Paper Title
A COMPARATIVE STUDY OF ANOMALY DETECTION ALGORITHMS FOR SOLAR PHOTOVOLTAIC DEFECTS

Abstract
In the advancing realm of photovoltaic (PV) solar panel inspection, incorporating machine learning (ML) models signifies a revolutionary approach to detecting and categorising defects. This paper presents a competitive study between four advanced anomaly detection models for solar photovoltaic defect detection, namely: Artificial Neural Network (ANN), Convolutional Neural Network (CNN), Deep Neural Network (DNN) and a composite model combining CNN and DNN model, to address the problem of anomaly detection for solar photovoltaic defects. Then, the performance of all models was evaluated based on accuracy, precision, recall values, and F1 score. The comparative study on solar panel defect detection reveals that the CNN and composite models exhibit superior overall performance compared to ANN and DNN in terms of accuracy, F1 score, and recall. Additionally, the results indicate that the CNN and composite models outperform other implemented models in defect detection across various defect types. However, it is noted that all models struggle to detect, to some extent, defect type D2 (Crack). This suggests the need for further refinement or specific adjustments to enhance detection accuracy for these particular defect. Keywords - Photovoltaic (PV) Anomaly Detection, Machine Learning