The ESO balances the electricity system in real time, relying on a mix of generation to meet constantly changing demand and ensure that, whatever the mix, electricity is always there when its needed.
Renewables are playing an increasing role in the GB electricity mix, in May 2019 the ESO reported the first ever fortnight of coal-free operation of Great Britain’s electricity system and aims to be able to operate a zero-carbon electricity system by 2025. Accurate forecasting of solar and wind generation is becoming essential to operating the system economically and efficiently, however, generation from these technologies is hard to predict – they are weather dependent and connected at a local rather than national level.
Researchers and doctoral students at The Alan Turing Institute helped the ESO explore ways of developing improved forecasting models for solar and wind power.
The new method stemmed from the Turing’s Data Study Groups (DSGs): week-long events bringing together data science, analytics, and mathematics to analyse real-world data science challenges. The ESO provided the challenge of investigating how a data-driven approach could help with real-time balancing of the electricity system.
A number of the methods proposed by the DSG participants had excellent potential for forecasting renewable energy generation. The ESO were keen to continue collaborating, so the ESO Energy Forecasting and Innovation teams worked with the Turing to scope an innovation project, funded by the NIA.
Historically, the ESO solar forecast took 2 basic variables, installed solar capacity and solar irradiance. Using a simple relationship between the two, they produced forecasts of solar generation output for different regions, summed up to give a total GB forecast.
The innovation project developed a new, random forest approach to arrive at the forecast. Looking at historic data and around 80 input variables, including temperature and much more granular solar irradiation data, the random forest model trains itself by finding hundreds of different mathematical pathways (or decision trees) to take those inputs, and arrive at the output generation figure. The model then forecasts by running the 80 new forecast weather variables through these decision trees, and takes the average as the new solar generation forecast.
The ESO took the new approach, and combined it with several other machine learning techniques in a multi-model ensemble forecast. Enhancing the project’s initial output with these different machine learning methods, ESO has built a solar forecasting system which is 33% more accurate.
Rob Rome, Commercial Operations Manager at the ESO, said: “Renewable sources of power are becoming a bigger part of the energy mix so it’s vital our forecasts of their output are as accurate as possible. The ESO’s dedicated innovation team are always looking at new techniques and methods to help us balance the system and this partnership with The Alan Turing Institute is a great example. Improved solar forecasts will help us run the system more efficiently, ultimately meaning lower bills for consumers. It will also enable more solar capacity to be connected and utilised, helping us to achieve our 2025 ambition to be able to operate a zero carbon electricity system.”
Andrew Duncan, Data-Centric Engineering Group Leader at The Alan Turing Institute added: “The project has opened up a lot of new avenues and ESO are interested in pushing other projects forward. There’s no shortage of problems to tackle".