Hybrid Rocket Engine Design Using Pairwise Ranking Surrogate-assisted Differential Evolution

Abstract

Real-world optimization problems often require time-consuming fitness evaluations because of simulations or complex numerical calculations. A hybrid rocket engine (HRE) design problem is a time-consuming real-world application. HRE is a rocket engine that stores either the oxidizer or fuel in solid form and the other in liquid form. A surrogate model can reduce the number of expensive objective/constraint function calculations for such expensive optimization problems by searching for promising solutions. This study applies a new surrogate-assisted constraint-handling evolutionary algorithm, CHDE+ELDR, to solve the HRE design problem with fewer evaluations. CHDE+ELDR combines CHDE, a constraint-handling differential evolution, and ELDR, a pairwise ranking surrogate model. We conduct an experiment on the HRE design problem with limited evaluations and compare CHDE+ELDR with CHDE without a surrogate model. The experimental results show that CHDE+ELDR can consistently obtain high-quality HRE designs with fewer evaluations than simple CHDE.