Study Of Alignment System For Solar Wafer Screen Printing Based On Machine Vision

Solar energy which is the most important new energy has a broad market while theenergy crisis is in an emergency. The solar wafer screen printing equipment determines theproduction key issues such as production efficiency and printing quality. The alignmentsystem of wafer and screen is a critical subsystem. This paper designs an vision-baseautomatic alignment system, studies the line detection algorithm based on RandomizedHough Transform (RHT) to locate wafer, and the circle detection algorithm to locate screen.A high precision and efficiency alignment system based on vision is designed combiningthe characteristics of the wafer and screen after analyzing the screen printing process and therequirements of alignment system. The line detection and circle test algorithm are proposed tolocate the wafer and the screen. The kinematics solutions of a plane parallel mechanismnamed UVW platform is deduced to correct the deviation between wafer and screen．Eight neighborhood tracking algorithm combining ellipse fitted with the least-squaresmethod is proposed to extract circle outline for circle location, and the2D disorder circlearray coordinates are sorted by limiting the distance and slope of coordinates, which is usedfor camera calibration and improves its adaptability. The relationship between the left andright sub-cameras is calibrated based on circle location algorithm.The RHT algorithm is studied to detect the wafer edge line which costs too much timeand space.3optimization methods are proposed to solve these problems. The hierarchicalpyramid image and interlaced block sampling strategy is used to reduce the edge points; theconcept of imitative gradient direction is proposed to classify the edge points; and fitting linewith least-squares method improves the detection accuracy. At last, the screen is located withcircle detection and the algorithm flow of alignment system is described.In the end of this paper, performance indicators of alignment system are tested. Theexperiment results prove that calibration accuracy is up to±2μm and circle location accuracyis10μm. The accuracy of the line detection is30μm and the time is20ms in most of theconditions, which meets the production requirements. The vision-base alignment system hasbeen applied in the solar wafer screen printing equipment.