A Decentralized Approach on the Energy Management in Buildings

Abstract

Buildings are accounted to consume around 40% of the total energy in Europe and USA, even more – up to 60% - in Canada and Scandinavian countries, and the rate of growth of the number of buildings follows the growth of the population. In this context, there is a vast literature dedicated to finding solutions for saving energy through innovative control systems of the heating, ventilation and lighting in residential and commercial buildings. This paper proposes a solution for controlling the HVAC system in small and medium size buildings by means of a distributed neural network, implemented using low cost microcontrollers, connected with passive infrared (PIR) motion detectors. This network is capable to detect and learn user occupancy and activity patterns, relevant for the energy management of the building. Unlike other similar systems, in our approach the energy management is entirely decentralized, relying exclusively on the above mentioned distributed ANN.