Transformer High-Voltage Primary Coil Quality Detection Method Based on Machine Vision

Aiming at the problems of low efficiency and accuracy in manual detection winding angle wire spacing during automatic high-voltage primary coils transmission distribution transformers, a scheme using machine vision is proposed. Firstly, coil image acquired by industrial camera, region segmented, ROI (region interest) pre-processed. For detection, we propose slicing method for graying to reduce interference caused uneven light irradiation. The gray converted binary image, skeleton extraction performed; identified Hough transform feature straight lines, then calculated. an intersection perpendicular lines method, which extracts edge coordinates contour images performs endpoint pixel expansion shape classification. Use vertical determine centroid outline, calculate distance adjacent centroid, obtain combining calibration. Comparison experiments have shown that solution has high (0.01 mm), error integrated results not higher than 10%, enables real-time status corrects process according visual result improve finished product quality coils.