Modeling Wind Speed Using Parametric and Non-Parametric Distribution Functions

Abstract

The variability of wind speed is modeled in the literature using probability distribution functions (PDFs). In many papers, the selection of PDFs is based solely on goodness of fit metrics without giving proper consideration as to whether these PDFs can model the wind speed range. A PDF's ability to model the wind speed range ensures that it can synthesize both the minimum and maximum wind speed at a site. A methodology to select PDFs that can be used to model wind speed is presented in this paper. The proposed methodology not only considers goodness of fit metrics when selecting PDFs but also their ability to model the wind speed range. The proposed methodology is compared with a commonly used methodology in the literature where the PDFs are selected based solely on goodness of fit metrics. Simulation results show that the proposed methodology chooses better PDFs than the commonly used methodology in the literature. Furthermore, it is shown that selecting PDFs using the commonly used methodology in the literature does not guarantee that the chosen PDFs will model the wind speed range.

Maximizing Returns and Minimizing Risks in Hybrid Renewable Energy Systems: A Stochastic Discounted Cash Flow Analysis of Wind and Photovoltaic Systems in Brazil

“…It is important to note that despite the numerous possibilities for representing wind and solar uncertainties [30][31][32], this study assumes a Gaussian probability distribution function with different parameters for each source. In Equation ( 3), the expected energy generation of the hybrid system is shown:…”