MULTI-AGENT SIMULATION OF HUMAN ACTIVITY FOR RESIDENTIAL ELECTRICAL LOAD AND DEMAND FORECASTING AND ITS APPLICATION TO ELECTRIC MOBILITY

We present our agent-based architecture focusing on occupant behavior for energy consumption simulations in the residential sector. Based on cognitive ergonomics work, it is aiming at a holistic modelling of human activity, considered as the driver of energy consumption. In our simulations, each household's occupant decides its own actions thanks to a decisional process that puts multiple aspects of human activity in balance and is focused on the interactions between individuals and their material and social environment. This decisional process and this interaction with the environment are guided by the individual's knowledge, preferences and goals, giving the occupants the ability to react and adapt themselves to changes such as electricity price, new appliances or special events. To illustrate the aggregative ambition of the architecture, we present how mobility and actions related to the charging of electric vehicles (EV) were naturally added to the architecture. This new model, centered on both EV use and charging behaviors, aims at assessing the impact of EV on the residential consumption and the electricity distribution network. Finally, the validation processes of our architecture as well as the preliminary validation of the EV model are presented.