Research Seminar on AI: Predictive Quality in Manufacturing - Can AI replace Metrology?
- +49 241 80 28394
- E-Mail schreiben
Dienstag, 16.11.2021, 16.00 Uhr
The objective of Predictive Quality is to optimize product and process-related quality characteristics by data-driven forecasts in manufacturing. From the many aspects of the research field, the talk focuses on predicting quality characteristics from process data. An important facet is the proof of suitability (i.e., uncertainty quantification) before any machine learning model can be trusted in production environments.
Simon Cramer will present the potentials and challenges of Bayesian models to replace/complement physical measurements for quality inspections. This class of models can differentiate between aleatoric and epistemic uncertainty, which are prime indicators for model calibration and confidence in the prediction. After successful interpretation in the sense of metrology, Bayesian models are top candidates for the broad application in industry.
Simon Cramer is a doctoral researcher with Prof. Schmitt at the Laboratory for Machine Tools and Production Engineering (WZL) since 2019. He received his M.Sc. and B.Sc. degrees in Computational Engineering Science (CES) from RWTH Aachen University. His research focuses on applying Bayesian Neural Networks to predict quality characteristics from process data and the application in industry.