Real-time vision-based detection of wire gapping on a rotating transformer core - Robotics Institute Carnegie Mellon University

Real-time vision-based detection of wire gapping on a rotating transformer core

C. J. Taylor, E. Evans, John M. Dolan, and Pradeep Khosla
Journal Article, SPIE Industrial Optical Sensors for Industrial Optical Sensors for Metrology and Inspection, Vol. 2349, pp. 235 - 246, November, 1994

Abstract

High-quality transformer winding requires precise measurement and control of the gapping between adjacent wires. We take a vision-based approach to the measurement subtask of determining the gaps between copper wires wound onto an oval transformer. The oval core shape, which can have an eccentricity as high as 2-to-1, leads to significant variations in surface normal and viewing distance. We use special lighting, a secondary mandrel shape sensor, and the specular reflection off the wires to give us an accurate model of the experimental geometry. We further exploit the vertical symmetry of the viewed region to condense our 2D image to a simple 1D signal containing reflectance peaks. after utilizing pattern recognition and some additional safety features to separate the wire peaks from background noise, we perform a least squares curve fit of the peaks to determine the subpixel maxima. the final algorithm is computationally fast and yields the desired wire gap in an absolute metric.

BibTeX

@article{Taylor-1994-13804,
author = {C. J. Taylor and E. Evans and John M. Dolan and Pradeep Khosla},
title = {Real-time vision-based detection of wire gapping on a rotating transformer core},
journal = {SPIE Industrial Optical Sensors for Industrial Optical Sensors for Metrology and Inspection},
year = {1994},
month = {November},
volume = {2349},
pages = {235 - 246},
}