Unmanned Aerial Vehicle (UAV) Decision-Making for Photovoltaic (PV) Plant Diagnostics Using Image and Electrical Data Analysis

Abstract

This paper aims to develop an unmanned aerial vehicle (UAV) decision-making platform for accurate photovoltaic (PV) plant diagnosis and optimum operation and maintenance (O&M) activities. A modular approach was followed to support the full plant characterization and performance assessment of the PV system under study by combining image processing with electrical data analytics and advanced computing capabilities. The proposed UAV platform incorporates an electroluminescence camera, a Red-Green-Blue camera, and a thermographic sensor. The system is controlled by a user-friendly geo-visualization interface with advanced reporting and operational functionalities. Robot operating, processing and decision-making capabilities are also included, adding the possibility to perform aerial flight inspection and predefined actuations by the UAV platform. The platform can propose remedy field actions to minimize the response and resolution times (i.e., the timing since an anomaly is detected and certain actions are carried out to mitigate the incident). Therefore, it enables near real-time monitoring of the PV plant, early detection, accurate localization of faults and geolocation of defects, leading to time- and cost-efficient PV plant diagnosis. The proposed platform was validated in a real environment that manages the complete process of data and images acquisition, cleaning and storage of the received information, image processing and data analysis for early fault detection, generation of recommendations for maintenance and prioritization/scheduling of field O&M actions. The results showed the efficacy of the utilized equipment and incorporated functionalities for raw data management, near real-time fault detection and localization of faulty modules, and accurate geolocation of the defects within the PV system using UAV-captured images complemented by electrical data analysis.