Application of adaptive multivariable Generalized Predictive Control to a HVAC system in real time

This paper presents the application of a Multivariable Generalized Predictive Controller (MGPC) for simultaneous temperature and humidity control in a Heating, Ventilating and Air-Conditioning (HVAC) system. The multivariable controlled process dynamics is modeled using a set of MISO models on-line identified from measured input-output process data. The controller synthesis is based on direct optimization of selected quadratic cost function with respect to amplitude and rate input constraints. Efficacy of the proposed adaptive MGPC algorithm is experimentally demonstrated on a laboratory-scale model of HVAC system. To control the air-conditioning part of system the designed multivariable predictive controller is considered in a cascade dual-rate control scheme with PID auxiliary controllers.