Discovering Complex Knowledge in Massive Building Operational Data Using Graph Mining for Building Energy Management

Abstract

Discovering useful knowledge from massive building operational data is considered as a promising way to improve building operational performance. Conventional data analytics can only handle data stored in a single two-dimensional data table, while lacking the ability to represent and analyze data in complex formats (e.g., multi-relational databases). Graphs are capable of integrating and representing various types of information, such as spatial information and affiliations. The knowledge discovery based on graph data can therefore be very helpful for revealing complex relationships in building operations. This study proposes a novel methodology for analyzing massive building operational data using graph-mining techniques. Two problems are specifically addressed, i.e., graph generation based on building operational data and knowledge discovery from graph data. The methodology has been applied to analyze the building operational data retrieved from a real building in Hong Kong. The research results show that the knowledge obtained is valuable to characterize complex building operation patterns and identify atypical operations.