Use of two-stage least-squares algorithms for identification of continuous systems with time delay based on pulse responses

This paper presents the so-called two-stage least-squares algorithms to deal with practical identification difficulties often encountered in field testing, such as unsteady initial states, unknown load disturbances, and noise-corrupted measurement. For step and ramp responses, a general linear regression equation is derived from multiple integration of the differential system equation. Four types of pulse inputs are then considered, each of which has its specific advantage in applications and can be represented as a combination of step or ramp inputs. Based on any of these pulse responses, the algorithms are able to overcome those difficulties and yield accurate parameter estimates in a sequential manner. The resulting identification scheme provides a convenient way to give the delay estimate along with system parameters and is robust with respect to noise and model structure mismatch.