Energy signature-based clustering using open data for urban building energy analysis toward carbon neutrality: A case study on electricity change under COVID-19

Abstract

Under the necessity for urban energy savings and the importance of energy demand changes, this paper proposes energy analytics of urban buildings using a novel clustering method with open data. In the proposed method, the concept of energy signatures is introduced, and the signatures are defined as the representative symbols in the symbolic hierarchical clustering. This method can advance the existing clustering method or symbolic aggregate approximation (SAX) with limited energy usage patterns by introducing energy signatures with various pieces of energy information into the symbolic transformation. The proposed method can scientifically support tracking building energy usage and patterns, evaluating the existing concepts (such as green retrofitting and zero energy buildings) and advanced technologies, and the decision-making process for new policies under the global carbon neutrality scenarios. In a case study applied to a city using open energy data (in South Korea), the proposed method determined five representative clusters/areas, revealing the open data quality problems (anomaly and missing data), energy usage changes (e.g., energy usage polarization) caused by COVID-19, and the necessity of classifying building types in terms of energy usage patterns. Specifically, the cluster for decreasing energy patterns accounted for approximately 37%, and the increasing patterns accounted for 25%. Educational buildings accounted for 70% of the decreasing patterns, and technical and medical research facilities accounted for 76% of the increasing patterns under the COVID-19. Approximately 93% of missing data was found in the residential buildings. Anomaly data accounted for 10.9% in the total data.