Research On Inspection Technology Of Drilling And Riveting Quality Defects Based On Vision

Drill-rivet machining is an important process in aircraft manufacturing and is used extensively in the production of structural components such as skins,wall panels and frames.There are generally three possible quality defects in drill-rivet produced structural components.The first is that the drill-rivet process is not carried out in the designed position,known as "missed machining".Secondly,drill-rivet machining is mistakenly carried out outside the designed position,known as "mismachining".Thirdly,the actual drill-rivet position deviates significantly from the design position and does not meet the machining accuracy requirements,known as "machining overrun".Whichever defect may affect the strength of the component,will directly affect the subsequent assembly of the component and product quality.Therefore,the detection of drill-rivet quality defects is one of the key aspects of aircraft production quality assurance.At present,most of the domestic aircraft mainframes still use manual visual inspection combined with special templates for the detection of drill-rivet processing quality defects.This method is inefficient,has a large error margin,and is subject to leakage and mis-inspection.With the widespread use of digital and automated processes in aircraft production,current inspection methods can no longer meet the technical requirements of high efficiency,reliability and accuracy.Therefore,the study of new drill-rivet quality defect detection methods is an urgent technical problem to be solved.The visual inspection technology based on image measurement has the technical advantages of high inspection accuracy,high efficiency and high reliability,and can obtain 2D images and 3D data of the surface of the product under test.This has been widely used in many aspects of quality inspection in industrial production.Currently,the technology is applied to drill-rivet machining process hole position detection,both to quickly detect the number of holes,but also to obtain high-precision drilling position,simultaneous detection of missed processing,processing overrun and other defects.In this thesis,based on binocular stereo vision measurement technology,a drill-rivet quality defect detection method is designed,and the rivet hole image feature recognition method and the hole feature matching algorithm in stereo image pairs are studied.This achieves a high precision 3D reconstruction of the hole position.By comparing with the design data,the drill-rivet defects are determined and the location and characteristics of the defects are directly reproduced on the product surface using DLP projection technology.The main research elements of this thesis are as follows:1.The problems faced by drill-rivet quality defect detection technology and the current status of research are analysed.A set of drilling-rivet quality defect detection and spatial reproduction method based on binocular stereo vision measurement technology is proposed.2.A rivet hole image feature recognition method is studied.For the hole feature matching problem in stereo image pairs,a feature point matching algorithm based on geometric constraints and Delaunay triangle network adjacency relationship is proposed.3.The construction of the hardware and software platform of the inspection system is completed,and the high-precision reconstruction of multiple rivet hole positions and the reliable detection of drill-rivet quality defects are realised.The location and type of defects are reproduced on the workpiece surface using DLP projection technology to assist manual re-inspection.