Digital Twin‐Driven Closed Loop for Photovoltaic Power Plant Cleaning Decision‐Making: Considering Spatial Heterogeneity and Self‐Learning Feedback
Digital Twins and Applications · 2026-01 · 3 (1)
DOI: 10.1049/dgt2.70043 ↗Kurz gefasst
Fachartikel in der Zeitschrift Digital Twins and Applications, online erschienen am 07.07.2026. Verfasst von Tianwen Zhang und 3 weiteren Autor:innen. Die Arbeit nennt 30 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).
Abstract (Originalsprache)
Soiling substantially erodes long‐term PV plant efficiency, making well‐timed cleaning economically critical. Existing methods are mostly open‐loop and hence lacking feedback and adaptation. This work proposes a closed‐loop digital twin framework linking physical and virtual plants via three components. The physics‐constrained Bayesian fusion of multi‐source irradiance enables soiling ratio estimation and counterfactual prediction under variable data quality. In addition, a heterogeneity index from string‐level efficiency differences switches diagnostic granularity across plant, zone, and equipment levels, relaxing uniform soiling assumptions. Also, an event‐driven mechanism uses counterfactual simulation to quantify gains, correct parameters and thresholds, and builds a cold‐start case library enabling continuous evolution. Controlled tests on 11 commercial plants yield a soiling ratio MAE of 2.37% (ordinary dust) and 2.47% (special soiling). A year‐long study at an industrial‐commercial plant in Gansu, China, shows the closed‐loop strategy yields roughly 13 times the net cleaning revenue of the open‐loop baseline. These results indicate improved economic return and self‐learning, advancing PV maintenance from open‐loop planning to self‐evolving operation.
Zitieren
APA
Zhang, T., Wang, R., Chen, L., & Yang, Q. (2026). Digital Twin‐Driven Closed Loop for Photovoltaic Power Plant Cleaning Decision‐Making: Considering Spatial Heterogeneity and Self‐Learning Feedback. Digital Twins and Applications, 3(1). https://doi.org/10.1049/dgt2.70043
Vancouver
Zhang T, Wang R, Chen L, Yang Q. Digital Twin‐Driven Closed Loop for Photovoltaic Power Plant Cleaning Decision‐Making: Considering Spatial Heterogeneity and Self‐Learning Feedback. Digital Twins and Applications. 2026;3(1). doi:10.1049/dgt2.70043
BibTeX
@article{Zhang2026,
author = {Zhang, Tianwen and Wang, Ruohong and Chen, Lu and Yang, Qiang},
title = {Digital Twin‐Driven Closed Loop for Photovoltaic Power Plant Cleaning Decision‐Making: Considering Spatial Heterogeneity and Self‐Learning Feedback},
journal = {Digital Twins and Applications},
year = {2026},
volume = {3},
number = {1},
doi = {10.1049/dgt2.70043}
}
EUVIMED