Solar panel flaws detected with 90% accuracy
The sun's power is a remarkable source of clean, renewable energy. Solar energy minimizes energy consumption, which benefits our planet. While most of us have seen solar panels on homes and buildings, many might not be aware that there are primarily two types of technology that use sunlight to produce energy.
First, solar thermal technology captures the heat of the sun for cooling and heating your home, among other purposes.
Photovoltaic (PV) technology, on the other hand, directly converts sunlight into electricity through panels made of semiconductor cells. They are designed to power up electrical devices and produce electricity for power grids.
Maintaining photovoltaic power plants is a huge job and that’s where CGI comes in.
The solution
Drones with thermal and RGB (Red Green Blue) cameras can inspect the photovoltaic plants, generating hundreds of images. After being processed, these images can be used to detect, with 90% accuracy, defects in the panels, as well as evaluate the plant's performance and enable the geopositioning of its location.
Using cloud architecture in Oracle and deep learning (Convolutional Neural Networks) with NVIDIA GPUS running in OCI, it is possible to analyze a huge data set of images about photovoltaic defects in a short amount of time. Moreover, a dashboard reports information about defects in panels, performance, and geopositioning with the objective to optimize the efficiency of the renewable energy through artificial intelligence (AI).
The result
The adoption of this solution helps our clients gain insights they can act on to improve their plant performance. By efficiently processing large volumes of images, the solution reduces the manual processing time from weeks to mere minutes, enabling increased frequency of inspections, reduced cost and greater reliability and speed in the photo analysis. Inspection history evolution helps clients to detect improvements and performance of the photovoltaic panels.
The images adhere to industry standards, and our technology ensures regulatory compliance, while remaining innovative in its design and approach. This solution has been distinguished by Oracle with awards at the Oracle CloudWorld 2023 event. First, in the 'Innovation' category of the 2023 Oracle Partner Award Cloud/Tech for Southern Europe and EMEA (Europe, Middle East and Africa), and then for the 2023 Oracle Worldwide Partner Award for Innovation.
