Middleware system utilises AI to coordinate building energy management
These days, buildings are carefully designed to be energy efficient. Yet, sometimes, they do not perform as intended. This discrepancy is called the energy performance gap. Construction can be inexact, introducing unintended gaps and other defects. Furthermore, once built, temperature control and ventilation systems may not perform exactly as expected. Building occupants may also behave unpredictably. Getting energy-efficient buildings to operate as intended is not simple. The EU-funded HIT2GAP (Highly Innovative building control Tools Tackling the energy performance GAP) project developed a new building energy management platform that takes the complicating factors into account to reduce the performance gap. The system, which includes modules that are based on Artificial Intelligence (AI) algorithms and techniques such as data mining and Knowledge Discovery, focuses on the operational phase of a building’s life, meaning the long period after construction and before decommissioning where tenants occupy the building. The project developed a new paradigm for building energy management platforms and marketed the resulting product.
Middleware platform and its modules store
The system is an open-source middleware platform called BEMServer. It receives information measured by building management systems and other sensor infrastructure. After centralised processing, various specialised modules access the information and provide actionable information to building managers and occupants. For example, the modules might identify faults or inappropriate settings concerning building equipment or alert to energy-inefficient user behaviours. All modules connect to the BEMServer. They can exchange information with other modules, making possible new data sets and combined analytics and therefore new advanced services.
User-created modules
Uniquely to HIT2GAP, modules can be created by third-party developers. BEMServer being deliberately open-source, developers do not have to recreate core functionalities of the middleware. “It’s a builders’ platform,” says project coordinator Pascale Brassier, “not unlike the iPhone operating system, whereby developers can create software for the platform to suit their own needs.” The system can therefore be modified and improved, and will remain up to date with emerging technologies. Fully commercial systems do not have this flexibility. Under the HIT2GAP concept, users only pay for the modules and not for the middleware BEMServer.
Enforcing Building Information Modelling with AI
Some of the modules utilise AI algorithms, including the Fault Detection and Diagnosis – Principal Component Analysis (FDD-PCA) module that utilises the HIT2GAP integrated data access to offer enhanced monitoring capabilities that evaluate building behaviour with respect to normal operation. If the module detects a fault, it allows for its isolation and for pinpointing the variables causing it, ensuring that maintenance crews can more easily identify causes or explain why the fault has occurred. Meanwhile, the Fault Detection and Diagnosis (FDD) uses machine learning based methods that continuously analyse time-series data from the BEMServer and provides a message to building operators if anomalies are detected. Finally, another module that utilises AI, ENIXO, is an application that produces recommendations about the energy waste situations in the whole system and sends messages to facility managers and other targeted users through the use of semantic reasoning. “So far, 12 modules have been delivered,” says Brassier. “BEMServer has been deployed at four pilot sites: in France, Ireland, Poland and Spain.” Each pilot selected the modules most relevant for their needs and for the different problems encountered in their building. BEMServer has been shown to work as intended, connecting buildings to information management modules. In testing, the system appropriately provided numerous warnings and alerts and helped building managers in decision making. Documentation for the system, and for most of the modules, has been made available to developers.
Next steps for BEMServer
Next, the team plans to develop further demonstration cases, and to extend the BEMServer through new or improved functions, including additional core functionalities based on AI. Future developments will focus on advanced preprocessing functions able to detect and correct data errors ensuring a minimum level of data quality. The team also plans to improve the user experience, including updated administrative and other functions. Soon, a second, more commercial version will be released to the marketplace.
Keywords
HIT2GAP, building, energy performance gap, BEMServer, modules, middleware, energy management, energy efficient, building management
