Solarwise Hybrid Solar Forecasting Model

Objective: Improve the accuracy of short-term solar energy (PV) forecasts by predicting how different cloud conditions affect output — helping solar operators better manage grid integration..

Methodology:

• Hybrid CNN/ResNet Model: Trained a hybrid neural network that combined image analysis (CNN) with time-sensitive data modeling (ResNet) to predict solar output based on sky condition images.

• Condition-Specific Forecasting: Built separate sub-models to handle four distinct weather types — clear, partly cloudy, mostly cloudy, and overcast. varying sky conditions. 

• Data & Training: Used 600 labeled sky images from Stanford’s dataset. The model was trained to make 10-minute-ahead PV output forecasts.

Key Outcomes:

• Achieved Root Mean Squared Error (RMSE) values as follows: Clear (1.91), Partly-cloudy (1.72), Mostly-cloudy (1.13), and Overcast (4.48). The model excelled with mostly-cloudy conditions but faced challenges with overcast scenarios, likely due to the unpredictability of dense cloud cover.

• Contribution & Future Scope: This hybrid approach could help solar operators better plan energy usage under changing weather conditions. With added meteorological inputs (like wind speed or humidity), the model could evolve into a real-time forecasting tool with broader applications in smart grid systems.