Increased energy efficiency in energy management in buildings through refined process modeling and optimization strategies


Energy management systems aim to control energy, cooling and heating systems in such a way that they take into account both the energy-efficient operation of these components and the comfort requirements of the residents. Weather-dependency and living habits are also important influencing factors. The hypothesis of the project was that substantial efficiency increases can be realistically expected through refined methods of modeling and optimization. Experiments with partners have shown that by integrating heuristic rules based on local weather forecasts, savings potentials of 5% in heating energy in winter and 10% cooling energy in summer are possible. In order to ensure the transferability of these savings potentials to individual building types, living situations and technical infrastructure, automated methods of Model Predictive Control (MPC) and Transfer Learning are required to derive the control rules from measurement data and data-based forecasts by learning from refined models. Such an approach is the basis for a higher degree of adaptability in control and thus also for the exploitation of further savings potentials.


In view of the increasing use and inclusion of alternative energy sources (photovoltaics, solar energy, biomass), the question arose of optimising the interaction of all energy producers and consumers across all energy sources in order to use the existing energy sources as energy-efficiently as possible, i.e. to create an optimal holistic energy management for a building. This enabled an automated adaptation of the control system (e.g. heating curve optimization).


Ganzheitliches Energiemanagement


Therefore, we had the following project goals:

  • Development and analysis of widely applicable methods from the field of Computational Intelligence for the creation of process models
  • Creation of concrete subsystem models for the areas of energy yield (PV, solar), energy consumption (electricity, heating, hot water...), and energy storage
  • Development and integration of semantically enriched models of user behavior
  • Development and evaluation of concrete MPC applications
  • MPC test environment based on building simulations
  • Integration in and evaluation of a real test object
  • Development of methods for simple parameterization and implementation of MPC approaches


In the project, a self-learning building model was developed that can take into account both user behaviour and weather forecasts in order to ultimately reduce energy costs while maintaining the same level of comfort.


 Zusammenspiel zwischen Energiekosten und Komfort

An mpcEnergy test environment was developed to simulate energy management tasks. This allows different optimization methods to be selected and evaluated for their potential savings. The operation of this mpcEnergy test environment is shown in the demonstration video below.

Project Data

Duration: 01.07.2012 – 31.12.2014

Budget: ca. 596.000 € total costs and ca. 298.000 € funding contribution


  • Software Competence Center Hagenberg GmbH (coordinator)
  • Johannes Kepler University Linz, Research Institute for Symbolic Computation (RISC)

Funding Program: 1st Upper Austrian Energy Research Programme

The project mpcEnergy was supported within the program Regionale Wett­bewerbs­fähigkeit OÖ 2007-2013 by the European Fund for Regional Development as well as the State of Upper Austria.

Video (77,5 MByte)



Bernhard Freudenthaler

Freudenthaler Bernhard

Area Manager Data Science
Phone: +43 50 343 850