Optimal Planning for Smart Energy Management of Residential Batteries

Tom Alard, Yvan Saeys, Arne Gevaert

Track: Type A (Regular Papers)

Keywords: Linear Programming, Time Series, Energy, Optimization

Abstract: In most residential batteries, the charge/discharge policy consists of a simple rule: charge whenever energy is abundant and discharge whenever it is scarce. However, this simple algorithm is far from optimal for households with dynamic energy contracts, where the price of energy can change every hour. In this work, we use time series forecasting on past energy production and consumption data to optimize the charge/discharge policy of residential batteries in households with solar panels and dynamic, day-ahead pricing energy contracts. The resulting method requires no external inputs, relying solely on past observations of energy production and consumption. We demonstrate that our method produces 80% optimal results on a historical dataset of Belgian households with residential batteries, which is more than twice as efficient as the baseline policy.

Serve As Reviewer: ~Arne_Gevaert1

Submission Number: 19