Applicant
Prof. Dr. Helena Liebelt
Institute Future Technologies
TH Deggendorf
Project Overview
Photovoltaic modules are exposed to mechanical, thermal, and environmental stress through their operational lifetime. Defects such as hotspots, soiling, or structural damage can reduce energy yield and affect operation reliability [1]. Modern approaches combine infrared thermography with automated image analysis to detect such conditions and support more efficient maintenance [2].
The project “Quanten-KI-basierte Klassifikation und Zustandsbewertung von Photovoltaikmodulen mittels Infrarotthermografie / Quantum AI Image Classification for Photovoltaic Diagnostics: A Use Case Introduction“ investigates the applicability of Quantum Machine Learning (QML) to this real-world classification task. The planned work is based on a publicly available dataset comprising around 20,000 infrared images of photovoltaic modules across twelve condition classes [3]. The project will explore hybrid quantum-classical approaches that combine classical image or feature processing with quantum-based classification. Attention will be given to Variational Quantum Circuits and Quantum Support Vector Machines, as well as to the challenge of efficient data representation under current NISQ constraints. The developed approaches will be systematically compared and evaluated with established classical machine-learning baselines.
References
[1] M. Köntges et al., “Review of failures of photovoltaic modules,” 2014.
[2] U. Jahn et al., Review on infrared and electroluminescence imaging for PV field applications: International Energy Agency Photovoltaic Power Systems Programme. International Energy Agency, 2018.
[3] M. Millendorf, E. Obropta, and N. Vadhavkar, “Infrared Solar Modules,” 2023, Kaggle. doi: 10.34740/KAGGLE/DSV/6866566.