Stiction Estimation Using Constrained Optimisation and Contour Map

Abstract

A closed-loop method for valve-stiction detection and quantification based on Hammerstein modelling is discussed. A suitable model structure of valve stiction is chosen prior to conducting valve-stiction detection and quantification. Given the stiction-model structure, a bounded search space of a stiction model is defined and a constrained optimisation problem is performed. The best unknown stiction-model parameters are found by satisfying a mean-squared error criterion within a space of valve stiction model parameters. For this purpose, a multi-start adaptive random search is used. The proposed strategy not only detects but also quantifies valve stiction. The validity of the proposed method is illustrated through industrial examples. Also, the closed-loop identifiability issue is addressed.