Continuous-time model identification for closed loop control performance assessment

The paper describes a system identification procedure for continuous-time transfer function models and demonstrates how it can be used for control performance assessment. Two normalized indices are derived for detecting oscillations and sluggish performance. The methods are designed for single-input single-output (SISO) control loops that experience periodic setpoint changes. An advantage of the system identification procedure is that it can operate on irregularly sampled data. The paper presents results from simulation tests and from tests on an heating, ventilating, and air-conditioning (HVAC) system serving a large commercial office building.