Diagnosis of Poor Control Loop Performance: Investigate Support Vector Regression Method on Stiction Quantification for Control Valve Nonlinearity

The control valve in process control undergoes wear and aging which causes valve stiction. The valve stiction problem effects the process performance in loops and eventually the quality of the product. Stiction study consists of four main parts, stiction modelling, stiction detection, stiction quantification and stiction compensation. The quantification technique for the valve stiction is very important aspect in process control. We consider that the presence of stiction in valve can be calculated using few techniques. The objective of this paper is to develop support vector regression (SVR) for the valve stiction nonlinearity. This technique is less explored and very rarely applied in stiction quantification. A proper algorithm has been introduced for the SVR based stiction quantification. The algorithm had been demonstrated through simulation based on multiple cases studies. The results of the experiment has been interpreted and analyzed. It also has been compared with other methods based on the efficiency and accuracy