A Hybrid Optimization Strategy Using Design-Space Evolution and POD-based Order Reduction
Satyajit S. Ghoman, Darius Sarhaddi, Ping Chen, Zhicun Wang, Rakesh K. Kapania · 12th AIAA Aviation Technology, Integration, and Operations (ATIO) Conference and 14th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference · 2012
This paper describes the development of an innovative hybrid optimization strategy that employs an evolutionary algorithm technique combined with a Proper Orthogonal Decomposition-based reduced order design-space scheme. The proper orthogonal decomposition method of extracting dominant modes from an ensemble of candidate congurations is used for the design space order reduction wherein the reduced number of dominant POD coecients act as the new shape design variables instead of large number of actual physical design variables. The snapshot of candidate population is updated iteratively using tness-driven evolutionary algorithm technique of health-based retention and modication. Established with the goal of developing a robust optimization framework, the proposed strategy capitalizes on the advantages of evolutionary algorithm as well as PODbased reduced order modeling, while overcoming the shortcomings inherent with these techniques. When linked with a streamlined multi-disciplinary optimization framework, this hybrid optimization strategy oers a computationally ecient methodology for problems having a high level of complexity with challenging design space that has a large number of physical design variables. The developed framework is demonstrated for its robustness on a non-conventional supersonic tailless air vehicle wing shape optimization problem.