Artificial neural network‐based Sobol algorithm for multi‐objective optimization of renewable energy supply in buildings: A transient approach.

Summary: The need to increase the share of renewable energy supply in buildings has been a subject of concern for many authors. Renewable energies can offset building electrical and thermal energy loads, thus reducing constrains on national grid networks. The aim of this paper is to construct, model and optimize a hybrid renewable energy supply system under transient conditions. The supply system consists of solar‐PV panels and vertical axis wind turbines, while batteries are used for energy storage. TRNSYS software is used to model the energy demand of an office building that operates under the average standard building as specified in EnergyPlus software. In order to determine the optimal size of the energy supply system an Artificial Neural Network (ANN) combined with a Sobol algorithm are employed. Two objective functions are developed to: (a) minimize the cost of purchasing solar‐PV panels and wind turbines, and (b) minimize the electricity required from the national grid network. The simulation and optimization results show that the best condition is achieved with 461 panels and one vertical axis wind turbine. At this best condition, the amount of electricity required from the national grid network is 5.55 MWh/year, and this will cost 2.4 $/h. Moreover, 3% of electricity will be obtained from the grid, 31% from batteries and 66% are supplied directly through panels and turbines. [ABSTRACT FROM AUTHOR]