SingaporeEnergy and utilities2021 – ongoingEnergy Market Authority
Machine-learning solar forecasts for grid operations
EMA completed a solar forecasting trial with average normalised error below 10%
Problem
Singapore's changing cloud cover and rainfall made solar generation difficult to anticipate. Unexpected changes could create imbalances between electricity supply and demand.
Approach
SERIS at the National University of Singapore developed the model with support from EMA and Meteorological Service Singapore. It combined rooftop irradiance measurements with satellite imagery, machine learning and numerical weather predictions. A year-long trial at EMA's Power System Control Centre ended in September 2022.
Results
Takeaway
The trial measured forecast accuracy before integration into dispatch systems. EMA still planned the operational system upgrade when it announced the results.
Categories
- Size
- Large enterprise
- Industry
- Energy and utilities
- Business function
- Production and quality
- Innovation type
- Process innovation
- AI technology
- Predictive analytics and ML
- Adoption
- Partner co-development
- Goal
- Quality and risk
- Evidence
- Company's own claim
Sources
Source summaries
EMA reported completion of the control-centre trial and average normalised forecast error below 10% up to 1 hour ahead. It planned to incorporate the forecasts into its Energy Management System by 2023 and provide them for electricity-market scheduling.
Full details are in the original sources.
