Home Publications Research Data Research Grants Equipment Inventions & Patents Awards Additional Research Activities Faculties & Institutions Research Areas

Benchmarking imaging-based inspection techniques for PV systems

Pillai DS, Thiruchuthan A, Mellit A, Bayindir AB, Buerhop-Lutz C, Marius Peters I, Kivambe MM, Abdallah AA, Bayhan S, Aissa B (2027)

---

Publication Type: Journal article, Review article

Publication year: 2027

Journal

Book Volume: 244

Article Number: 117457

DOI: 10.1016/j.rser.2026.117457

Abstract

Photovoltaic modules experience gradual degradation and sudden failures that reduce energy yield, reliability, and safety, motivating the use of imaging-based diagnostic techniques. Imaging approaches enable the detection of electrical and physical defects that are often invisible through conventional visual inspection. Studies have reported that microcracks, hotspots, potential-induced degradation, and light-induced degradation can noticeably affect PV module reliability. This review therefore examines both established field-deployed diagnostic techniques (technology readiness level > 5) and emerging approaches that are under development (technology readiness level <5). Most existing studies focus on the development or evaluation of individual imaging techniques, while some review articles discuss multiple methods without providing detailed comparisons between conventional and emerging diagnostic approaches across different fault types. As a result, a systematic comparison of the diagnostic capabilities of the available imaging modalities remains limited. To address this gap, this review presents a structured study and fault-centric benchmarking of various imaging-based PV inspection techniques, emphasizing fault visibility and diagnostic relevance across imaging modalities rather than relying solely on reported accuracy metrics. Additionally, a hybrid scope–mapping systematic review methodology is applied, in which peer-reviewed studies are screened, classified, and synthesized based on fault type and technological maturity. Based on results reported in the literature, machine learning-assisted infrared thermography has achieved detection accuracies of 94–98%, while deep learning-based electroluminescence methods have reported accuracies of up to 97.8%. Ultraviolet fluorescence techniques have demonstrated crack detection rates exceeding 91% and inspection throughput up to 10–15 times higher than near-infrared inspection under comparable operating conditions. These performance values originate from different studies, datasets, and experimental conditions and are therefore intended to illustrate representative capabilities rather than enable direct comparison between techniques. Emerging approaches such as daylight luminescence and magnetic-field-based diagnostics are also gaining attention, although their broader use remains limited by operational complexity and signal-to-noise challenges.

Involved external institutions

How to cite

APA:

Pillai, D.S., Thiruchuthan, A., Mellit, A., Bayindir, A.B., Buerhop-Lutz, C., Marius Peters, I.,... Aissa, B. (2027). Benchmarking imaging-based inspection techniques for PV systems. Renewable and Sustainable Energy Reviews, 244. https://doi.org/10.1016/j.rser.2026.117457

MLA:

Pillai, Dhanup S., et al. "Benchmarking imaging-based inspection techniques for PV systems." Renewable and Sustainable Energy Reviews 244 (2027).

BibTeX: Download