Online Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 354 - Experimental Design and Reliability
Type: Contributed
Date/Time: Thursday, August 12, 2021 : 10:00 AM to 11:50 AM
Sponsor: Section on Physical and Engineering Sciences
Abstract #318726
Title: Optimal Design of Experiment for Validation of Complex Computer Models from Operational Data
Author(s): Ethan Davis* and Jon Stallrich
Companies: North Carolina State University and North Carolina State
Keywords: Sensor placement; Experimental design; Non-linear model; Sampling theory; Finite Element Model; Ordinary Differential Equations

Computational models comprised of systems of ordinary differential equations (ODEs) are used in a variety of domains to represent real world systems. These systems, which are often derived through subject matter expertise, require validation using data from operational use of the represented system. Given the high dimensional, non-linear setting, proper instrumentation of systems for validation testing poses unique and unanswered design challenges. Despite the importance of validating ODE computational models, little work has been performed to compare design construction measures with respect to their ability to recover model parameters. We discuss existing design measures from a statistical perspective and explore their efficacy using a simulation study of a Kirchoff-Love thin plate. Borrowing ideas from classical design theory and high dimensional sampling theory, we propose two algorithmic improvements to current ODE non-linear optimal design. We also present a simulation-based method to determine the number of sensors needed for validation testing of ODE computational models using operational data.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2021 program