Automatic Monitoring and Detection of Hot Spot on Photovoltaic Panel Based on YOLOv5

The research contented the development of an automatic monitoring system for photovoltaic (PV) panel array with hot-spot detection capability through applying YOLOv5 deep learning model on PV thermal images. The system uses a sliding mechanism and automatic capture thermal images at each panel posit...

Full description

Saved in:

Bibliographic Details
Published in:	2023 International Conference on Advanced Computing and Analytics (ACOMPA) pp. 59 - 63
Main Authors:	Hoang Khang, Nguyen Phuoc, Duc, Doan Xuan, Nhan, Nguyen Chi, Van Tuan, Huynh
Format:	Conference Proceeding
Language:	English
Published:	IEEE 22-11-2023
Subjects:	Analytical models Automation Computational modeling Deep learning monitoring system photovoltaic panel Photovoltaic systems solar power system Training YOLO
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!

Description
Summary:	The research contented the development of an automatic monitoring system for photovoltaic (PV) panel array with hot-spot detection capability through applying YOLOv5 deep learning model on PV thermal images. The system uses a sliding mechanism and automatic capture thermal images at each panel position with a thermal camera and a controller system. The designed system afterward is used to collect an image dataset for training a deep learning model of YOLOv5 architecture to detect hot-spot area from the panel's thermal image. The training result of the YOLOv5 model from a dataset of 120 images with train/valid ratio of 90/30 gives out the accuracy of 81% of hot-spot detection. The trained model is integrated into the system operation to automatically perform daily monitoring for panels in the photovoltaic array.
DOI:	10.1109/ACOMPA61072.2023.00019