The present PhD Thesis provides an overview on the results of a research project at
the Vienna University of Technology since 2012. The publications originated in the
course of a cooperation project between the Institute of Mechanics and Mechatronics
(Division of Control and Process Automation), the FH Joanneum Kapfenberg (University
of applied science) as research partner, and evon-automation GmbH, Gleisdorf as
industrial partner. The project has been funded by the Austrian Research Promotion
Agency (FFG No. 832103).
A contemporary issue of potentials for saving energy in buildings is discussed in this
work. The research was focused on the development of new methodologies for smart and
energy-efficient building automation systems. In this context a new nonlinear model
predictive control strategy has been developed for a specific building. Moreover, the
commissioning in a this real building succeeded with excellent results.
In this PhD Thesis a nonlinear model predictive control (MPC) concept for complex
office buildings is presented. Conflicting optimization goals naturally arise in buildings,
where the maximization of user comfort versus the minimization of energy consumption
poses the main challenge. MPC technologies are able to reduce the energy demand
while increasing the user comfort, by taking weather predictions and/or occupancy information
into account. Dynamic thermal behavior of buildings is typically nonlinear,
thus, for controlling a suitable model is necessary. In this work the overall nonlinear
building model is a data-driven black-box model, which can be directly used for
controller design. The proposed modeling approach is applicable for other complex
buildings. For the demonstration building the complex nonlinear optimization problem
has been split into a set of less complex subproblems (different building zones). For
each zone an independent nonlinear MPC (fuzzy MPC - FMPC) is designed. Because
of an integrated thermal activated building system couplings between different zones
occur, thus, the optimization goal is to find a cooperation between the zones. The
overall problem formulation leads to a cooperation of the FMPCs, to a cooperative
fuzzy MPC (CFMPC), where an underlying cooperative iteration-loop guarantees convergence.
Closed-loop stability and convergence of the cooperative iteration-loop has
been proven for the CFMPC concept. Additionally, this concept is suitable for complex
multi-zone office buildings, as it can optimally deal with input and output constraints
as well as with disturbances.