تصوير كهروضوئي عالي الدقة معزز بالذكاء الاصطناعي للكشف المبكر عن عيوب التلامس داخل الخلايا الشمسية

Abstract

 This study aims to develop an intelligent system based on artificial intelligence for the early detection of defects in solar cells using high resolution photovoltaic imaging techniques. An integrated methodology was designed that combines digital image processing with deep learning algorithms particularly Convolutional Neural Networks (CNNs)to analyze images and extract patterns indicative of internal and external cell defects. The proposed system relied on a diverse database containing images of both healthy and defective cells. The model was trained on this dataset after applying enhancement and normalization procedures to improve prediction accuracy.

Experimental results showed that the model achieved a detection accuracy ranging between 96.4% and 97.8%, with a sensitivity of 96.5% and a precision of 98.2%, confirming its efficiency in distinguishing between different types of defects such as micro-cracks, interruptions, and non-uniform reflection areas. The results indicate that the proposed system can be deployed as an intelligent platform for industrial-scale solar cell inspection. Moreover, the ROC curve analysis demonstrated high performance stability across test samples not used during training.

The heatmaps generated by the model provide an accurate visual tool for locating defects within the cells, helping accelerate inspection processes and reduce human errors by more than 80% compared to traditional methods such as infrared imaging. The system also shows strong potential for integration into industrial production lines to enable real-time and non-destructive solar cell inspection.

This study highlights the significant potential of artificial intelligence to enhance the reliability and quality of visual inspection in photovoltaic industries and represents a step toward developing autonomous monitoring systems capable of supporting smart transformation in the field of sustainable solar energy.