Closed-loop step identification of low-order continuous-time process model with time delay for enhanced controller autotuning

Abstract

A continuous-time model identification method is proposed based on a closed-loop step response test for obtaining a low-order process model to facilitate control design in industrial engineering practices. By introducing a damping factor to the closed-loop step response for realisation of the Laplace transform, a frequency response estimation algorithm is developed in terms of the closed-loop system configuration used for identification. Correspondingly, two model identification algorithms for obtaining the widely used low-order process models of first-order-plus-dead-time (FOPDT) and second-order-plus-dead-time (SOPDT) are derived analytically respectively. To improve model fitting accuracy over a specified frequency range interested to controller tuning, in particular for identification of a higher order process, another identification algorithm is given based on weighted least-squares fitting of the process frequency response points estimated in the specified frequency range. Using the identified low-order models, a model-based controller design and tuning method is proposed for improving load disturbance rejection. Illustrative examples from the recent literature are used to demonstrate the effectiveness and merits of the proposed identification algorithms and controller tuning method.