Design of non-linear model predictive control for temperature stabilization of a parabolic trough solar-thermal power plant

Kurzfassung

At the present time, Nonlinear Model Predictive Control (NMPC) has already found its application to parabolic trough solar-thermal power plants, based on oil as a heat transfer fluid. Besides that, certain commercially successful configurations of such systems, making use of water-steam separators in the solar field, with water as a heat transfer fluid already exist. Nevertheless, there still remains a significant direction for further improvement and price reduction of this renewable energy technology. One promising option are plants with the so-called once-through mode and water as a heat transfer fluid. Such technical config-uration would afford to achieve a higher efficiency in comparison to the oil-based parabol-ic troughs plants, while considerably reducing the necessary investments in comparison to the plants making use of additional water-steam separators. However, the once-through-based parabolic trough plants are challenging to control. This master thesis is focused on the design of an NMPC for the stabilization of the outlet temperature of a parabolic trough solar power plant. Furthermore, the concern of observer design, as of an obligatory prerequisite for an NMPC algorithm, is addressed. In the framework of the present master thesis an observability analysis of the given system is performed and a few most commonly used state estimation techniques are examined. As a result, some of them get discarded and eventually particle filter is chosen, since it is more appropriate for nonlinear state estimation than the Kalman-filter-based methods, like extended Kalman filter or unscented Kalman filter. After the subsequent development of the NMPC algorithm, its performance is compared to a control scheme based on proportional-integral controllers. Based on results of that comparison, conclusion are drawn and suggestions for future research are given.