MODELOS LINEARES E NÃO LINEARES NA MODELAGEM DO PREÇO SPOT DE ENERGIA ELÉTRICA DO BRASIL / USING LINEAR AND NON-LINEAR APPROACHES TO MODEL THE BRAZILIAN ELECTRICITY SPOT PRICE SERIES

AUTOR(ES)
DATA DE PUBLICAÇÃO

2003

RESUMO

In this dissertation, modeling strategies are presented involving linear and non-linear time series models to model the spot price of Brazil`s electrical energy market. It has been used, among the linear models, the modeling approach of Box, Jenkins and Reinsel (1994) i.e., ARIMA(p,d,q) models, and dynamic regression. Among the non-linear ones, the chosen model was the STAR developed, initially, by Chan and Tong (1986) and, later, by Teräsvirta (1994). For this model, the Lagrange Multipliers test, to measure the degree of non linearity of the series , as well as to evaluate the estimated model was used. Moreover, it was also used a proposal for the initial values of the optimization algorithm, developed by Franses and Dijk (2000). The smoothed Kalman filter estimates were used in order to provide values for the spot price series during the energy shortage period.

ASSUNTO(S)

arima(p,d,q) arima(p,d,q) star model regressao dinamica modelo star dynamic regression

ACESSO AO ARTIGO

Documentos Relacionados