SMU Data Science Review


This paper details how to predict solar radiation at a location for the next few hours using machine learning techniques like Facebook’s Prophet, and Amazon’s DeepAR+. Multiple techniques like AutoRegressive (ARIMA) and Exponential Smoothing (ES) have been used to forecast solar radiation, but they lack accuracy and are not scalable. Whereas Prophet, and Amazon’s DeepAR+ are scalable, accurate, and easily integrated into other machine learning techniques. This will be the first time where the combination of these techniques along with Linear Regression, Random Forest, XGBoost and Decision Tree will be leveraged to forecast solar radiation for the short term. Predicting solar energy accurately depends on multiple factors (including weather conditions) that make forecasting highly resource-intensive, and accuracy remains a challenge. Improving the accuracy of the short-term forecast of solar energy production would provide a massive value to the companies operating IoT Devices and drones to have a more efficient operation and reduced cost. The objective is to improve the accuracy of forecasting short-term solar radiation to power drones and IoT devices, leveraging the ensemble techniques by combing the outcome of Prophet and DeepAR+. Facebook’s Prophet, and Amazon’s DeepAR+ used to carry out shortterm solar forecasting can be scaled by leveraging the supercomputer. Amazon’s DeepAR+ runs on the AWS cloud platform, so they align well with scaling and bring in all the enhancement that comes with cloud technology. Multiple models were used to identify the best way to forecast short-term solar radiation. Random Forest and ensemble models outperformed the Facebook Prophet and Amazon’s DeepAR+, achieving a coefficient of determination R2 of 99 % in Dallas, Texas. Ensemble Model was created to minimize the bias and variance of the outcome.

Included in

Data Science Commons