Keeping up with the renewable energy transition
summary
Python framework that produces two-hour forecasts at five-minute resolution and predicts the critical 15–25 minute window used to size MFRR activations. It composes reusable data and transformer nodes, runs on OpenShift, and currently uses fast linear models (scikit-learn Ridge) with real‑time, historical and planned data inputs. Models are trained manually today but deployed for streaming inference and return predictions in seconds. Operational results show rest imbalance reduced vs benchmark across most bidding areas and for Sweden overall.
AI-written summary · about
chapters (9)
- 0:00Introduction to talk
- 1:12Speaker introduction
- 3:21Energy transition challenges
- 6:14Imbalance forecasting goals
- 10:23Model framework evolution
- 16:02New framework implementation
- 19:15Model performance metrics
- 20:36Summary and future work
- 20:56Q&A session
related talks
31mHow Python is supporting Australia’s energy transition to 100% renewable
10mWhy Your Forecasting Transformer Isn’t Working (And How To Fix It...
26mDriving towards a cleaner future through data
28mNixtla: Deep Learning for Forecasting...
22mFaster, Smarter Climate Mitigation Using Python 🌳
33mPython for Water Forecasting Services
31mUsing Python to drive Australian water availability forecasting
47mCreating a Contemporary Risk Management System Using Python