A cloud-based Energy Information System (EIS) for innovative energy management in buildings: the case of Politecnico di Torino

The building sector accounts for more than a third of global energy consumption, and nearly 90 percent of total energy consumption in the building lifecycle depends on its operation. In recent years, the spread of IoT technologies in buildings and the adoption of pervasive smart-metering systems have enabled the acquisition of a massive amount of high-frequency energy-related data. Leveraging this data to extract and formalize knowledge is essential to characterize the actual performance of buildings during operation and take action to reduce energy consumption, prevent energy waste, and promote a more efficient way of managing buildings. A valuable tool employed to monitor, analyze and control building energy systems by taking advantage of advanced data analysis technologies are so-called energy management and information systems (EMIS). EMISs are often designed as monolithic software deployed on physical servers and thus are unable to scale properly to support computationally demanding real-time applications. In addition, EMISs are usually tailored to the building-specific monitoring system, which leads to a lack of interoperability and raises challenges when integrating different advanced services based on data-driven techniques. This work presents the design of microservice-based Energy Management and Information System cloud architecture and the implementation of a forecast and anomaly detection application. The architecture, based on the Kubernetes container cluster, enabled fine-grained system decoupling, reliability, scalability, and ease of system maintenance while optimizing resource utilization, interoperability, and integration, creating a robust environment to analyze cross-domain data and developing innovative data-driven EIS services. The forecast and anomaly detection application was tested and deployed online on the photovoltaic plant of the Politecnico di Torino campus, leading to the development of a tool useful to support energy management through an effective prediction of energy demand at a daily scale and anomaly alerting.