Full Text:
<7>
CLC number:
On-line Access: 2026-03-09
Received: 2025-11-10
Revision Accepted: 2026-02-09
Crosschecked: 0000-00-00
Cited: 0
Clicked: 24
Physics-informed deep learning for data-efficient and robust PV power forecasting
| |||||||||||||
Chang HUANG1, Xuanbin HUANG1, Jinmin GUO2, Zhuo CAO1, Ting HE1,Wentao SHANG1. Physics-informed deep learning for data-efficient and robust PV power forecasting[J]. Journal of Zhejiang University Science A, 1998, -1(-1): .
@article{title="Physics-informed deep learning for data-efficient and robust PV power forecasting",
author="Chang HUANG1, Xuanbin HUANG1, Jinmin GUO2, Zhuo CAO1, Ting HE1,Wentao SHANG1",
journal="Journal of Zhejiang University Science A",
volume="-1",
number="-1",
pages="",
year="1998",
publisher="Zhejiang University Press & Springer",
doi="10.1631/jzus.A2500582"
}
%0 Journal Article
%T Physics-informed deep learning for data-efficient and robust PV power forecasting
%A Chang HUANG1
%A Xuanbin HUANG1
%A Jinmin GUO2
%A Zhuo CAO1
%A Ting HE1
%A Wentao SHANG1
%J Journal of Zhejiang University SCIENCE A
%V -1
%N -1
%P
%@ 1673-565X
%D 1998
%I Zhejiang University Press & Springer
%DOI 10.1631/jzus.A2500582
TY - JOUR
T1 - Physics-informed deep learning for data-efficient and robust PV power forecasting
A1 - Chang HUANG1
A1 - Xuanbin HUANG1
A1 - Jinmin GUO2
A1 - Zhuo CAO1
A1 - Ting HE1
A1 - Wentao SHANG1
J0 - Journal of Zhejiang University Science A
VL - -1
IS - -1
SP -
EP -
%@ 1673-565X
Y1 - 1998
PB - Zhejiang University Press & Springer
ER -
DOI - 10.1631/jzus.A2500582
Abstract: Photovoltaic (PV) power forecasting is challenged by its inherent variability. Pure data-driven models struggle with generalization under data scarcity and complex weather conditions. In this paper, we introduce a physics-Informed Deep Learning hybrid model (PIDL-HM) that systematically generates physically-grounded input features (e.g., plane-of-array irradiance and module temperature) for a Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM), establishing a principled integration framework beyond simple ensemble methods. Rigorously validated across multiple PV power plants in China and Australia for 15-min, 4-h, and 24-h forecasting, our approach demonstrates superior performance of up to 8.93% in root mean square error (RMSE) compared to a purely data-driven baseline. Crucially, the model shows remarkable data Efficiency, maintaining high accuracy with only three months of training data, and exceptional robustness, providing a 6.87% improvement in performance under strong cross-seasonal data distribution shifts. This work has provided a reliable and data-efficient forecasting solution, establishing the PIDL-HM as a foundational element for next-generation forecasting systems.
Open peer comments: Debate/Discuss/Question/Opinion
<1>