Detection And Diagnosis Method Of Control Loops Oscillation Based On Control Valve Stiction

Loop oscillation is a common problem in the process of industrial production,which is caused by improper controller parameters,valve stiction and external periodic disturbance.The existence of oscillation may result in the decrease of product quality,the increase of energy consumption and the shortening of equipment life,and so on,thus affecting the overall efficiency of equipment.Therefore,the detection and diagnosis of the control loop is of great significance to the process control,the main work of this paper is as follows:The method based on Empirical Mode Decomposition(EMD)and Auto-correlation Function(ACF)is applied to detect the oscillation of the control loops.The detected signal is decomposed into Intrinsic Mode Function(IMF)with different frequencies by EMD,and then the indexes of oscillation are calculated by the ACFs of the IMF signals The method can be used to identify multiple oscillating signals,and is robust to the noise signals.Finally,the oscillation period and amplitude of oscillation can be given..The simulation results show that the method can detect the oscillation accurately and reliably.Valve stiction is one of the main causes of the control loop oscillation in industrial processes.Most of the existing control valve stiction detection methods are based on specific assumptions,only for a specific range.Aiming at the limitation that area peak method only applies to the self-regulating plant,this paper presents an improved shaped analysis method to apply to both the self-regulating plant and integrating process.Considering the obvious difference of the second derivative of error signal waves in two different plants,a new index is added to further stiction detect.The method overcomes the limitation of the area peak method only applicable to a particular range.Finally,the simulation experiment validates the effectiveness and reliability of the proposed method.Several methods have been developed to address the diagnosis problem by focusing on only one of the causes for oscillations.In this work,an algorithm based on EMD and closed-loop testing principle is presented for identification of multiple root causes for oscillations in closed-loop systems.Making use of the property that EMD can decompose non-stationary time series into a finite set of IMFs,and then the closed-loop testing principle can exactly isolate the source of oscillation corresponding oscillatory IMF by injecting extra testing signal.The improved approach is used for determination of multiple causes for oscillation in linear SISO systems.Finally,the simulation experiment on the continuous stirring and mixing process validates the effectiveness and reliability of the proposed method.