An Ensemble Method for Aggregated Baseline Load Estimation: From Probabilistic Perspective

Demand response (DR) is regarded as an alternative for new infrastructure to meet the peak load, which curbs the greenhouse gas emission. As an intermediary agent, DR aggregator (DRA) needs to comprehend aggregated baseline load (ABL) in the DR event. However, there are a few studies estimating ABL from an aggregator's perspective. Therefore, this work proposes a new ensemble method for probabilistic ABL estimation, which utilizes multiple clustering algorithms with varying cluster numbers to produce multiple customer divisions. And the combing weights for the ABL results, produced by the multiple customer divisions, are determined by linear programming problems. We demonstrate the proposed method using real world dataset, Low Carbon London trail. For probabilistic estimation, it has good performance with high credit. We show that no single individual method can achieve good performance all the time. Ensemble method is a better choice to obtain relatively stable and good results.