Comparação do desempenho de diferentes modelos de distribuição de probabilidade aplicados a dados de velocidade de vento onshore e offshore na região do nordeste brasileiro

• Autor(a) principal: Lins, Davi Ribeiro
• Data de Publicação: 2021
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da Universidade Federal do Ceará (UFC)
• Texto Completo: http://www.repositorio.ufc.br/handle/riufc/57995
• topic: Energia eólica offshore; Modelagem da velocidade de vento; Distribuições de velocidade do vento não convencionais; Análise estatística de múltiplos critérios

The objective of this dissertation is to evaluate the performance difference presented by different distribution models when applied to the onshore and offshore wind speed data, where the onshore data were measured at two stations located in Northeast Brazil, and the offshore data were measured by two ocean buoys located in the South Atlantic. Five distributions were used to model wind speed, namely: Weibull Distribution (W), Nakagami Distribution (N), Extended Generalized Lindley Distribution (EGL), Generalized Gamma Distribution (GG) and Generalized Extreme Value Distribution (GEV). The distribution parameters were estimated using the following numerical methods: Maximum Likelihood Method (MLE), Modified Maximum Likelihood Method (MMLE) and the Method of Multi-objective Moments (MUOM). In addition, three goodness of fit tests were used to choose the model that presented the best fit to the data, namely: Kolmogorov-Smirnov test (KS), Asymmetry Deviation and Kurtosis (DSK) and the Akaike Information Criterion (AIC). The results of these tests were normalized and unified in a parameter called Total Error (ET), where the closer to zero this ET the better the accuracy of the distribution. In both onshore locations, due to the pattern presented by the results, the only conclusion obtained was that the three parameters distributions EGL and GG were, in general, superior to the other distributions. In the offshore wind data from buoy 1, the EGLMLE (ET = 0), WMUOM (ET = 0.085504) and GGMLE (ET = 0.130876) distributions showed the best precision in adjusting the histogram. According to the results of buoy 2, the best precision was obtained by the EGLMLE (ET = 0), GGMLE (ET = 0.052179) and WMUOM (ET = 0.058615) distributions. For all four locations, the distribution GEVMMLE presented the highest Total Errors (ET) and, consequently, the worst precisions. According to the results, there is no single distribution model that is more appropriate to fit a set of wind speed data, and it is always necessary to apply a study to know which of the available distributions is the most appropriate for the case in question, being also necessary to verify which parameter estimation method is the most suitable for a given distribution.