Robust Step-Like Identification of Low-Order Process Model Under Nonzero Initial Conditions and Disturbance

Nonsteady initial process states, measurement noise and unexpected load disturbance are practical difficulties associated with model identification from step response tests. A robust identification method is proposed to overcome these practical problems. The proposed step-like test differs from the conventional step test in that not only the process transient response to the step change, but also the subsequent transient response from removing the step change, are used for model identification. Based on a general low-order model structure, linear regression equations are established through multiple integrals for parameter estimation. The influence of nonzero initial process states and load disturbance is specifically considered in such a linear regression equation. A feasible instrumental variable (IV) method is also given with strict proof for consistent estimation against measurement noise. Illustrative examples from the recent literature are performed to show the effectiveness and merits of the proposed identification method.