A win-win algorithm for aggregated residential energy management: resource optimisation and user acceptance learning

This paper proposes a solution based on Multi Agent System to study a residential Demand Side Management (DSM) program with a centralised approach. It focuses on minimising the cost considering different energy sources, such as photovoltaic panels and energy storage system, while optimally scheduling the appliances that can be shifted in time. The cost minimisation is formulated as a Mixed Integer Linear Programming (MILP) problem. The optimal allocation of the shiftable loads takes into account the modelled users’ preferences that are learnt by means of an algorithm based on an explore-exploit strategy. From the results, it emerges that a win-win situation could be achieved if user preference are considered.These benefits include savings and users’ satisfaction.