Série e Transformada de Fourier Aplicadas no Preenchimento de Falhas de Séries Temporais de Intensidade do Vento na Central Nuclear Almirante álvaro Alberto, Rio de Janeiro - Brasil

Abstract

In this work is proposed a methodology, based on the Fourier Series and Fourier Transform, as a gap filling technique in a time series of wind intensity from a meteorological tower located at Almirante Alvaro Alberto Nuclear Power Plant (CNAAA), Rio de Janeiro - Brazil. The time series comprises the period 1982-2001, with three periods without gaps. The periods ranged from July/1988 to September/1988, from October/1988 to December/1988 and from September/1999 to May/2000. The series were vectorially decomposed into zonal and meridional components, after which artificial gaps were generated. The Fourier series were used and later the coefficients of the Fourier transform were calculated. The cut limits were chosen for each spectrum of each component of the wind. The new coefficients were applied in the Inverse Fourier Transform for the fault times, generating predicted values of the vector wind components, to fill the artificial gaps. The forecasts were multiplied by gain factors so that their intensity was better adjusted. The three periods were statistically evaluated based on the linear regression model (MLR), linear correlation coefficient (r), curve fitting, standard error of estimation (EPE, ms -1 ), concordance index (r 2 ) and index of Willmot’s agreement and confidence index (c). The visual analysis of series reconstructions indicates that the method reproduces the wind intensities in a reasonably satisfactory way. In spite of this, the statistical analysis of index (c), in general, showed a poor performance of the wind components in the three periods, and may be due to the phase difference demonstrated in the reconstructions. The methodology can be applied, with caveats, to the other existing towers in the CNAAA, only being restricted the condition of transition of the flow regime between sea breeze circulation and katatic wind, followed by the gain factor for larger time series.