Prediction of electrical consumptions using a bio-inspired behavioral model,

Continuously growing electrical energy requirements have become an issue as resources to produce it are limited. Numerous studies have shown that human behavior plays a key role in buildings consumption. Usual stochastic behavioral models rely solely on databases to model behaviors. Thus once the database becomes obsolete due to changes on habits or users, the whole modeling process has to be carried out again. We propose a new Multi-agent model based on a psychological scheme. It is able to simulate energy consumption accurately and to generate realistic profiles of users that are not in the database. We conclude showing that the bayesian framework is a valuable tool to establish a link between psychological model and its computer implementation.