A Novel Lightweight Transformer for Photovoltaic Power Forecasting with Gap-Aware Reconstruction and Residual-Based Diagnostics
Journal of Intelligent Decision Making and Information Science · 03.10.2026 · 3 (16s) · 2431-2450
DOI: 10.17762/jidmis.v3.5179 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Intelligent Decision Making and Information Science, online erschienen am 03.10.2026. Verfasst von C. Swetha Priya. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
The reliability of the electric grid and the uncertainties of the network depend on the accurate predictions of solar active power and the residual diagnostic reasoning. Further, heterogeneity in photovoltaic (PV) configurations, data gaps, and task-specific AI models limit generalizability. To address these limitations, this study proposes a novel lightweight Interpretable Solar Transformer (IST) framework. To impute missing PV time-series data, we developed a lightweight Transformer framework that performs multistage, gap-aware data reconstruction. Utilizing a single encoder block, the proposed framework is subsequently trained on the reconstructed dataset to predict active power output. Furthermore, we performed fault diagnosis and anomaly detection using a residual-based reasoning module. Additionally, we evaluated the generalizability of the proposed framework across four benchmark PV technology datasets. The proposed framework demonstrates strong performance with RMSE values as low as 0.2795 and R2 scores up to 0.9697. It is also observed that IST has fewer trainable parameters (89,397) and a smaller model size (1.088 MB) than the existing grid transformers. Finally, we evaluated the robustness of the proposed framework and observed that the anomaly rate remained below 3%. Experimental results demonstrate that our framework significantly improves prediction accuracy, generalization, and robustness compared to existing schemes.
Zitieren
APA
Priya, C. S. (2026). A Novel Lightweight Transformer for Photovoltaic Power Forecasting with Gap-Aware Reconstruction and Residual-Based Diagnostics. Journal of Intelligent Decision Making and Information Science, 3(16s), 2431-2450. https://doi.org/10.17762/jidmis.v3.5179
Vancouver
Priya CS. A Novel Lightweight Transformer for Photovoltaic Power Forecasting with Gap-Aware Reconstruction and Residual-Based Diagnostics. Journal of Intelligent Decision Making and Information Science. 2026;3(16s):2431-2450. doi:10.17762/jidmis.v3.5179
BibTeX
@article{Priya2026,
author = {Priya, C. Swetha},
title = {A Novel Lightweight Transformer for Photovoltaic Power Forecasting with Gap-Aware Reconstruction and Residual-Based Diagnostics},
journal = {Journal of Intelligent Decision Making and Information Science},
year = {2026},
volume = {3},
number = {16s},
pages = {2431--2450},
doi = {10.17762/jidmis.v3.5179}
}
EUVIMED