LSTM-based Load Curtailment Prediction in Demand Response System

Demand response (DR) which reduces energy consumption at energy crisis is an interesting technology in the field of smart grid. However, it has a special characteristic that its energy reduction is very unstable. This uncertainty can break the balance between energy supply and demand because we may miscalculate the real energy demand. In this paper, the load curtailment prediction algorithm is proposed. A two-layer long short-term memory (LSTM) is applied based on the past energy consumption data. Real data from two demand response service providers are utilized to develop and validate the proposed prediction model. In the result, the proposed model provided 31%p improved result compared to the baseline past averaging method.