The Detection And Quantification For Valve Stiction Based On Fuzzy Clustering

In the process industry,pneumatic control valves are the most widely used actuators,and their reliability and safety are directly related to the control performance of running control system.Since control valves are installed in the environment with high temperature and pressure or other complex conditions,working directly with the process medium for a long time,thereby larger friction between components is easily increased,leading occurrence of stiction nonlinearity.Valve stiction will make a great influence on the performance of actuator and the controller,in severe cases,may cause oscillation of the control loop.Therefore,the research of valve stiction is also an important part of performance monitoring for control system.To detect stiction nonlinearity and quantify its degree is basic for guiding engineers to take measures for valve maintenance and the performance improvement of the control loops,and is of great significance to realize the process industry running at “safety,stability,long run-length,full capacity and excellent” environment.The main contents of this paper are summarized as follows:1、 An improved method based on fuzzy clustering for detecting valve stiction is proposed.The existing detection method may easily make wrong diagnosis for detection of valve stiction nonlinearity,whose basic principles and problems of are analyzed in this paper.On this basis,the improvement is proposed which includes an additional part of the convex-shape identification and a new definition of stiction index.The proposed method can effectively recognize the control loop oscillation caused by the external disturbance or valve stiction and has the ability of anti-disturbance.An experiment study is used to verify the effectiveness and accuracy of the improved method for detection,and it is suitable for the flow control loops or the control loops equipped with smart valve.2、 A new algorithm to quantify valve stiction based on fuzzy clustering is proposed.Firstly the problem and the basic solution of stiction quantification are analyzed,then stiction nonlinearity quantification algorithm based on ellipse fitting method is discussed and studied,and then the validity and shortcomings of this kind quantification where the shape of fitted ellipse is sensitive to the controller parameters are analyzed.On the basis of the existing research,a new quantification algorithm is proposed and the effective distance between cluster centers is considered as estimated value of stiction.This algorithm is tested for the actual data in the process industry to verify the effectiveness of the method and its possibility use in future,completing the estimation of the degree of valve stiction nonlinearity.3、 From the practical application point of view,combined with existing experimental conditions,a system software for valve stiction detection and quantification is designed based on proposed method to promote the practical application.Detailed description of the software design and functional operation are to describe the practicality of the software,including the module function of the welcome interface,data import,image plotting,parameters of detection result display,toolbox and others,and the operation of each part are also displayed.At last,the research content of this paper is summarized and further discusses and prospects of the research are introduced.