*********************************
There is now a CONTENT FREEZE for Mercury while we switch to a new platform. It began on Friday, March 10 at 6pm and will end on Wednesday, March 15 at noon. No new content can be created during this time, but all material in the system as of the beginning of the freeze will be migrated to the new platform, including users and groups. Functionally the new site is identical to the old one. webteam@gatech.edu
*********************************
Austin Gabhart
(Advisor: Prof. Dimitri N. Mavris]
will propose a master’s thesis entitled,
Method for Robust Design Optimization of Systems-of-Systems (MRDOSS)
On
Tuesday, July 19 at 12:00 p.m. EDT
Collaborative Visualization Environment (CoVE)
Weber Space Science and Technology Building (SST II)
or
Abstract
Uncertainty is inherent in any design problem. Its significance increases when the problem involves probabilistic interactions or a lengthy design cycle. Uncertainty has a particular significance in Systems-of-Systems (SoS) analysis, where the designer must predict how the capability of many constituent systems might advance with time. These challenges have led engineers to develop the concept of robust design to minimize the effect of these uncertainties on the performance of the system being designed. However, in the literature, these approaches tend to be limited to one source of uncertainty, either ranges of input values or stochastic outputs, and have limited capabilities to ensure a feasible output.
The proposed work seeks to characterize the problem space of robust design optimization for multiple sources of uncertainty and present a method select robust feasible designs. The proposed method evaluates a large set of designs through a random design of experiments for both performance and robustness metrics. The infeasible designs are then filtered out, and a multi-attribute decision-making approach is applied to the remaining designs. Since some robustness metrics require Monte Carlo methods to evaluate, standard multi-objective optimization methods can become computationally costly; the implementation of random design of experiments is expected to mitigate this effect. The proposed method will be compared to multi-objective optimization techniques to ensure accuracy and efficiency. This comparison will be conducted with a hypersonic mission SoS. The problem characterization will utilize the same SoS to evaluate the variation in the Pareto Frontier as robustness metrics are added as objectives. The intended contribution of this thesis is to provide a computationally efficient methodology to select robust designs while ensuring feasibility and satisfying requirements.
Committee
