Autonomous UAV-Based Inspection Methods for Photovoltaic Arrays Using AI-Driven Defect Detection
Disciplines
Aeronautical Vehicles | Electromagnetics and Photonics | Other Electrical and Computer Engineering | Other Mechanical Engineering | Power and Energy | Systems Engineering and Multidisciplinary Design Optimization
Abstract (300 words maximum)
Sustainable energy sources such as PV (Photovoltaic) arrays have expanded significantly over the past century, and now require a new generation of high-fidelity, cost-effective, and mobile inspection methods. UAVs (Unmanned Aerial Vehicles) provide a unique capability to quickly and effectively improve operational linearity in large-scale PV installations. UAVs can access and inspect areas that are difficult to access, and can be equipped with radiometric thermal imaging systems, enabling rapid and accurate assessment of industrial scale industrial sites.
In this project, we leveraged radiometric thermal imaging to acquire precise temperature data and apply physical analysis along with computational methods to properly detect and diagnose potential structural or electrical faults in PV arrays. This thermal data allows us to reliably identify defects such as hot spots, microcracks, and conductor failures with considerable accuracy and consistency. The reliability of this data allowed us to properly train AI (Artificial Intelligence) and ML (Machine Learning) programs to automatically recognize and classify these defects. Specifically, we have employed YOLOv12 for automated pattern recognition, identifying defect structures consistent with those included in the training models.
By eliminating the need for manual inspections and extensive training periods, our integration of UAV technology in conjunction with AI-driven analysis methods will allow power operators to scale PV installations more efficiently while minimizing energy losses due to undetected faults. Through the methods used in this project, the research team is introducing a promising framework for improving the reliability and sustainability of large-scale PV installations.
Use of AI Disclaimer
no
Academic department under which the project should be listed
SPCEET – Mechanical Engineering
Primary Investigator (PI) Name
Adeel Khalid
Autonomous UAV-Based Inspection Methods for Photovoltaic Arrays Using AI-Driven Defect Detection
Sustainable energy sources such as PV (Photovoltaic) arrays have expanded significantly over the past century, and now require a new generation of high-fidelity, cost-effective, and mobile inspection methods. UAVs (Unmanned Aerial Vehicles) provide a unique capability to quickly and effectively improve operational linearity in large-scale PV installations. UAVs can access and inspect areas that are difficult to access, and can be equipped with radiometric thermal imaging systems, enabling rapid and accurate assessment of industrial scale industrial sites.
In this project, we leveraged radiometric thermal imaging to acquire precise temperature data and apply physical analysis along with computational methods to properly detect and diagnose potential structural or electrical faults in PV arrays. This thermal data allows us to reliably identify defects such as hot spots, microcracks, and conductor failures with considerable accuracy and consistency. The reliability of this data allowed us to properly train AI (Artificial Intelligence) and ML (Machine Learning) programs to automatically recognize and classify these defects. Specifically, we have employed YOLOv12 for automated pattern recognition, identifying defect structures consistent with those included in the training models.
By eliminating the need for manual inspections and extensive training periods, our integration of UAV technology in conjunction with AI-driven analysis methods will allow power operators to scale PV installations more efficiently while minimizing energy losses due to undetected faults. Through the methods used in this project, the research team is introducing a promising framework for improving the reliability and sustainability of large-scale PV installations.