Data-Based Method To Diagnose Valve Stiction with Variable Reference Signal

Abstract

Stiction is a well-known villain in industry because of the limit-cycle imposed on the controller. Several methodologies are reported in the literature to automatically detect this problem using only normal operating data. However, this becomes more difficult when the loop with stiction is affected by disturbances or the sticky valve is inside a cascade loop. This study proposes two methods to automatically diagnose valve stiction when the reference signal is variable and centers primarily on recognizing triangular or sinusoidal patterns. The first method is based on the slope of the signal peaks and the second on data segmentation. These techniques were compared to a curve-fitting method, providing similar results when the reference is fixed. However, for processes significantly affected by disturbances or when the sticky valve was inside a cascade loop, stiction detection was better for both methods proposed. These results are corroborated by simulation and industrial data.

Stiction detection in low sampling rate signals

“…Yamashita method [16] -Recognizes the direction of movement in the MV(OP) diagram, then the sequence of directions is classified in typical sequences for the case with stiction and general combinations. Peak slope method [21] -Calculates the slope of each peak and valley of the signal and compares with the slope of the standard triangular and sinusoidal signals. Curve-fitting method [18] -Fits sinusoidal and triangular functions for each signal half-cycle.…”

“…The relation between the number of directions in each class is used for oscillation diagnosis. Zone segmentation method [21] -Calculates the signal distribution for each half-cycle and compares it with the distribution of the standard triangular and sinusoidal signals. Better fit to triangular function indicates the presence of stiction, better fit to the sinusoidal function indicates another cause of oscillation.…”

“…The two methods proposed in Dambros et al [21] are applicable to processes with a variable mean, caused mainly by external disturbances or cascade controller. The two methods proposed in Dambros et al [21] are applicable to processes with a variable mean, caused mainly by external disturbances or cascade controller.…”

“…Better fit to triangular function indicates the presence of stiction, better fit to the sinusoidal function indicates another cause of oscillation. Peak slope method [21] -Calculates the slope of each peak and valley of the signal and compares with the slope of the standard triangular and sinusoidal signals. Slope closer to the slope of the triangular signal indicates stiction presence; closer to the sinusoidal signal, another cause of oscillation.…”

“…The regularity indexes [23] are respectively equal to 1.82 and 10.1 (where an index larger than 1 indicates regularity), only one oscillation is found, and the signals do not have a variable mean based on visual inspection. The stiction index for all methods and different sampling is compared with the index for the original signal (dashed lines) extracted directly from the original works of Jelali and Scali [6] and Dambros et al [21] and is shown in Figure 12 and Figure 13.…”