Automatic Detection of Stiction in Actuators: a Technique to Reduce the Number of Uncertain Cases

The presence of stiction in actuators is responsible for deterioration of performance in control loops: the possibility of automatic detection of this phenomenon from the analysis of recorded data is certainly very important and should be included in a monitoring system. In the paper it is shown that the application on plant data of a largely diffused technique (Horch, 1999), may not be able to detect the presence of stiction in some cases. More in general, application of the technique in simulation on different stiction models, shows that, by varying process parameters, stiction can be hidden and the uncertainty zone (gray zone: where is not possible to take any decision) may become very large. A new technique, based on the shape of loop signals originated by stiction, is then proposed. Even though it is not able to guarantee detection of friction phenomena in all cases, (impossible for every technique, without any further knowledge on the process), the proposed technique allows a significant reduction of the gray zone; this is illustrated by intensive simulation and by application on plant data. Copyright c ©2004 IFAC