Fitting methods of two-parameter Weibull of wind series and Electric-wind potential estimation.

Abstract

The aim of the present work is to compare interpolation methods of the probability density function with Weibull distribution. This process is fundamental for the wind resource analysis through the Electric-wind potential estimation. The information of this study corresponds to temporal series of wind data from 7 meteorological stations, located in the Provinces of Galápagos, Carchi, Tungurahua, Bolívar, and Loja. These series were processed by the interpolation methods of linearization, graphical, moments and maximum likelihood estimation (MLE). The comparison is based on the interpolation graphics with the relative frequencies of wind data, and the residual analysis according to regression methods. Moreover, it analyzed the contrast of the mean of wind speed, mean square error (MSE) and correlation index, among the methods studied. This analysis is the basis for the Electric-wind potential estimation, according to the methodology proposed by Jijón et al. [1], based on the wind mean calculated by each method. In the results of this work, the MLE method had the lowest MSE and the highest correlation index by the most wind data series analyzed; and the linearization method had the lowest mean. In consequence, the Electric-wind potential shows the high sensibility under the mean of wind speeds calculated by every method. According to the limitation identified in each method, it is recommended that this type of analysis might be replicable to study prioritized places for wind farms.