ENERGY / FORECASTING
Solar Power Forecasting.
Time-series and machine learning approaches to hourly photovoltaic power forecasting.
The challenge
Solar generation depends on changing irradiance, temperature, wind, and cloud conditions. Accurate forecasts support planning and energy-market decisions.
The approach
Explored persistence, statistical time-series models, machine learning, and physical features such as plane-of-array irradiance and cell temperature.
Key learning
Lagged irradiance features were influential, while cloudy winter days proved harder to predict. Full evaluation and methodology can be shared in a future technical write-up.