Publications and deliverables
Deliverable 4.1
Title: | Semantic-aware energy optimization algorithm |
Due date: | 2017-08-31 |
Executive summary:
This document summarizes the results of work package 4, i.e. the design of the semantic-aware energy optimization algorithm. In general, the optimization tries to find a schedule of using the available resources in an efficient way while satisfying comfort needs of building users. Thus, an optimization solution, i.e. a schedule, is calculated by a hybrid algorithm that minimizes the sum of comfort dissatisfaction and energy costs. The JAVA-based optimization component is integrated as all other components into the proof-of-concept system. Communication with the semantic core and its ontology uses the semantic interface. In order to ensure universal applicability of the optimization approach, the necessary information is modeled in the ontology, which was developed in work package 2. Then, configuration of the optimization is done automatically by reading this context information and initializing the optimization problem. Moreover, this deliverable addresses the inclusion of results from the clustering, correlation analysis, and pattern detection done in work package 3. An optimization run is executed prior to a period of time that should be optimized. Thus, the optimizer is not a real-time control optimizer but a scheduler that plans the set point changes and takes into account external and internal influences.
The semantic-aware energy optimization is characterized as human-in-the-loop system as there is always interaction with the users of the building. Even if all devices are controlled automatically, the user is still involved through the defined requirements and constraints. In order to see the influence of the user behavior of applying different set points to the heating system instead of the optimized suggestion, a parameter variation of different heating set points was made by means of an EnergyPlus simulation.
Additionally, work package 4 deals with the integration of demand response use cases from work package 1 into the optimization system. Thus, the inputs and outputs, which are exchanged between the optimization system and the various actors in the smart grid, are defined per use case. Finally, the integration of the use cases into the optimization procedure is discussed.