The applicability of zero inflated beta distributions for stochastic modeling of PV plants' power output

Abstract

Uncertainty of generation is the inherent property of renewable energy sources such as photovoltaic panels. Issues such as optimal choice of equipment, energy management, and reliability of energy supply in distribution grids and microgrids should take into account this uncertainty. Usually this type of uncertainty is modeled by setting distributions of climate conditions (solar irradiation, temperature, wind speed). The specific values of power output in this case are determined by weather conditions and specifications of equipment. Such approach demands huge amount of systematic meteorological observations. This paper investigates the setting of generation output for photovoltaic plants without distributions of weather conditions. The proposed method assumes the use of hourly average generation data. The data are divided into samples. The samples are fitted by degenerated, beta or zero inflated beta distributions. Degenerated and beta distributions are presented as special cases of the more general zero inflated beta distributions. Correctness of fitting the distributions is verified by Kolmogorov-Smirnov tests.