Fostering Data Fluency: Building a Strong Data…
Estimate the wind power production profile by comparing weather forecast for the next day with a history of past forecasts
Even if there are technological solutions for storing electricity, they are not available at the moment for large scale deployment. Thus, the balance between production and consumption must be constantly checked on the electricity grid.
In order to forecast the evolution of the load, power grids need to have accurate algorithms. However, the development of decentralized renewable energies installed directly on the distribution network, (for example wind and photovoltaic power plants) makes this task more difficult.
Consumption and production are highly weather-sensitive, and monitoring energy trades on the lines is strongly related to forecast of meteorological hazards .
Sia Partners designed a bot based on the “Analog Ensemble” method to provide probabilistic forecasts for wind power production on wind farms. Getting wind farms power production in a production history, makes it possible to estimate the wind power production profile for the next day.
This bot uses a 2017 history composed of the real power production data from a French major energy company and the weather forecasts of the corresponding location from Etalab.
The idea is to simulate the wind on one day thanks to a Weibull distribution function (with customizable shape and scale) and to use the Analog Ensemble method to identify analogs and draw the resulting power production forecast.
If you wish to learn more about Back to The Future bot, use the link bellow to access a demo !
To learn more about Heka and its ecosystem