by using multi-thread technology.On this basis,the system was tested several times on the spot and the results verified that the system measurement accuracy and detection speed meet the requirements.