Session: CIE-02 / 03 / 04 AMS: Joint Topics

Paper Number: 141592

141592 - Characterizing Model Fidelity Through a Set-Based Approach 

Simulation models are often used in design to predict system performance for several use cases including trade space exploration, decision-making, and validation and verification. Models are abstractions of reality and do not contain all the phenomena and details in the real world. During the model development process, decisions are made regarding what parameters are important and must be included in the model, the physical phenomena included and not included in the model, and the mathematical representations of the physical phenomena. This fact brings many concerns, such as "How can I trust this model?" or "How should I choose between models?". In modeling and simulation (M&S), the concept of fidelity explains how a model differs from reality. This paper proposes a set-based definition of fidelity based on the reduction of information through several domains throughout the model development process. The proposed approach differs from uncertainty and accuracy-based definitions of model fidelity. The set-based approach for defining model fidelity is demonstrated based on a US Department of Defense ground vehicle gradeabilty test.  An example of a ground vehicle conducting a gradeability test demonstrates the reduction in information from reality, the known world, testing, modeling, and simulation. Overall, this set-based fidelity approach bridges disparate definitions of fidelity and creates a greater understanding of how models reflect reality.

Presenting Author: Evan Taylor Clemson University

Presenting Author Biography: Evan is a PhD student at Clemson University. His research is focused on establishing a systematic method for computing model fidelity and developing approaches to select the appropriate model to support design decisions.

Authors:

Evan Taylor Clemson University
Edward Louis Clemson University
Gregory Mocko Clemson University
Gang Li Clemson University
Evan Hybl Clemson University