Research On Detection Method Of Aircraft Skin Riveting Quality Based On Machine Vision

The airframe components of the aircraft are composed of tens of thousands of parts,and their connection methods are mainly riveted,so the detection of riveting quality is a key link in the entire aircraft assembly process.At present,most production workshops still use the traditional manual detection method for testing,which is inefficient,high in capital investment and may be missed,and can no longer meet the needs of modern industrial intelligent detection.Industrial inspection technology based on machine vision is widely used in the production and manufacturing process of modern industry because of its high detection efficiency,non-contact,online detection and analysis,and real-time evaluation.In this paper,the riveting hole of aircraft skin and the riveted pier head after riveting are used as the detection objects,and a machine vision-based detection method for the riveting quality of aircraft skin is studied.The research content mainly includes the following parts:Firstly,the aircraft skin riveting quality image is pre-processed,mainly including image grayscale processing,filtering processing,thresholding segmentation,morphological processing and edge detection.Image grayscale processing is mainly to reduce the amount of data that needs to be processed for future image processing work.By comparing the experimental results of common filtering methods,it is found that median filtering and bilateral filtering can denoise the rivet hole image and rivet image respectively,while retaining the image edge details.Image thresholding segmentation can transform images into binary images,and this paper compares and studies several different threshold segmentation methods,and finally selects the most suitable threshold for binary segmentation of images.Then,morphological processing methods were studied to process the image of rivet pier heads.The operation of edge detection is mainly to obtain the edge contour of the image,and after experimental comparison,it is found that Canny edge detection has the best effect,in order to improve the detection accuracy,this paper adopts the improved algorithm of polynomial interpolation subpixel edge positioning to extract the edge of the image and obtain high-quality image edge information.Secondly,the extraction technology of aircraft skin riveting quality feature elements is studied,mainly including the detection and extraction of riveting hole aperture(i.e.riveting hole diameter),roundness and burr.Detection and extraction of the diameter and roundness of rivet pier heads after riveting.According to the feature elements to be extracted,the image after preprocessing the riveting hole image was measured by the least squares fitting circle method to obtain the riveting hole aperture and the roundness error evaluation of the riveting hole was evaluated by the least squares method to extract the roundness feature of the riveting hole,while the detection and extraction of the riveting pier head diameter was detected by the Hough transform circle detection method,the extraction of the roundness feature of the rivet pier head was extracted by the minimum external circle method.For the detection of riveting hole burr,this paper scans the contour points of riveting hole,and compares the distance from the center of the circle to the edge contour point with the set threshold,and then finds the suspicious burr points of riveting hole and makes statistics.Finally,the diagnosis model of least squares support vector machine based on particle swarm parameter optimization is studied to realize the diagnosis of riveting quality.Including the construction of mass diagnosis model,the selection of kernel function and parameter optimization based on particle swarm.During the experiment,it was found that the particle swarm algorithm had some drawbacks,and this paper improved the particle swarm algorithm.The experimental results show that the method of aircraft skin riveting quality inspection based on machine vision proposed in this paper can meet the inspection requirements.