This Research aims at early detection of impending failures during tribological experiments e.g., pitting, recognition of characteristic features and their evolution over time. The objective of this research is the development of an ML algorithm for the real-time classification of the state of operation as well as the detection and prediction of impending failures in a defined laboratory experiment and Implementation of trained algorithm in a real-time data acquisition (DAQ) system.

In addition to early-stage failure prediction, AI and machine learning approaches will also be used to predict the probability of a tribo-layer formation, which strongly depends on the interaction between the involved materials and lubricants as well as the different operating conditions. Furthermore, AI and ML methods are useful to support the characterization and classification of surface topography and/or the state of lubrication during a tribological experiment and thus may provide helpful data for the design process of dry and lubricated machine elements.

Rollers of a tribometer with simplified data path, as well as visualized results of a test on the tribometer

© Surya Kannan Peesapati