An Accurate, Light-Weight Wind Speed Predictor for Renewable Energy Management Systems

Saira Al-Zadjali, Ahmed Al Maashri, Amer Al‐Hinai, Sultan Al-Yahyai, Mostafa Bakhtvar

Energies · 15 November 2019

In one sentence. Lightweight perturbed-observation ensemble neural network nowcasts wind speed at masts, cutting NRMSE by 20.9% and running an order of magnitude faster than rivals.

Abstract

This paper proposes an approach for accurate wind speed forecasting. While previous works have proposed approaches that have either underperformed in accuracy or were too computationally intensive, the work described in this paper was implemented using a computationally efficient model. This model provides wind speed nowcasting using a combination of perturbed observation ensemble networks and artificial neural networks. The model was validated and evaluated via simulation using data that were measured from wind masts. The simulation results show that the proposed model improved the normalized root mean square error by 20.9% compared to other contending approaches. In terms of prediction interval coverage probability, our proposed model shows a 17.8% improvement, all while using a smaller number of neural networks. Furthermore, the proposed model has an execution time that is one order of magnitude faster than other contenders.

Labels

Labels and summary are generated by Claude and reviewed weekly. Wrong? Edit this entry.