Prepared by: Ali Farhang Mehr
Group Lead: Irem Tumer
Complex Systems Design Group
Title: Planning and Design of Experiments for Measurement of Physical Parameters
Application Context: The proposed approach is directly applicable when the distribution
of a physical parameter should be approximated or modeled. For example:
Example 1: To find the temperature distribution of upper
troposphere/lower stratosphere (that directly affects flight missions).
Example 2: To approximate the concentration and density distribution
in the Atlantic Ocean (that can help model the Thermohaline
Circulation). etc.
Objective: The objective of the proposed research is to identify the 'best' points, in terms
of location and number, where measurements must be made so that 'the most accurate'
approximation model can be obtained for a physical phenomenon.
Approach: An information-theoretic approach is used to quantify the expected 'worth of
information' obtained by conducting a particular measurement (i.e. experiment). An
interim Bayesian approximation model is then built that can be used to estimate
parameter distribution based on the information at hand. This can be used, in a sequential
manner, to plan the next experiments such that 'maximum information is obtained about
the unknown distribution of the physical parameter with minimum number of
experiments'.
Our Related Research Articles:
"Bayesian Meta-Modeling of Engineering Design Simulations: A Sequential Approach
with Adaptation to Irregularities in the Response Behavior", International Journal for
Numerical Methods in Engineering Vol. 32(4), pp. 491-514
"Meta-Modeling of Multi-Response Engineering Simulations", 10th AIAA/ISSMO
Multidisciplinary Analysis and Optimization Conference, Albany, NY