Environmental Sensor-Less Hybrid Analytical-Machine Learning (ESHAML) framework for ultra-fast solar irradiance estimation in climate-sensitive real-time applications: experimental validation

Basit Öğe Kaydı
dc.authorid0000-0002-4049-0716
dc.authorid0000-0001-6944-4775
dc.contributor.authorHarrison, Ambe
dc.contributor.authorMbasso, Wulfran Fendzi
dc.contributor.authorDagal, Idriss
dc.contributor.authorAlombah, Njimboh Henry
dc.contributor.authorJangir, Pradeep
dc.contributor.authorAl-Gahtani, Saad F.
dc.contributor.authorElbarbary, Z. M. S.
dc.date.accessioned2026-01-31T15:08:20Z
dc.date.available2026-01-31T15:08:20Z
dc.date.issued2026
dc.departmentİstanbul Beykent Üniversitesi
dc.description.abstractAccurate real-time solar irradiance (SI) estimation is pivotal for optimizing climate-sensitive system, yet existing methods often rely on costly environmental sensors or complex measurements, limiting their applicability in resource-constrained regions. This study introduces the Environmental Sensor-less Hybrid Analytical-Machine Learning (ESHAML) Framework, a novel solution for ultra-fast SI estimation tailored for climate-sensitive applications, such Renewable Energy Systems. ESHAML synergizes an analytical Photovoltaic (PV) model, derived from electrical characteristics, with a Machine Learning (ML) based- Bayesian regularization-trained neural network for temperature approximation, eliminating environmental sensor dependencies. A fractional-area Maximum Power Point Tracking (MPPT) algorithm achieves rapid convergence (under 20 ms) and stable PV operation, outperforming metaheuristic methods like Particle Swarm Optimization (PSO), Grey Wolf Optimization (GWO), Improved Differential Evolution (IDE), and Salp Swarm Optimization (SSO) in speed and stability. The framework leverages a novel synthetic data generation model with adaptive resolution for low-irradiance regimes, enhancing ML robustness in data-scarce environments. Validated through simulations and experimental studies using the IoT-based Trio-PV Monitor in Cameroon's solar-rich Dschang region, ESHAML achieves a Mean Absolute Percentage Error (MAPE) of 2.67-4.94 %, surpassing sensor-dependent approaches (e.g., Carrasco et al.'s analytical methods) and measurement-intensive neural networks (e.g., Chouay & Quassaid's dual-variable models). Comparative analyses highlight ESHAML's superior attributes: zero environmental sensors, seamless integration into existing PV infrastructures, and millisecond-level irradiance estimation. By addressing sensor reliance, data scarcity, and integration complexity, ESHAML emerges as a transformative tool for advancing solar energy utilization in Sub-Saharan Africa and beyond, aligning with global sustainable energy objectives.
dc.description.sponsorshipDeanship of Research and Graduate Studies, King Khalid University, through the Small Research Project [RGP2/108/46]
dc.description.sponsorshipThis work was supported by the Deanship of Research and Graduate Studies, King Khalid University, through the Small Research Project under Grant RGP2/108/46.
dc.identifier.doi10.1016/j.measurement.2025.118635
dc.identifier.issn0263-2241
dc.identifier.issn1873-412X
dc.identifier.scopus2-s2.0-105012616156
dc.identifier.scopusqualityQ1
dc.identifier.urihttps://doi.org./10.1016/j.measurement.2025.118635
dc.identifier.urihttps://hdl.handle.net/20.500.12662/10660
dc.identifier.volume257
dc.identifier.wosWOS:001581163800005
dc.identifier.wosqualityQ1
dc.indekslendigikaynakWeb of Science
dc.indekslendigikaynakScopus
dc.language.isoen
dc.publisherElsevier Sci Ltd
dc.relation.ispartofMeasurement
dc.relation.publicationcategoryMakale - Uluslararası Hakemli Dergi - Kurum Öğretim Elemanı
dc.rightsinfo:eu-repo/semantics/closedAccess
dc.snmzKA_WoS_20260128
dc.subjectSolar irradiance (SI)
dc.subjectEnvironmental Sensor-less Hybrid Analytical-Machine Learning (ESHAML)
dc.subjectMachine Learning (ML)
dc.subjectphotovoltaic (PV)
dc.subjectTrio-PV monitor
dc.subjectArtificial intelligence
dc.subjectFractional-area Maximum Power Point
dc.subjectTracking (MPPT)
dc.titleEnvironmental Sensor-Less Hybrid Analytical-Machine Learning (ESHAML) framework for ultra-fast solar irradiance estimation in climate-sensitive real-time applications: experimental validation
dc.typeArticle

Dosyalar

Koleksiyon

WoS İndeksli Yayınlar Koleksiyonu
Scopus İndeksli Yayınlar Koleksiyonu