Search papers, labs, and topics across Lattice.
This study investigates a surrogate-assisted optimization framework for designing a current-excited coil-core system, integrating a finite-element method (FEM) model with a Gaussian-process (GP) surrogate to navigate complex design constraints. The research reveals that the effectiveness of optimization methods, such as expected improvement (EI) Bayesian optimization, COBYLA, and BOBYQA, varies significantly depending on the evaluation budget, with BOBYQA achieving the highest mean terminal response. Importantly, the findings indicate that different optimization strategies excel under varying conditions, emphasizing the need for tailored approaches in simulation-driven design processes.
Optimization strategies yield different performance outcomes based on evaluation budgets, challenging conventional assumptions about their effectiveness in coil design.
This work evaluates surrogate-assisted optimization of a seven-parameter current-excited coil--core benchmark subject to geometric, manufacturing, and separate core and copper mass constraints. A Python--MPh--COMSOL workflow couples a two-dimensional axisymmetric finite-element method (FEM) model to a Matern 5/2 Gaussian-process (GP) probabilistic surrogate. Here, physics-constrained denotes a design problem evaluated by a governing-equation FEM model and restricted by explicit physical, geometric, manufacturing, and material-allocation constraints; it does not denote a physics-informed GP architecture. Sequential Bayesian optimization (BO) ranks candidates using expected improvement (EI), and every reported incumbent is verified by FEM. Five paired runs show that optimizer ranking depends on the available FEM-evaluation budget: EI--BO improves rapidly at small continuation budgets, COBYLA is stronger at the earliest checkpoint, and BOBYQA attains the highest mean terminal response. A retrospective finite-pool study further finds no robust endpoint advantage of EI over posterior-mean ranking on this smooth response surface. The broader result is that early progress, terminal response, information use, and wall-clock cost can favor different methods in simulation-driven design. A selected-design check at a common total current preserves the observed BOBYQA--COBYLA--EI-BO ordering. The conclusions nevertheless remain conditional on this axisymmetric benchmark and do not establish a fixed-current optimum, fixed-power performance, or electrical-efficiency superiority.
