Optimal Sizing of Hybrid AC/DC Distributed Generation Systems with On-Site Renewables for Off-Grid Power Supply to Remote Areas

This study proposes a self-contained hybrid power system with DC and AC voltage buses connected to renewable and conventional energy sources to supply power to remote areas. The conventional sources are meant to serve as a backup for renewables to meet the demand during power shortages in the system. This research focuses on integrating an optimal structure of an off-grid system consisting of PV and WT generators with a battery energy storage system (BESS) and a diesel generator set (DGS) to ensure high reliability of power supply to a remote area. We contribute a machine learning algorithm (TLBO) for optimal sizing of the installed capacity of the PV-WT-BESS-DGS hybrid system components. The algorithm relies on the original multi-objective function to determine the most advantageous values of unit capacities for PV and WT plants, as well as the number of batteries and diesel generators, to meet the demand in the hybrid system under different scenarios of PV and WT power shortage. The optimization problem is solved considering the variability of meteorological conditions as manifested in the random and uncertain intermittency of PV and WT generation as well as in the stochastic nature of the electricity consumption process. To enhance the efficiency of its solution, this study seeks to cluster these processes and identify common scenarios for the data set of hourly measurements of PV and WT generation and loads during the year. We use k-means clustering to determine the number of clusters for PV, WT, and loads for representative geographical areas. The study reports the results of the analysis of optimal configurations for a hybrid AC/DC system made up of PV, WT, BESS, and DPS in these areas.