-- This project is inactive -- 

Title: Watt-Sun: A Multi-Scale, Multi-Model, Machine-Learning Solar Forecasting Technology
Funding Opportunity: Solar Forecasting 
SunShot Subprogram: Systems Integration

As part of this project, new solar forecasting technology will be developed that leverages big data processing, deep machine learning, and cloud modeling integrated in a universal platform with an open architecture. Similar to the Watson computer system, this proposed technology will leverage deep machine learning and self-adjusting voting algorithms to decide between various forecasting models and expert systems. This approach will yield the best forecasts and more importantly, continuously improve and adjust as the system is operating and evolving.

APPROACH

Similar to the recently demonstrated Watson computer system, the proposed technology will leverage deep machine learning and self-adjusting voting algorithms to decide between various forecasting models and expert systems. This approach will yield the best forecasts and more importantly, continuously improve and adjust as the system is operating and evolving. The solar forecasting framework is independent of proprietary weather or solar radiation models and enables the technology to scale and to be adopted by solar producers, electrical utilities, independent system operators (ISOs), and other stakeholders. The forecasting will also be validated at multiple sites with significantly different weather patterns. The team will work closely with utilities, solar power producers, and ISOs to integrate the proposed technology and to determine the value of solar forecasting on daily operation, load modeling, optimizing spinning reserve, and day ahead planning.

INNOVATIONS

The goal of the project is the development and demonstration of an improved solar forecasting technology (short: Watt-sun), which leverages new data processing technologies and optimal blending between different models and expert systems using deep machine learning methods. The technology promises significant advances in accuracy of solar forecasting as measured by existing or new metrics which themselves will also be developed within the scope of this project. This solar forecasting framework is independent of proprietary weather or solar radiation models and enables the technology to scale and to be adopted by solar producers, electrical utilities, independent system operators, and other stakeholders. The developed technology will be integrated into the operation of at least one Independent System Operator (ISO) and one utility.

OUTCOMES