Quick et al. (2023) present Stochastic Gradient Descent (SGD) in wind farm optimisation and compare it to conventional deterministic gradient-based optimisation. The SGD method estimates the annual energy production using a Monte Carlo simulation that randomly samples the input distributions of atmospheric conditions (wind speed and direction). SGD reduces the computational time compared to the deterministic model as it does not require all atmospheric conditions to be evaluated for every optimisation iteration. The SGD approach can naturally incorporate uncertainty quantification and is ideal for large-scale wind farm optimisation. It is readily accessible in the open-source TOPFARM package.

Quick, Julian, Pierre-Elouan Rethore, Mads Mølgaard Pedersen, Rafael Valotta Rodrigues, and Mikkel Friis-Møller. 2023. “Stochastic Gradient Descent for Wind Farm Optimization.” Wind Energy Science 8 (8): 1235–50. https://doi.org/10.5194/wes-8-1235-2023.