Identifying Wear Particles With Their Shape Attribute

Abstract

Wear particles are produced in all oil-lubricated machines when mechanical parts come in contact. These particles are generated in differing sizes, altering quantities, with specific composition, and morphology. The particles when examined by experts provide vital information concerning machine health to operate and maintain safety, efficiency, and economy. As condition monitoring becomes increasingly important and needs to be carried out regularly, a custom-designed software system is developed to investigate the wear particles by using computer vision and image processing techniques. Identifying abnormal wear at an earlier stage avoids wearing failure modes of components that may lead to the breakdown of the machinery. Conventional methods are used to diagnose wear conditions, however; the use of six morphological attributes of particle size, shape, edge details, color, surface texture, and thickness ratio is the basis of constructing an image analysis system to strengthen wear judgments. The particle shape is one of the most important attributes, which is investigated in this paper. Various parameters are used to extract shape features to identify six wear particle types including severe sliding wear, cutting edge, non-metallic, fatigue, water, and fiber particles. Particle identification is based on eight geometric features which are extracted by the image analysis system and MATLAB code. Tests are conducted to achieve 90% accuracy.