Coordinated optimization of robustness and flexibility of building heating systems for demand response control considering prediction uncertainty

Abstract

Robust strategies are essentially needed in building energy systems for meeting load demands stably under uncertainty, whereas demand response (DR) can serve the grid through flexible operation and achieve remarkable economic benefits. However, formulating optimal control strategies that enable weighing robustness and flexibility for energy systems is quite challenging and rarely addressed owing to their conflicting control objectives. In this paper, a dual-objective coordinated optimization approach is proposed and applied to the heating system of an office building equipped with heat pumps, to balance the robustness and flexibility of DR control under building load uncertainty. The load distribution interval is predicted by the quantile regression neural network model. And optimal DR strategies are performed via modulating indoor temperature set-points. Using the genetic algorithm and multi-objective decision-making optimization framework, the optimal strategy can be obtained by weighing the confidence level of load prediction interval and the operating cost of flexible operation. The results show that, compared with the robust optimal strategy, the coordinated optimal strategy reduces the operating cost by 35.8% with robustness that meets actual demands in 80% of the cases. Compared to the flexible optimal strategy, it sacrifices only 0.05% cost but increases the load guaranteed hours by 33.3%.