EUVIMEDOpdag EUVIMED
Open access · CC BY 4.0

Multivariate PV Power Forecasting via CPO-Optimized Deep Spatiotemporal Networks

EnHui Qu, LiLi Wang, RuiXia Xie

Journal of Computer Science and Frontier Technologies · 2026-03-09 · 2 (3) · 71

DOI: 10.63313/jcsft.9049 ↗

At a glance

Journal article in Journal of Computer Science and Frontier Technologies, published online on 2026-03-09. Written by EnHui Qu and 2 co-authors. 21 references. The full text is freely available (open access, CC BY 4.0).

AbstractBibliografiske oplysningerCiteRelated studies

Abstract (original language)

Accurate photovoltaic (PV) power forecasting is critical for grid stability, yet it remains challenging due to the inherent volatility of meteorological conditions. Existing hybrid models often struggle with manual parameter sensitivity, inadequate local feature extraction, and computationally expensive cross-channel interactions. To address these gaps, this paper proposes a novel multivariate forecasting framework: CPO-VMD-TCN-SOFTS. The Crested Porcupine Optimizer (CPO) is employed to adaptively optimize Variational Mode Decomposition (VMD), effectively mitigating mode mixing. A Temporal Convolutional Network (TCN) extracts deep temporal dependencies, while the SOFTS architecture utilizes a centralized STAR mechanism to achieve cross-channel feature fusion with strict linear complexity. Comprehensive multi-resolution, all-season, and cross-site experiments demonstrate the model's superiority. The proposed architecture significantly enhances predictive accuracy and exhibits robust generalization, maintaining an average R2 above 0.9860 across diverse sites, thereby providing reliable support for smart grid dispatching.

Cite

APA

Qu, E., Wang, L., & Xie, R. (2026). Multivariate PV Power Forecasting via CPO-Optimized Deep Spatiotemporal Networks. Journal of Computer Science and Frontier Technologies, 2(3), 71. https://doi.org/10.63313/jcsft.9049

Vancouver

Qu E, Wang L, Xie R. Multivariate PV Power Forecasting via CPO-Optimized Deep Spatiotemporal Networks. Journal of Computer Science and Frontier Technologies. 2026;2(3):71. doi:10.63313/jcsft.9049

BibTeX
@article{Qu2026,
  author = {Qu, EnHui and Wang, LiLi and Xie, RuiXia},
  title = {Multivariate PV Power Forecasting via CPO-Optimized Deep Spatiotemporal Networks},
  journal = {Journal of Computer Science and Frontier Technologies},
  year = {2026},
  volume = {2},
  number = {3},
  pages = {71},
  doi = {10.63313/jcsft.9049}
}
Start for free →