Smart Monitoring of Solar Photovoltaic Panels by the Approach of Machine Learning

Abstract

Due to the increasingly evident climate change and the recent energy crisis triggered by regional conflicts, the exploitation of renewable energy has become one of the pressing tasks today. In light of this background, today the photovoltaic (PV) market is booming globally. This makes the management and maintenance of solar PV panels a new area of business, as neglecting it may not only lead to significant financial losses but also the failure to combat climate change and the energy crisis. In fact, in the long service life of solar panels, they do face many risks that may degrade their power generation performance, damage their structures, or even cause the complete loss of their power generation capacity. It is hoped that these problems can be found and resolved as soon as possible. However, in reality, this is a challenging task as a large solar power plant may contain hundreds of thousands of solar PV panels. To address this issue, a smart solar panel condition monitoring technique is studied in this paper with the aid of the U-Net neural network and Support vector machine (SVM). The research result has shown that using the developed technique, the thermal infrared images that are remotely collected by drones from solar power plants can be rapidly and automatically processed, analysed and classified with reasonable accuracy. This is of great significance to further improve the asset management skills in solar power plants.