<ul><li>Exploration of novel architectures using physics-based simulation presents challenges for optimization algorithms.</li><li>Surrogate-Based Optimization (SBO) algorithms, specifically Bayesian Optimization (BO) using Gaussian Process (GP) models, address the challenges.</li><li>Strategies are investigated for satisfying hidden constraints in BO algorithms, including rejection of failed points, replacing failed points, and predicting the failure region.</li><li>A mixed-discrete GP is found to achieve the best performance in predicting the Probability of Viability (PoV), demonstrated in solving a jet engine architecture problem.</li></ul>

Surrogate-based optimization of system architectures subject to hidden constraints

Discover more