Short-Term Forecasting of CO2 Emission Intensity in Power Grids by Machine Learning

Abstract

A machine learning algorithm is developed to forecast the CO 2 emission intensities in electrical power grids in the Danish bidding zone DK2, distinguishing between average and marginal emissions. The analysis was done on data set comprised of a large number (473) of explanatory variables such as power production, demand, import, weather conditions etc. collected from selected neighboring zones. The number was reduced to less than 30 using both LASSO (a penalized linear regression analysis) and a forward feature selection algorithm. Three linear regression models that capture different aspects of the data (non-linearities and coupling of variables etc.) were created and combined into a final model using Softmax weighted average. Cross-validation is performed for debiasing and autoregressive moving average model (ARIMA) implemented to correct the residuals, making the final model the variant with exogenous inputs (ARIMAX). The forecasts with the corresponding uncertainties are given for two time horizons, below and above six hours. Marginal emissions came up highly independent of any conditions in the DK2 zone, suggesting that the marginal generators are located in the neighbouring zones.The developed methodology can be applied to any bidding zone in the European electricity network without requiring detailed knowledge about the zone.

Control of heat pumps with CO2 emission intensity forecasts

“…[°C] 18 Case Real seeks to only switch on the heat pump during low emission periods. The radiator system does this well, but it is clearly limited by the maximum indoor temperature limit and the power input decays immediately to avoid temperature violation.…”

“…24 hour horizon CO 2 emission forecasts presented in the related paper, [18], are used. The real time values are presented in Sec.…”

“…This provides an upper limit of CO 2 savings. • Case Real : This takes the CO 2 emission forecast developed in [18] and represents the performance of the MPC with real forecasts. • Case Trivial : This makes no use of forecasts and will thus result in a non predictive controller that simply controls the heat pump keeping the temperature at the lower limit if possible.…”

“…A recent study, [18], investigates marginal emissions and uses estimates provided by Tomorrow 1 to develop a 24-hour forecast using a machine learning approach on historical data. The CO 2 estimates are calculated with the empirical approach developed in [19] using historical data from European bidding zones.…”