Wind speed time series analysis using TBATS decomposition and moving blocks bootstrap

Abstract

The nature of the so called renewable energy (i.e. wind, solar, wave and tidal) led studies to refer to it as a stochastic generation processes and its dynamic behavior strengthens the need for robust and accurate analysis tools integrated to energy systems. The steadily increasing penetration of wind energy around the world raises a demand to better understand the behavior of historical data to infer indispensable information for the future market. This paper proposes an approach to analyze a wind speed time series using a recent decomposition method called TBATS and Moving Blocks Bootstrap for scenario generation. Results indicated complex multi-seasonality patterns as the simulated scenarios could represent statistically the mean; however, only 75% of the variance on the best scenario was explained, suggesting nonlinearity.