Using the Entropy for Typical Load Curve Classification

Clusters of energy consumers can be utilized to improve the accuracy of load forecasting or to introduce differentiated and personalized tariffs according to the consumers energy demand. In this paper, we aim to classify energy load curves according to their similarity with other households. We present the entropy as quantitative metric for Typical Load Curve Classification and clustering. Furthermore, we present the likelihood of time periods to uniquely distinguish load curves of residential households and approximate the minimal required time resolution for classification tasks. Results, using a real world data set, we confirm the practical relevance and feasibility of our approach.