A Multi-Fidelity Kriging-Based Experiment Optimization Framework with an Augmented Lagrangian Method for Distributed Optimal Design

Distributed optimal design brings significant solutions for experiment optimization in complex engineering design problems. A Kriging-based augmented Lagrangian Method is proposed with the help of the Multi-fidelity Hamiltonian Kriging (MHK) surrogate model. The Multi-fidelity Hamiltonian Kriging-based Augmented Lagrangian Method (MHK-ALM) uses subsystem surrogate models constructed from multi-fidelity data to speed up the inner loop solution of ALM, while also reducing the iterations of the outer loop of ALM. The MHK-ALM is illustrated with one numerical simulation of a multi-fidelity constrained NASA speed reducer problem, demonstrated with a multidisciplinary design optimization of a solid-propellant ballistic missile. The engineering application of the multidisciplinary design optimization (MDO) problem shows that the proposed method can perform precisely over certain advanced surrogate-based optimization frameworks. The MHK-ALM can be applied for any other distributed optimal design problems where one need complex subsystem decomposition.