A Comparison of MILP and Metaheuristic Approaches for Implementation of a Home Energy Management System under Dynamic Tariffs

This paper compares two different methodological approaches - a mixed-integer linear programing (MILP) model and a metaheuristic (a genetic algorithm, GA) – to be embedded in a Home Energy Management System (HEMS) with the aim to make the integrated optimization of energy resources under dynamic tariffs. Different types of demand-side resources, including shiftable, interruptible and thermostatically controlled loads as well as local generation and storage, have been considered. The objective is to minimize the electricity cost including the monetization of the dissatisfaction of end-users with possible changes of load operation. Since these two objectives are in conflict, a compromise solution is sought according to the end-user's profile. Different end-users' preferences are considered embodying different sensitivity levels of end-users to the cost and the energy service satisfaction.