Quantification of Aleatoric and Epistemic Uncertainty in Computational Models of Complex Systems

Abstract

For complex engineering systems, testing-based assessment is increasingly sought to be replaced by simulations using detailed computational models. This is due to a lack of experimental data and/or resources to conduct these experiments at the system level. Components, which are part of the system, are usually cheaper to build and test relative to the system itself. The availability of component data coupled with the lack of system data and the complexity of the system being model leads to a need to build models in a building-block or hierarchical manner. This approach takes advantage of data at the component level by guiding the development of each component model. These models are then coupled to form the system model. Quantification of uncertainty in a system response is required to establish the confidence in representing the actual system behavior. To be accurate, this quantification needs to include both aleatoric uncertainty (due to natural variability) and epistemic uncertainty (due to lack of or incomplete knowledge). This paper proposes a framework based on Bayes networks that uses the available data at multiple levels of complexity (i.e. components, subsystem, etc) and allows quantification and propagation of both types of uncertainty in a system model prediction. A method to incorporate epistemic uncertainty given in terms of intervals on a model parameter is presented and a numerical example demonstrating the approach is shown. These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 6 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.