The Detection And Quantification Of The Valve Stiction

The control system in the process industry are essentially nonlinear,and the nonlinear process is generally researched as the linear one for the linear object is rare. The valve is the most commonly used actuator in the control loops, its performance will affect the control loops directly. According to statistics, about30%of control loops in the process industry are in a state of oscillation, in which1/3due to the valve stiction. Therefore, stiction is the most common problem in the industrial control precess, and its existence seriously impacts on the performance of the control system.The detection and quantification of stiction is of great importance to improve the accuracy of the control loops. In recent years, a lot of researchers have presented different kinds of theories and methods for the stiction modeling and detection, but there is still a lack in the field of the quantification.In this paper, firstly, an analysis is conducted based on the stiction models which established by foreign researchers.A new model is presented which can accurately reflect the true characteristics of the valve after the analyzing on the advantages and disadvantages of all the models. And it is chosen as the standard. Secondly, the methods in common use about detection are researched in this paper. However, one or more limitations in the area, such as speed, accuracy, simplicity constrain the application of the theories and methods. Moreover, there is still a lack in the field of the stiction quantification. Based on the theories and applications of the previous research, a new detection method based on the wave segmentation and fitting and a new quantification method based on the modified fuzzy C-means clustering algorithm are presented in this paper. Then, the method is applied in the real process industry mainly as motor-vehicle chassis industry. Lastly, this method is applied to numerous processes by software.Numerous simulations are conducted according to existing stiction model and he real process industry data. The result of simulations show the reasonable of the model and effectiveness and accuracy of the new detection and quantification method. The results in this paper can not only be applied to the detection and quantification of the control loops, but also improve the product quality, reduce energy consumption and maintain the stability of the process, as well as reduce the intensity of the operation, and so on. The study of the stiction quantification can not only fill in the gaps in our country, but also significantly increase the economic efficiency of the process industry with its broad and huge prospect of application.