Estimation of distribution function for control valve stiction estimation

There are many methods for detection and estimation of control valve stiction. There exists, however, to the best of authors’ knowledge, no method to determine the statistical property of the stiction detection and estimation. The challenge lies in the memory nonlinearity of the stiction that creates difficulty in determining statistical property of the estimation. In this work, a bootstrap based approach is used to estimate the statistical distribution of stiction parameter estimation. The method is applied to a simulation example as well as several industry data sets to demonstrate its effectiveness.

► 30% control performance problems are caused by valve fault, particularly stiction.
► No method available to estimate the uncertainty of stiction quantification.
► We develop a method to estimate statistical distribution of stiction parameters.
► The results are verified by simulations as well as by real industrial data.