Theromgraphic Analysis of Photovoltaics

The Intersection of Computer Vision and Green Energy

Abstract

By 2050, the world will need an estimated 20 terawatts of non-CO2 energy to curb the impending greenhouse gas production crisis. Photovoltaic (PV) solar technologies have the potential to deliver this desperately needed green energy. However, defects on PV panels lead to reduced production of harvested solar energy. Thus, inspection of solar panels is essential in efficient operation of solar panels. However, human analysis is subjective and time consuming made all the more difficult by remote panel locations and the large-scale of PV farms. Automatic methods for defect detection with long-wave infrared thermography (IR) have been proposed, but the existing solutions are costly. This six month project will investigate low-cost IR vision systems for defect detection of PV panels. Novel computer vision algorithms will be developed that will allow utilities to remotely assess the health of panels at quickly at a large-scale.

Personnel

  • PI: Alberto C. Cruz
  • Co-PI: Saeed Jafarzadeh
  • Kody Bryant
  • Alex Rinaldi
  • Gordon Griesel

Publications

  • Kody Bryan, "Cost Efficient Long-wave Infrared Thermography to Detect Faults in Photovoltaics," Southern California Conference on Undergraduate Research, 2015.
  • Alberto C. Cruz, Saeed Jafarzadeh, Alex Rinaldi and Saeed Jafarzadeh. "Deep Autoencoders for Super-resolution of IR Thermography." International Conference on Pattern Recognition. In preparation.