Modelling and quantification of valve stiction by unknown input estimation

The presence of nonlinearities, such as stiction, and deadband in a control valve limits the control loop performance. Valve stiction is one of the most common causes of oscillations in industrial process control loops. In this work, we propose a novel approach to estimate the valve position using unknown input estimation. The estimation algorithm is a numerical method based on maximum likelihood. With estimated valve position, we can detect and also quantify the amount of stiction. The main advantages of the proposed method are numerical stability and computational efficiency. The efficiency of the method has been demonstrated through simulation examples.