A two-stage wind speed model for multiple wind farms considering autocorrelations and cross-correlations

Abstract

Collected wind speed time series (WSTS) has three major characteristics: randomness, autocorrelation and cross-correlation, which have significant effects on the wind speed modeling for power systems containing wind energies. Most WSTS models only consider some of the above characteristics, which may significantly reduce the computation accuracy on the analysis of wind-integrated power systems. This paper presents a two-stage model for WSTS at multiple wind sites. This model considers the wind speed autocorrelation for each WSTS in the first stage, and wind speed cross-correlation for all WSTSs in the second stage. The inverse transformation is used to derive the analytical correlation relationship between multiple WSTSs and multiple time series of normal distribution (TSND). Then modeling multiple WSTSs with given correlations can be done by building multiple TSNDs that meet appropriate autocorrelations and cross-correlations using an autoregressive model. Case studies demonstrate that the proposed model is capable of simulating WSTS with higher accuracy than the improved correlation method, the time-shifting technique, and the Copula method.