Generalized, Deployable, Insolation Forecasts

Clean Power Research Computer Science, 2022–23

Liaison(s): Thomas Haley POM ’02, Alex Kubiniec
Advisor(s): Julie Medero
Students(s): Kendah Abughararh, Max Hui (PM-S), Malia Morgan (PM-F), Hallie Seay

Solar utilities rely on accurate sunlight forecasts to determine how much power they will produce on a given day. Clean Power Research provides these forecasts, but existing methods take too much time and computation to forecast for a given location immediately. To explore machine learning solutions, our Clinic team constructed a pipeline to retrieve and transform data, train predictive models and evaluate their performance. We used this to investigate new data sources and the viability of two different machine learning approaches.