May 15–17, 2017 in Prague, Czech Republic
[Proceedings] [Sessions] [Authors] [Schedule] [Further material]

Papers by Dirk Müller:

Title: A Platform for the Agent-based Control of HVAC Systems
Authors: Roozbeh Sangi, Felix Bünning, Johannes Fütterer and Dirk Müller
Abstract:The amount of energy used for heating and cooling in the building sector is about one third of the total energy consumed in the world. The finiteness of natural energy resources on the one hand, and the ever-increasing demand for energy in the world on the other hand, necessitate the development of systematic approaches for improving the efficiency of building energy systems as well as minimizing the usage of primary energy resources and the damaging impacts on the environment. Attempts to tackle these problems have led to modern complex energy concepts for buildings, which have consequently initiated a need for new control strategies for them. Multi-agent control, which is known with other names like agent-based control, offers a promising solution to these challenges. To the knowledge of the authors, there are 96 platforms in different programming languages available, which are mostly java-based and mainly used in logistic applications, but there is no platform in the modeling language Modelica, which is widely used for simulation of dynamic systems, especially buildings performance simulation. This lack motivated the authors to develop a platform for agent-based control of HAVC systems. The platform eliminates the dependency of models developed in Modelica on an extra interface, which is usually required to couple the models to the platforms written in any programming languages other than Modelica. This paper presents the structure of the platform and explains how the agents' communications work. The flexibility of the optimization objective is ensured through the definition of readily interchangeable cost functions. The applicability and functionality of the platform are proved by applying the platform in the control of building energy systems examples.
Links: Full paper, Additional material

Title: Dymola-JADE Co-Simulation for Agent-Based Control in Office Spaces
Authors: Ana Constantin, Artur Löwen, Ferdinanda Ponci, Kristian Huchtemann and Dirk Müller
Abstract:This paper presents an application of coupling Modelica under Dymola and JADE to test novel agent-based control for office spaces. The office space with a coupled energy system and weather boundary conditions are modeled in Dymola. The agent platform is programmed in JADE, where the agents communicate with each other to control the technical equipment used to deliver thermal energy to the room. Heating experiments, run for a one room scenario, using a radiator, show better system reaction to the comfort desires of the user when compared to a control with a thermostatic valve, while having similar energy consumption. While the agents run in real time, the simulation in Dymola runs more quickly. We focus on the particularities of the connection for co-simulation to insure smooth transferability of the experiments from simulation to field test, where the energy system as well as the agent platform would be running in real time.
Links: Full paper