Master's Thesis Michael Verunica


Integration of a complex thermal Model into a modelpredictive Control with respect to real-time Requirements

Increasing demand for volatile energy storage has shown that bulk memory storage, which balances
generation- and consumption- fluctuations, gets more important in contemporary research. Those
technologies became important to the usage in storage power stations, because of rapid developments
in the section of electrochemical storage. Nevertheless, a more efficient application is only
manageable through ultra-flexible storage management systems. Those systems take thermal behavior
of the battery systems into account, since high temperatures have a significant influence on
the durability of the sensitive batteries. The thesis at hand considers this aspect and occupies the integration
of a thermal plant model into a model based predictive regulation. Regarding the example
of the M5BAT – project, different integration concepts, based on state observers, will be presented.
With respect to a small computation and the particular suitability for application with optimization
problems, attempts were made to transfer the complex plant model into a linear shape. It showed
that especially the discrete switching operations were hard to predict. A non linear state observer
achieved significantly better results, whereby, however, the convergence of the observer needs to be
estimated critical.