Research On On-line Inspection Method Of Spinning Forming Quality Based On Machine Vision

With modern manufacturing industry developing towards high efficiency,high precision and low cost,on-line inspection of product forming quality which can realize real-time monitoring of forming process,has become an important way to improve the intelligent level of manufacturing industry.Spinning process is characterized by poin loading and continuous local forming,which leads to wrinkles,cracks and out of tolerance for form accuracy easily.Because spun parst are rotating at high speed during spinning forming,the traditional method of manual inspection for forming quality can't achieve real-time inspection.Therefore,accurate on-line inspection of defects and form accuracy has become basis and key to realize spinning forming automation and intellectualization.In this paper,the spun parts were taken as research objects.Machine vision technology was proposed to realize on-line inspection of spinning forming quality.Aiming at typical defects such as cracks and wrinkles,form accuracy such as straightness and circularity in spinning,a systematic study was carried out on construction of on-line image acquisition system,image preprocessing,feature extraction,fast recognition for defects,automatic calculation for form accuracy,software development and verification.Aiming at the problem that it is difficult to acquire high quality and real-time images under rotation of high speed,the overall architecture scheme of spinning forming image on-line acquisition system was designed.By using three light sources placed on the left,middle and right,clear and uniform illumination of part's surface can be realized.The speed of camera acquisition was controlled by PLC controller and encoder to ensure image acquisition in real-time.An on-line image acquisition system was established from three aspects of fixing,connecting and communication,and camera calibration was carried out for the system.Aiming at complex background interference of on-line image acquisition,a ROI extraction algorithm for automatic separation of spun part region and background was proposed.Aiming at image blurring caused by high-speed rotation,Wiener filter was used for image deblurring.Aiming at noise pollution in the process of image formation,transmission and storage,the Gauss filter was used for image denoising.In order to highlight contour of images and facilitate edge inspection,Laplacian operator was used to sharpen images.In order to solve the problem that high and low thresholds of Canny operator are difficult to determine,Canny operator was improved based on Otsu method for automatic calculation of image high and low thresholds.Thus,the accurate recognition of image edge of spun parts can be realized.According to change of gray level for spun parts with cracks,a recognition method based on Euler number was proposed.Interference of surface stain was eliminated by circularity R and aspect ratio D.The shape of spun part's opening contour was taken as feature.Based on first derivative and second derivative,a method for judging whether contour was a curve and number of waves in the curve was proposed.According to the wave height and wave distance in the curve,interference of small waves was prevented.The wrinkle defect was recognized based on number of effective waves.Linear fitting of contour points on both sides of part was based on least square method.Straightness of spinning parts was calculated by distance between contour and fitting line.Aiming at the problem that it is impossible to obtain cross-section image of spun parts,a method for calculating roundness based on eight images was proposed by rotation of parts.Eight images were acquired uniformly from 0 to 360 degrees according to rotation angle.The distance between contours on both sides of parts was taken as a diameter.Circularity was calculated by least square method.Based on the above research,an on-line spinning forming quality inspection software with five functional modules of user login,image acquisition,camera calibration,parameter setting and image processing was developed by WinForm framework.The result of off-line inspection of spinning forming quality was compared with software.The reliability of software was verified by comparision.The results show that recognition accuracy of fracture and wrinkle is 97.8% and 98.9% which take 29.2 ms and 38.5 ms.Measurement accuracy of straightness and circularity reaches 0.05 mm which take 98.5ms and 69.3ms.The on-line inspection method of spinning forming quality studied in this paper can realize rapid recognition of defects and accurate calculation of straightness and circularity.It is proved that the on-line inspection method has high accuracy and efficiency.