Ability of Black-Box Optimisation to Efficiently Perform Simulation Studies in Power Engineering

Abstract In this study, the potential of so-called black-box optimisation (BBO) to increase efficiency simulation studies in power engineering is evaluated. Three algorithms (“Multilevel Coordinate Search” (MCS) and “Stable Noisy Optimization by Branch Fit” (SNOBFIT) Huyer Neumaier “blackbox: A Procedure for Parallel Expensive Black-box Functions” (blackbox) Knysh Korkolis) are implemented MATLAB compared solving two use cases: analysis maximum rotational speed a gas turbine after load rejection identification transfer function parameters measurements. The first case has high computational cost, whereas second computationally cheap. For each run algorithms, accuracy found solution number simulations or evaluations needed determine optimum overall runtime used identify comparison currently methods. All methods provide solutions optima that at least 99.8% accurate reference objective functions differs significantly but cannot be directly as only SNOBFIT algorithm does stop when not improve further, other predefined evaluations. Therefore, shortest both examples. expensive simulations, it shown parallelisation (SNOBFIT blackbox) quantisation input variables essential algorithmic performance. overspeed analysis, can compete with procedure concerning runtime. Further will have investigate whether applied BBO outperform problems higher dimensionality.