A BAYESIAN APPROACH FOR THE OPTIMAL INTEGRATION OF RENEWABLE ENERGY SOURCES IN DISTRIBUTION NETWORKS OVER MULTI-YEAR HORIZONS

We propose a method to optimise the planning strategy of an active distribution network. The problem is formulated as the search for the planning strategy parameters minimising antagonist objectives. These objectives are computed using a numerical simulator of the distribution networks and stochastic scenarios. Since simulations take a high amount of CPU time, we suggest using Bayesian optimisation algorithms, where the costs are modelled with Gaussian random processes. The main idea is to compute predictions of the costs and uncertainty intervals, which are then used to guide the optimisation algorithm. A case study illustrates the performance of the proposed method.