Full Description

The present paper addresses the modeling of a heating, ventilating, and air-conditioning (HVAC) system while such a system is operating under normal conditions. This so-called on-line identification problem is the first part of an adaptive controller design. Several recursive estimation techniques are reviewed for application to HVAC systems. The Recursive Maximum Likelihood (RML) method is selected based on its outstanding performance. A model of single-zone HVAC system is derived. From the multi-input multi-output (MIMO) nature of the model, a modification procedure for the RML algorithm is designed to enable a SISO (single-input single-output) tool to accommodate a MIMO system. The modified algorithm is then applied to the simple single-zone model and yields stable and convergent, parameter estimates, which are in excellent agreement with the parameters in the simulator.

Units: SI

Citation: Symposium, ASHRAE Transactions, 1987, vol. 93, pt. 1, New York, NY