An intelligent multi objective optimization based supervised classification model and experimental analysis for the performance of a renewable hybrid system

Abstract

The current energy structure based on fossil fuels has become unsustainable due to the climate crisis, costs, and supply security, bringing hybrid renewable systems that combine solar and wind energy to the forefront. In this study, the electricity generation performance of a hybrid system comprising a monocrystalline photovoltaic (PV) panel and a micro wind turbine (WT) was examined in detail under the climatic conditions of Elazig province. During the experiments, wind speed, solar radiation, ambient temperature, relative humidity, PV panel surface temperature, and PV and WT power outputs were recorded every 10 min using a common time base to ensure a comprehensive, synchronized database. In the first stage, two different blade geometries were tested for the three-bladed micro WT: Type-2 blades with a length of 60 cm and a weight of 280 g, and Type-1 blades with a length of 50 cm and a weight of 200 g. The Type-2 blade enabled power generation at lower air speeds (3.1 m/s) and produced approximately 28% more power than the Type-1 blade. The most suitable wing type was selected based on the entire operating range, and the PV + WT hybrid system was tested under real outdoor conditions with this configuration. Thus, the temporal variation of solar and wind-based production, seasonal effects, and hybrid contribution were detailed. The extensive dataset obtained was analyzed using explainable artificial intelligence (XAI) based algorithms dependent on climatic parameters; transparent rules were obtained that describe how climatic parameters such as temperature, relative humidity, wind speed, and solar radiation affect both PV and WT power generation over the course of a year. Thus, hybrid system outputs were defined directly using understandable engineering statements in an if-then style, rather than classical regression coefficients. This comprehensive PV + WT + XAI approach, rarely reported in the literature, proposes a novel and reusable method for the design and operation of small-scale hybrid systems.