Instrumentação e identificação de um ambiente predial visando controle preditivo do conforto térmico / Instrumentation and building identification aiming predictive control of thermal comfort




The technological basis of Ambient Intelligence is a network of processing units that collect sensor information, process them and act on the environment through actuators, providing the occupants a diversity of services. In this scenario, this manuscript presents a first-principles structured identification approach for a thermal environment consisting of a conference room of the Automation and Robotics Laboratory (LARA) located at the University of Brasília (building SG11), aiming an efficient model that can lead to successful controller designs suitable for energy-saving and comfort control strategies (ISO7730), such as predictive HVAC control. The data is obtained in a wireless sensor network using the ZigBee/IEEE 802.15.4 protocol. The obtained first-principles model simplifies the parameters identification, associating physical meaning to the internal model states and allowing easier parameters adjusts in case of environment changes, with no need to proceed the entire identification process, as in the black box method. Since the parameters acquisition from first-principles structured model is not trivial, some procedures for obtaining them were evaluated. In particular, good results were achieved using stochastic filtering methods, such as the extended Kalman filter.


automação predial ambient intelligenct engenharia eletrica identificação de sistemas modelamento first-principles

Documentos Relacionados