Bearing Corrosion Analysis Using Machine Vision and Computational Algorithms

摘要

This paper presents a new corrosion detection system to identify, accurately measure and categorize corrosion on bearings. The new corrosion detection system uses machine vision technology hardware and computer software to identify and classify corrosion and accurately determine the percentage of affected surface area on test specimens. This new methodology reduces human error, improves accuracy and increases repeatability of consecutive measurements relative to the traditional visual evaluation of corrosion. At present, there are several test methods that use bearings to determine the corrosion prevention properties of lubricating greases. They include the EMCOR method (ASTM D6138) and two other standard corrosion tests (ASTM D1743 and D5969). The downside of these methods is that they rely on subjective visual evaluation of corrosion and an approximate rating system.