Evaluation-aware optimizer improves tuning under strict budget limits
REMBGO for Evaluation-Limited Optimization: CEC2022 Benchmarking and EAF-STATCOM Generalization
Neural and Evolutionary Computing
Summary
Optimization methods often assume they can try many options, but in real-world engineering, evaluations are expensive and limited. The authors present REMBGO, which refines an existing method to work better when only a small number of evaluations are allowed. They tested REMBGO on standard benchmark problems and a power controller tuning task, showing that its benefits depend on how many evaluations are possible. REMBGO is not always better but is designed to work well when resources are tight.
What this means in practice
- •For control engineers: Tune controller parameters for power systems efficiently when simulation evaluations are costly and limited.
- •For optimization practitioners: Optimize complex engineering problems under strict limits on function evaluations using budget-aware metaheuristics like REMBGO.
Authors
Vikram Chekuri, Narapareddy Ramarao, Prashanth N A
Abstract
Metaheuristic comparisons are often reported at evaluation budgets far larger than an expensive engineering simulation can afford. This paper presents REMBGO, an evaluation-budget-aware refinement of EMBGO. Its contribution is not any single operator: it is the integration of population preservation, scale-normalized exploration and recovery mechanisms whose parent-relative benefit is concentrated under restricted function-evaluation budgets. REMBGO is evaluated on CEC2022 at a primary, application-aligned budget of 1040 function evaluations, with a surrounding budget ladder reported descriptively to characterise budget dependence rather than as a family of controlled hypotheses. A component-removal study finds no individual mechanism whose removal survives multiplicity correction, consistent with an integrated rather than a single-operator contribution. The algorithm is then applied to EAF--STATCOM controller tuning, and the resulting frozen gain vectors are transferred without retuning to twenty pre-specified unseen stochastic arc realizations ($210$ fixed-gain simulations, all feasible). The selected gain-vector main effect accounts for only $3.6$--$5.8\%$ of the crossed validation variation, against approximately $61.4\%$ for interaction plus residual, and the matched REMBGO--EMBGO paired differences are controller-dependent in sign and not statistically significant. REMBGO is therefore presented as a resource-aware refinement whose benefit must be stated together with the evaluation budget, not as a universally superior optimizer.