DETERMINISTIC AND PROBABILISTIC WIND SPEED FORECASTING EMPLOYING A HYBRID DEEP LEARNING MODEL AND QUANTILE REGRESSION

Abstract

The negative effect of primary sources on the environment and the exhaustion of fossil fuels brings a serious challenge to human society. To address these issues, renewable energy sources (RESs) are considered to be the power of the future, and wind energy is expected to be an important part of this evolution. However, the uncertainty of wind speed has posed significant hurdles to wind power development. Thanks to the tremendous breakthrough in Artificial Intelligence, a novel method to handle this problem is proposed. This paper illustrates a wind speed prediction scheme based on Long Short-Term Memory (LSTM) neural network and Self-Attention mechanism (SAM) with different forecast horizon. In addition to conducting point forecast, the proposed model combined with Quantile Regression is employed to implement interval forecast. The predicted results on two given datasets demonstrate that the proposed model outperforms three predictive benchmark models.