Super-Resolution Model Based on Wavelet Domain Loss for Improving Wind Speed Forecasting Accuracy 


Vol. 13,  No. 12, pp. 710-718, Dec.  2024
https://doi.org/10.3745/TKIPS.2024.13.12.710


PDF
  Abstract

This study proposes a novel GAN-based Super-Resolution (SR) model leveraging wavelet domain loss to improve the accuracy of wind speed prediction. The proposed model integrates an enhanced generator structure into the existing Wavelet-Guided Super-Resolution (WGSR) framework to transform low-resolution wind speed images into high-resolution representations. The generator is composed of three modules: feature extraction, feature refinement, and reconstruction. Dense blocks and Channel Instance Residual Blocks (CIRB) with instance normalization are incorporated to reduce distortion and improve reconstruction performance. Experimental results using the NREL WIND Toolkit dataset demonstrate that the proposed model achieves performance improvements of 9.57% in PSNR, 21.20% in SSIM, and 21.05% in RMSE compared to the original WGSR’s RRDBNet. This indicates that the model effectively enhances wind speedimages from 10 km per pixel resolution to 2 km per pixel resolution, suggesting its potential applicability in weather forecasting and energy management applications.

  Statistics
  Cite this article

[IEEE Style]

W. Jo, S. Park, Y. Kim, S. Jung, C. Sim, "Super-Resolution Model Based on Wavelet Domain Loss for Improving Wind Speed Forecasting Accuracy," The Transactions of the Korea Information Processing Society, vol. 13, no. 12, pp. 710-718, 2024. DOI: https://doi.org/10.3745/TKIPS.2024.13.12.710.

[ACM Style]

Wonji Jo, Sung-wook Park, Yong-seok Kim, Se-hoon Jung, and Chun-bo Sim. 2024. Super-Resolution Model Based on Wavelet Domain Loss for Improving Wind Speed Forecasting Accuracy. The Transactions of the Korea Information Processing Society, 13, 12, (2024), 710-718. DOI: https://doi.org/10.3745/TKIPS.2024.13.12.710.

Forms

Search
(IN TITLE, AUTHOR, ABSTRACT,KEYWORDS)

Advanced Search ▼

POPULAR KEYWORDS
(TOP 10 KEYWORDS)

Recent Publications
(LAST 3 YEARS)

Old Journals

Indexing

All publications of TKIPS is indexed in DOI, EBSCO, Google Scholar, Crossref, and CrossCheck.

TKIPS is also selected as the Journal for Accreditation by NRF (National Research Foundation of Korea).

Related Journals