The UTC Graduate School is pleased to announce that Garrick Muncie will present Master’s research titled, SOLAR PANEL DAMAGE IDENTIFICATION USING TENSORFLOW LITE on 03/04/2024 at 1040 in ECS 426 Maytag room. Everyone is invited to attend.
Engineering
Chair: Abdul Ofoli
Co-Chair:
Abstract:
The number of utility-scale photovoltaic (PV) installations are on the rise. Which saw a 50% rise compared to the first eight months of 2022. (Lawrence Berkeley National Laboratory et al.) With larger scale installations quicker ways of identifying and locating damaged PV arrays is need. Our solution is to use drones to capture aerial photos and use TensorFlow-Lite and Keras Deep-Learning methods to determine if a Panel has issues. The prepossessed model shows high performance (Accuracy score of 92.89% and a F1 Score of 92.92%) with an execution time of 0.303 seconds per picture.