A academic research project under Dr. Hari Om Bansal, Associate Professor, BITS Pilani. This project proposes a novel sets of inputs to Deep Neural Networks, that comprehensively describe the ambient conditons of a location, and ways to handle and train on such inputs.

Dataset: Collected through the Photovoltaic simulation dataset from System Advisor Model (SAM), an application developed by National Renewable Energy Laboratory, USA
(https://sam.nrel.gov/)

This dataset consists of half-hourly weather files for 49 different cities in USA.

Results:

Actual optimized power produced by a 2mx2m generic PV cell:

Results from the model:

Power prediction error: 0.47%