EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Artificial neural network-based modelling of photovoltaic/thermal systems: a review

Shamsa Said Al Jahwari, Tengku Juhana Bt Tengku Hashim, Hussein A Kazem, Karmila Binti Kamil

Open Engineering · 23.01.2026 · 16 (1)

DOI: 10.1515/eng-2025-0190 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Open Engineering, online erschienen am 24.08.2026. Verfasst von Shamsa Said Al Jahwari und 3 weiteren Autor:innen. Die Arbeit nennt 71 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Integration of the hybrid Photovoltaic/Thermal (PVT) system to existing energy and power grid is increasing exponentially, because of its superior energy production, combining both electric power and thermal energy utilization. However, despite accurate modeling of PVT’s complex performance, its non-linear behavior across many different operating conditions still poses significant challenge in the case of using conventional analytical and experimental methods, which are mostly involve complex computations hence consuming valuable time and become limited in scalability for more complex datasets. This review is the first to provide a focused, quantitative comparison of ANN architectures specifically for PVT systems, synthesizing reported R 2 , MSE, and RMSE values across more than 50 studies and configurations (air-, water-, and nanofluid-based). It further identifies underexplored areas such as deep and temporal learning for PVT fault detection and proposes a structured research agenda on physics-informed and metaheuristic-optimized ANN models tailored to PVT applications. The evaluation and review in this paper synthesizes research findings from numerous studies and reports, showcasing that ANN-based models, particularly MLP, SVM and hybrid metaheuristic-optimized modeling architectures like Particle Swarm Optimization (PSO), achieve significant prediction or forecasting accuracy with R 2 values typically exceeding 0.9500 for both electricity and thermal energy output, while its mean squared error (MSE) values are recorded at 0.0007. Further, review highlights a critical significance of utilizing high-quality and controlled experimental dataset used to do robust training plus validation of the predictive models. Therefore, the comparative analysis reveals the superiority of ANN architectures over other AI techniques, while also identifying persistent research gaps, including the under exploitation of deep learning and temporal models for fault detection. Finally, this study outlines possible directions for future research, including hybrid physics-informed AI modelling, integration with IoT and big data, and the development of standardized benchmarks that may support progress toward the next generation of intelligent PVT structures.

Zitieren

APA

Jahwari, S. S. A., Hashim, T. J. B. T., Kazem, H. A., & Kamil, K. B. (2026). Artificial neural network-based modelling of photovoltaic/thermal systems: a review. Open Engineering, 16(1). https://doi.org/10.1515/eng-2025-0190

Vancouver

Jahwari SSA, Hashim TJBT, Kazem HA, Kamil KB. Artificial neural network-based modelling of photovoltaic/thermal systems: a review. Open Engineering. 2026;16(1). doi:10.1515/eng-2025-0190

BibTeX
@article{Jahwari2026,
  author = {Jahwari, Shamsa Said Al and Hashim, Tengku Juhana Bt Tengku and Kazem, Hussein A and Kamil, Karmila Binti},
  title = {Artificial neural network-based modelling of photovoltaic/thermal systems: a review},
  journal = {Open Engineering},
  year = {2026},
  volume = {16},
  number = {1},
  doi = {10.1515/eng-2025-0190}
}
Kostenlos starten →