ME 4550: Computational Uncertainty Quantification for Engineering Applications

Course Description

This course focuses on a probabilistic (Bayesian) treatment of uncertainties in modeling a system's behavior, specifically adapted to mechanical engineering and bioengineering problems. This treatment extends to both making predictions under uncertainty (predictive modeling) as well as updating our knowledge about the system model using data/measurements (Bayesian inference). The course will emphasize both (i) applications with physical models as well as (ii) applications with statistical models build entirely based on data/inference. The course will start by briefly reviewing the foundations of probability as a multi-valued logic that quantifies all our available knowledge about a real system and its environment. This leads to a rigorous meaning for the probabilistic model for a system. The main part of the course focuses on computational tools for (i) predictive analysis (i.e., uncertainty propagation) and (b) Bayesian system estimation/identification (i.e., model updating based on experimental observations). (OC).