An Automated System for Surface Damage Detection Using Support Vector Machine.

The global objective of this paper was to build an automated prediction system for surface damage. Practically, the damage initiates from the free surface because of the high-stress concentration that presents in valleys of the surface profile. Hence, the surface condition is a major factor in the fatigue strength of the metal. In this paper, the surface condition has been measured using an optical confocal measurement system (Alicona). Arithmetical mean height and Surface Flatness have been selected as input data source for the machine learning model. The machine learning model was built using the Support Vector Machine method. The role of this model is to select the best surface parameters to detect surface damage. The results show that the Surface Flatness parameter provides better prediction for surface damage than the Arithmetical mean height parameter. [ABSTRACT FROM AUTHOR]