Wear analysis in HFRR test samples using image processing.
wear, lubricity, image processing, Artificial Neural Networks.
Tribological tests are developed as a controlled way to evaluate the wear mechanism acting between metallic surfaces, as well as to observe the influence of the type of lubricant used. The lubricity assessment of a lubricant is standardized by the HFRR (High Frequency Reciprocating Rig) test, which is given by the sphere-disc tribological system in lubricated contact, and it produces as a result, images from the wear and the scar diameter is extracted, defining the Wear Scar Diameter - WSD. From a set of samples of different fuels applied as lubricants, images of the worn surfaces were obtained. Thus, it is proposed in this work to explore other characteristics of the images in addition to the WSD, thus
allowing a better description of wear and the type of lubricant used. With the images acquired in the HFRR test, image processing techniques were applied using the Matlab software and the OpenCV library to obtain quantitative
parameters. From this information, an Artificial Neural Network was able to classify new images according to the type of fluid applied in the lubrication, demonstrating the use of artificial intelligence to identify and classify wear patterns
from the analysis of their images.