Image Analytic Inspections of Solar Panels

Small unmanned aerial systems (sUAS), or drones, are versatile tools whose applications are just beginning to be explored.  In the summer of 2018, I am working with an engineer in the Naval Facilities and Engineering Command (EXWC), Public Works Division, to explore ways to use computer-assisted methods to make UAS-based methods for inspecting solar panel arrays economical and efficient.

Our goal is to use image analysis and techniques from computer vision to identify malfunctions in solar panels and their photovoltaic cells while a sUAS is conducting an inspection using multispectral sensors.  The method we aim to develop will allow the Navy to inspect its growing renewable energy facilities, domestic and international, more economically and without putting its employees at risk.

To be successful, we need to know how solar panels appear when they are functioning correctly and when they are malfunctioning.  The Navy has loaned the University a small collection of solar panels and a ‘load bank’ that we will use to gather initial images of panels.  Our experimental station, which is located at CI Park, is pictured below.

The panels are cordoned off because, when functioning correctly, they present an electrocution danger.  Please don’t touch.

Fall 2018, I will be working with students in a course on image classification and shape description that will expend the summer work.  My goal is to have students develop methods that will build on the successes of the summer.  Check back here for updates to find out what they are able to do.