A Decision Analysis Perspective on Multiple Response Robust Optimization
Abstract
Decisions in which multiple objectives must be optimized simultaneously occur frequently in government, military, and industrial settings. One method a decision maker may use to assist in such decisions is the application of a desirability function. An informed specification of the desirability function's parameters is essential to accurately describe the decision maker's value trade-offs and risk preference. This thesis uses utility transversality to analyze the implicit trade-off and risk attitude assumptions attendant to the desirability function. The desirability function does not explicitly account for response variability. A robust solution takes not only the expected response into account, but also its variance. Assessing a utility function over desirability as a means to describe the decision maker's risk attitude produces a robust operating solution consistent with those preferences. This thesis examines robustness as it applies to the desirability function in a manufacturing experiment example. Different levels of diplomatic, informational, military, and economic (DIME) instruments of national policy are investigated to examine their effect on the political, military, economic, social, infrastructure, and information (PMESII) systems of a nation. AFRL's National Operational Environment Model (NOEM) serves as a basis for identifying a robust national policy in a scenario involving the Democratic Republic of Congo.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 2012
- Accession Number
- ADA558279
Entities
People
- Jonathan S. Findley
Organizations
- Air Force Institute of Technology