Super multi-step wind speed forecasting system with training set extension and horizontal–vertical integration neural network

Ling Liu and Jujie Wang

Applied Energy, 2021, vol. 292, issue C, No S0306261921003895

Abstract: With the rapid development of wind energy utilization, people pay more and more attention to improving the accuracy of wind speed prediction. However, the unstable and nonlinear nature of wind speed makes it difficult to predict accurately. Due to the accumulation of prediction errors, the prediction steps in many studies are very short. In this paper, a novel super multi-step forecasting system with structure of ”data extraction - horizontal multi-output prediction - vertical integration prediction” is constructed to make 7-day, 15-day and 30-day wind speed prediction without updating the data source. A new mutual information matrix is designed to select the decomposition number of the singular spectrum analysis. Two different types of bidirectional long short-term memory network are constructed to realize horizontal–vertical integration prediction. A new training set extension method is applied to improve the generalization ability of the neural network. Three wind speed data sources and five comparison models are used to evaluate the performance of the proposed system. The empirical results illustrate that: (a) the maximum mean absolute errors of prediction results of three sites are 1.371 m/s, 0.567 m/s and 0.580 m/s, respectively, which shows the model has better prediction performance; (b) the mean absolute errors of the three sites increased by -0.050 m/s, 0.079 m/s and 0.113 m/s, respectively, which shows the proposed system has no obvious accumulative errors; (c) the proposed forecasting system shows better performance than the comparison models.

Keywords: Super multi-step forecasting; Training set extension; Singular spectrum analysis; Long short-term memory (search for similar items in EconPapers)
Date: 2021
References: View references in EconPapers View complete reference list from CitEc
Citations: View citations in EconPapers (7)

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S0306261921003895
Full text for ScienceDirect subscribers only

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:eee:appene:v:292:y:2021:i:c:s0306261921003895

Ordering information: This journal article can be ordered from
http://www.elsevier.com/wps/find/journaldescription.cws_home/405891/bibliographic
http://www.elsevier. ... 405891/bibliographic

DOI: 10.1016/j.apenergy.2021.116908

Access Statistics for this article

Applied Energy is currently edited by J. Yan

More articles in Applied Energy from Elsevier
Bibliographic data for series maintained by Catherine Liu ().