A novel two-stage interval prediction method based on minimal gated memory network for clustered wind power forecasting

As the global demand for clean renewable energy has grown, the contribution of wind power in grid systems has significantly improved. Wind power predictions play an important role in the stable and safe operation of grid systems. Considering the shortcomings of traditional wind power point predictions, a novel two-stage short-term wind power interval prediction method is proposed in this study. In the proposed method, the minimal gated memory (MGM) network and improved interval width adaptive adjustment strategy, which is an approach that is designed to adjust the prediction interval (PI) labels, are combined for short-term interval predictions of wind power. First, the point model for subsequence data based on the MGM network is proposed. Subsequently, the interval model is proposed to obtain the final PIs of wind power using the improved interval width adaptive adjustment strategy. Finally, with the purpose of verifying the prediction performance of the proposed model, two datasets and four representative benchmark models are implemented for comparative experiments. The superiority of the proposed model is demonstrated via experimental results, which show that the proposed model can obtain suitable wind power intervals with high confidence and quality.

Automated summary

A novel two-stage short-term wind power interval prediction method is proposed in this study, which uses the minimal gated memory network and improved interval width adaptive adjustment strategy for prediction. Experimental results demonstrate that the proposed model can obtain suitable wind power intervals with high confidence and quality.