Classification of Wear Particles Based on Qualitative Morphological Features

Wear debris morphology is closely related to the wear mode and mechanism occurred. Image recognition of wear particles is, therefore, a powerful tool in wear monitoring. An algorithm of classification of wear particles is proposed based on qualitative morphological features. The standard classes are presented as a set of vectors of coded ratings. Descriptions of the standards are based on the knowledge-base of experts. A distance between the particle and the standard classes in the multidimensional space of features showed rating of the similarity. The classification of particles is determined by identifying the closet standard. The coding of the semantic features of the morphological feature of wear particles was demonstrated to be useful for classification with statistical methods. The results showed that the presented method was satisfactory in solving practical problems.