Renewable Generation Forecasting

AI forecasts renewable power output so operators can plan dispatch and commitments ahead of time.

Business impact

• Forecast accuracy — One deployment reported a 31% improvement in solar prediction accuracy.
• Operating-cost variation — One study reported operating-cost variation of up to 10% under forecast error.
• Value of wind output — One deployment reportedly increased the value of wind output by roughly 20%.

Data requirements

• Weather data (tabular) — Used as an input to predict renewable output.
• Historical renewable generation data (tabular) — Used to train and evaluate forecasting models.
• Distributed photovoltaic site data (tabular) — Supports short-horizon solar forecasting across sites.
• Wind-farm data (tabular) — Supports wind power forecasting and trust assessment.

AI methods and techniques

• Machine learning — Used to map weather and generation data to future output.
• Neural network — Used for photovoltaic and wind forecasting models.
• Multi-layer perceptron neural network — Used in distributed solar forecasting.
• Predict-then-optimize — Forecasts are fed into an optimization step for operating decisions.
• Explainable AI — Used to assess forecast trustworthiness and variable importance.

Industries

Real-world evidence

4 documented case studies on record.

Companies using this: Australian Renewable Energy Agency, Politecnico di Milano, EPFL, Google DeepMind.