Agent Based Control


Agent Based Control Strategies for non-residential Buildings

Simple Model ABC Copyright: © RWTH Aachen Simple model for a multi-agent system for temperature control in an office room

Non-residential buildings with complex energy systems often do not reach the expected targets for energy efficiency and indoor comfort, Because of lack of effective coordination between resources and implementation issues of the centralized control system. This project proposes the use of multi-agent systems as a solution to this problem.

The new main building of the E.ON ERC will serve as a demonstration site for this new technology. This project focuses on the air conditioning system for the offices, as the installed technical equipment provides an excellent base for agent communication. Each office room is equipped with a user interface and a façade ventilation unit. The energy for heating and cooling, is delivered by water/air heat exchangers supplied by an array of central generation units: a heat pump, a combined heat and power unit and a boiler.

In a first phase an ontology for multi-agent systems performing energy management in buildings is defined, adapting and extending ontologies in literature. The introduction of new agents must be possible, without reconfiguring the existing architecture or changing the communications protocols. Furthermore data models for the application in the E.ON ERC main building are developed, as well as reference use cases.

A general method, based on a minimal set of variables, will be developed for the evaluation of user comfort. Then a decision making process for the optimal user comfort depending on the individuals present in the room will be developed. In this view, the possibility of adopting self-learning comfort agents will be investigated.

As the field of agent based control for building services is new, the system will first be developed and implemented in simulation, using Dymola / Modelica for the building and equipment models, and in JADE for the agents. Connecting the agents with the simulation models will be realized as software-in-the-loop. Subsequently the system will be implemented in the building.

In order to assess the performance of the method, the extensive monitoring system installed in the building will be used, particularly in reference rooms, where the energy supplied to the room is measured in detail.

We gratefully acknowledge the financial support of E.ON ERC gGmbH.

The final report of the project can be found here: ABC final report