Times series forecasting for urban building energy consumption based on graph convolutional network

The world is increasingly urbanizing, and to improve urban sustainability, many cities adopt ambitious energy-saving strategies through retrofitting existing buildings and constructing new communities. In this situation, an accurate urban building energy model (UBEM) is the foundation to support the design of energy-efficient communities. However, current UBEM are ineffective to capture the inter-building interdependency due to their dynamic and non-linear characteristics. Those conventional models either ignored or oversimplified these building interdependencies, which can substantially affect the accuracy of urban energy modeling. To fill the research gap, this study proposes a novel data-driven UBEN synthesizing the solar-based building interdependency and spatio-temporal graph convolutional network (ST-GCN) algorithm. Especially, we took a university campus located in the downtown area of Atlanta as an example to predict the hourly energy consumption. Furthermore, we tested the feasibility of the ST-GCN model by comparing the performance of the ST-GCN model with other common time-series machine learning models. The results indicate that the ST-GCN model overall outperforms in different scenarios, the mean absolute percentage error of ST-GCN is around 5%. More importantly, the accuracy of ST-GCN is enhanced when simulating buildings with higher edge weight and in-degrees, this phenomenon is magnified in summer daytime and winter daytime, which validated the interpretability of the ST-GCN models. After discussion, it is found that data-driven models integrated with engineering or physics knowledge can significantly improve urban building energy use prediction.