Heuristic Optimization Approaches for the Smart Home Appliance Scheduling Problem

With the emerging of smart grids as an answer to the need of sustainability, it grows in relevance the ability to manage the energy at our disposal. This is of utmost importance in residential settings where the demand is dependant on the family behaviors and the possibility of having other distributed energy sources (i.e. Solar Panels, Plug-In Electric Vehicles). Reliable energy Management Systems help in this task by scheduling the appliances in order to optimize a specific aspect of the energy demand. This thesis studies the effect of different optimization approaches for the power scheduling problem in residential settings. We start from a Mixed-Integer Linear Program as a base and build an heuristic approach based on the Tabu Search algorithm, improved with an expanding neighborhood. Moreover, the problem is formulated as a Markov Decision Process and a Dynamic Programming formulation of the problem is developed. The different approaches are then compared and analyzed with different pricing schemes and different Renewable Energy Sources settings, such as Solar Panel power generation on different weather conditions.