Effect of Forecaster Performance on Peak Shaving in a University Building by Battery Scheduling

摘要

The objective of this project is to develop and evaluate a load forecasting technique that gives input to a demand management algorithm for a building in the presence of a Time-of-Use demand rate structure. The objective of the demand management controller is to minimize the annual electricity bill with the help of behind-the-meter energy storage by shifting the load partially from peak pricing periods. The forecasting algorithm uses a two-stage method where a deterministic part uses an exponentially weighted moving average (EWMA) model accounting for longer-term trends and a third-order regression model (AR3) accounting for the short-term variations. It is shown that a very small increase in forecasting error can result in a significant loss of the annual savings achieved by a demand response controller. A building in University Campus with a highly dynamic load profile is considered for studies.