Vision-Based Piston Rod Surface Damage Detection System Application Research

The piston rod is the only actuating element in the cylinder,and its surface quality directly affects the service life of the cylinder,as well as whether the pushing force of the cylinder can reach the rated value under the specified working environment,which puts forward high requirements on the surface quality of the piston rod.At the same time,with the development of industrial automation level in China,enterprises also need to improve market competitiveness by enhancing production efficiency,and automatic loading technology based on machine vision guidance is of great practical significance to improve production efficiency.The main object of this research is a piston rod of 8 mm diameter,and the related research is conducted for the problems of automatic feeding of piston rods and surface damage during the machining of piston rods,and the main research contents are as follows.(1)To propose a damage detection method based on multiple light complementary to the phenomenon of random distribution of damage location and angle on the surface of piston rod to reduce the leakage rate of defective products in piston rod.(2)The piston rod is pushed above the feeding table by the lifting mechanism of the mechanical system,and the vision system of the feeding module acquires the image of the piston rod on the feeding table and processes it to extract the coordinates of the piston rod center point position,usually the position and attitude of the piston rod on the feeding table are different,and this paper classifies the piston rod position on the feeding table into horizontal,inclined and vertical by angle calculation and connected domain width calculation,and processes them separately.Finally,the coordinate position of the center point of the piston rod on the feeding table is identified and sent to the lower machine to drive the single-axis servo slide to drive the lifting mechanism to the designated position to lift the piston rod to feed the inspection module.(3)Taking the piston rod surface image as the research object,the technical difficulties are the uncertainty of the location of the milled cake on the piston rod surface image,the Non-parallel of the horizontal CMOS chip array of the line array camera and the incident light line of the strip light source causing the uneven light and darkness of the left and right pieces of the image,and the unclean cleaning of the piston rod surface causing the mistaken detection of oil or water stains as surface damage.In view of the above difficulties,this thesis proposes a piston rod surface image milling cake removal method based on transformation,a gray scale fast correction method based on scale factor image and a Canny edge detection algorithm based on adaptive high and low threshold parameters,respectively.Finally,based on the studied piston rod surface damage detection algorithm,the piston rod surface damage detection system is designed and the stability of the visual inspection algorithm is ensured through verification experiments in industrial sites.A piston rod is detected within 3s,and the overall leakage rate is less than 0.05% and the false detection rate is less than 15%,which meets the detection accuracy requirements of the piston rod surface damage detection system.