- Industrial Session -
Tool Degradation Estimation with Ensemble Learning

Estimating tool degradation is crucial to prevent additional maintenance and production costs in any machining process. To address this challenge, we present a method to estimate tool degradation condition with few data using digital signal processing (DSP) and an ensemble learning framework. In this particular case, we deal with vibration data recorded from few drilling sessions. Because of the small number of data, we found out that using a single learner does not yield desirable estimates. This led us to investigate the potential of ensemble learning. By carefully selecting features using DSP and strategically combining several learners' results, we were able to obtain more accurate degradation estimation results even with a small number of training data.

PRESENTER

AI Research & Innovation Hub

Takahiro Kusunoki (Principal)
Jerelyn Co (Data scientist)
Mary Grace Malana (Data scientist)

APPLICATION