Modeling, Detection And Compensation For Control Valve Stiction

Pneumatic control valves are the most commonly used actuators in process industry for their high performance. Friction appears frequently in the packing boxes around the valve stem because control valves are in direct contact with process fluids when used in process industry field. Stiction or high static friction always causes oscillations in control loops which limits the control loop performance. With the increasing demand of the quality, the safety and the pursuit of economic profits, the processes are more complex that the performance can't be improved if ignoring stiction in control valves. So, stiction is being increasingly concerned by domestic and foreign researchers in recent years. Based on former researches and related references, the thesis focuses on stiction research from theoretical and practical aspect. The main contents contain the following parts:1,The phenomena of stiction is described and the causes of stiction are also analyzed. Stiction models based on physical principles and data-driven ones are listed and analyzed. Then an improved data-driven stiction model is proposed. And a simulation platform, which is developed to analyze stiction, is introduced to demonstrate the improved stiction model's usefulness. Furthermore, describing function analysis is used to analyze S and J parameters stiction model to gain insights into the conditions under which oscillations might arise.2,Two non-invasive methods of stiction detection are proposed, dealing with data from control loops. One of them is based on time series while the other on spectrum analysis. The effectiveness of the two proposed methods is demonstrated on our simulation platform by simulating and industrial data.3,Based on approaches for stiction compensation available, two improved PI control methods are developed to eliminate oscillations in control loops caused by valve stiction. One of them combines Proportional control with Proportional-Integralcontrol, while the other combines Proportional-Integral control with Fuzzy control. And the effectiveness of the improved PI control methods is demonstrated by simulating.