Genetic programming for photovoltaic plant output forecasting

• Published in: Solar energy Vol. 105; pp. 264 - 273
• Main Authors: Russo, M., Leotta, G., Pugliatti, P.M., Gigliucci, G.
• Format: Journal Article
• Language: English
• Published: Kidlington Elsevier Ltd 01.07.2014; Elsevier; Pergamon Press Inc
• Subjects: Applied sciences; Artificial intelligence; Artificial neural network; Distributed computing system; Energy; Energy efficiency; Exact sciences and technology; Genetic algorithms; Genetic programming; Hybrid models; Mathematical models; Natural energy; Photovoltaic cells; Solar energy; Italy; Europe; Artificial neural network; Hybrid models; Artificial intelligence; Distributed computing system; Genetic programming; Programming; Neural network; Distributed system; Distributed computing; Forecasting; Single crystal; Distributed processing; Accuracy; Genetic algorithm; High efficiency; Solar energy; Cloudiness; Hybrid model; Silicon; Mathematical model; Comparative study
• ISSN: 0038-092X; 1471-1257
• DOI: 10.1016/j.solener.2014.02.021

•Using genetic programming we identified models for the power output of a string.•Only the actual power and the time difference with respect to sunrise are necessary.•74 inputs were investigated, but no cloud information was considered. In this paper we have identified several mathematical models for predicting the solar power output of a 1.05kWp Monocrystalline Silicon high-efficiency photovoltaic string located at the ENEL Catania site, Italy. The data we used corresponds to 15min of averaged power generated over a whole year (2010). A tool named the Brain Project was used. It follows a distributed genetic programming approach. Seventy-four inputs were investigated for our purposes, but no cloud information was considered. The accuracy of all the models was evaluated and compared to other approaches. Among these, the simpler models, that foresee only two inputs perform similarly to our more complex models and to several others in literature.