Due to the 2012 change in the renewable energy act the feed-in tariffs, and therefore the number of newly installed photovoltaic systems decreased dramatically in Germany. Particularly in the residential sector as the biggest market new business ideas for photovoltaic systems were developed. Hence a photovoltaic and combined heat-and-power system, which provides not only electricity but also heat. This complex system requires flexible control strategies. A new combined control algorithm is proposed that in contrast to the standard strategies can operate even under incorrect weather and load forecasts without creating discomfort. Furthermore is it applicable for cloud-based solutions. In this paper a high-level model predictive control based on a mixed integer linear programming problem is combined with an additional low-level, rule-based controller. The low-level control compares the set-points of the optimization with the actual values and corrects the set-points according to each system component until the next optimization takes place. The results show that a hybrid system can be successfully controlled by a combined control approach. In case of a cloud-based application the need for an optimization can be reduced by a factor of four without diminishing comfort or facing much higher operational costs. It has also been shown that the combined control algorithm can be used as an energy management of microgrids. (C) 2017 Elsevier Ltd. All rights reserved.