Research Summaries

Back A New Mathematical Framework for Design Under Uncertainty

Fiscal Year 2014
Division Graduate School of Operational & Information Sciences
Department Operations Research
Investigator(s) Royset, Johannes O.
Sponsor Defense Advanced Research Projects Agency (DoD)
Summary The objective of this proposal is to overcome the challenges in engineering design under uncertainty by developing a new theoretical framework based on stochastic approaches for risk-averse design and optimization. To this end, we propose to combine stochastic multi-fidelity models and other sources of information about system performance (including experimental data and/or expert opinions) using novel data fusion and machine learning approaches. This will lead to a comprehensive description of system performance with less uncertainty than in the case of single-source/model analysis, it will allow the exploration of broader design spaces, and it will enable us to endow final designs with rigorous certificates of quality. None of the methods in current engineering practice can consistently account for the errors and uncertainties present in the complex sequence of analyses embedded in the design process, and, furthermore, key statistical relationships among those analyses are usually ignored. In addition, all decisions regarding the order in which to use available analysis and optimization tools are usually based on experience and intuition, which often yields under-performance and inefficient use of available resources. The multi-information-source optimization framework we propose allows us to decide on the timing and execution of a large ensemble of predictive tools including stochastic simulations, assimilation of noise-corrupted experimental data, and subject-matter expert opinions.
Keywords
Publications Publications, theses (not shown) and data repositories will be added to the portal record when information is available in FAIRS and brought back to the portal
Data Publications, theses (not shown) and data repositories will be added to the portal record when information is available in FAIRS and brought back to the portal