Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
1736765 | Energy Reports | 2015 | 9 Pages |
The utilization of wind energy, as a booming technology in the field of renewable energies, has been highly regarded around the world. Quantification of uncertainties associated with accurate wind speed forecasts is essential for regulating wind power generation and integration. However, it remains difficult work primarily due to the stochastic and nonlinear characteristics of wind speed series. Traditional models for wind speed forecasting mostly focus on generating certain predictive values, which cannot properly handle uncertainties. For quantifying potential uncertainties, a hybrid model constructed by the Cuckoo Search Optimization (CSO)-based Back Propagation Neural Network (BPNN) is proposed to establish wind speed interval forecasts (IFs) by estimating the lower and upper bounds. The quality of IFs is assessed quantitatively using IFs coverage probability (IFCP) and IFs normalized average width (IFNAW). Moreover, to assess the overall quality of IFs comprehensively, a tradeoff between informativeness (IFNAW) and validity (IFCP) of IFs is examined by coverage width-based criteria (CWC). As an applicative study, wind speeds from the Xinjiang Region in China are used to validate the proposed hybrid model. The results demonstrate that the proposed model can construct higher quality IFs for short-term wind speed forecasts.