ETFA 2009 14th IEEE INTERNATIONAL CONFERENCE ON EMERGING TECHNOLOGIES AND FACTORY AUTOMATION September 22-26, 2009, Mallor lorca, ca, Spain C ONTROL SYSTEM AND FAULT DETECTION ALGORITHM FOR A RESTORED TEETH FATIGUE ASSAY MACHINE I. Peñarrocha, M. Orellana ipenarro@esid.uji.es Dept. d’Enginyeria de Sistemes Industrials i Disseny. Universitat Jaume I de Castelló, Spain
S UMMARY Introduction Machine requirements Machine and control system description Machine operation Fault detection algorithm Experimental results Conclusions 2 ETFA 2009 , September 22-26, 2009, Mall llor orca, Spain in
INTRODUCTION Necessities for dental restorations in Dentistry field Optimal design of the restored teeth crown stump Posts of different material and geometry root post Tested with time-varying forces Knowing which component fails first adhesive 3 ETFA 2009 , September 22-26, 2009, Mall llor orca, Spain in
M ACHINE REQUIREMENTS Test different posts (material and geometry) of restored teeth Emulate the mastication forces Application of time-varying forces Number of mastication cycles before the failure of the tooth Fault detection Knowing which component fails first Automatic stop of forces application 4 ETFA 2009 , September 22-26, 2009, Mall llor orca, Spain in
FATIGUE ASSAY MACHINE DESCRIPTION 5 ETFA 2009 , September 22-26, 2009, Mall llor orca, Spain in
CONTROL SYSTEM DESCRIPTION 6 ETFA 2009 , September 22-26, 2009, Mall llor orca, Spain in
MACHINE OPERATION Force calibration: Torque reference (u k ) vs. Measured force (F k ) 7 ETFA 2009 , September 22-26, 2009, Mall llor orca, Spain in
MACHINE OPERATION Specimen fatigue assay From required force (F k ) to Torque reference (u k ) Measured position (y k ) depends on the fault (f k ) 8 ETFA 2009 , September 22-26, 2009, Mall llor orca, Spain in
FAULT DETECTION ALGORITHM Model-based fault detection Noisy model: Prediction model: Prediction error: Expected (fault free) prediction error: This error can be bounded during the identification process Fault detection algorithm: Failure occurs with probability of 99.74% if 9 ETFA 2009 , September 22-26, 2009, Mall llor orca, Spain in
FAULT DETECTION ALGORITHM Identification of the prediction model: LTV model 10 ETFA 2009 , September 22-26, 2009, Mall llor orca, Spain in
FAULT DETECTION ALGORITHM Prediction error 11 ETFA 2009 , September 22-26, 2009, Mall llor orca, Spain in
FAULT DETECTION ALGORITHM FDA: RLS Adaptive threshold 12 ETFA 2009 , September 22-26, 2009, Mall llor orca, Spain in
FAULT DETECTION ALGORITHM Tuning process Forgetting factor: As big as possible to detect slow changes on the process Bigh enough to avoid changes on the parameters due to faults Filter pole used to approximate : Big enough to approximate properly Small enough to initiate the application False alarm ratio related parameter: Small enough to detect the faults as fast as possible Big enough to avoid false alarms 13 ETFA 2009 , September 22-26, 2009, Mall llor orca, Spain in
EXPERIMENTAL RESULTS Prediction error and threshold 14 ETFA 2009 , September 22-26, 2009, Mall llor orca, Spain in
EXPERIMENTAL RESULTS Tooth failure 15 ETFA 2009 , September 22-26, 2009, Mall llor orca, Spain in
CONCLUSIONS Developement of a fatigue assay machine for studying the dynamical behaviour of restored teeth The control system has been described The machine operation has been described The fault detection algorithm has been addressed The tunning process has been carried out Experimental results have been discussed 16 ETFA 2009 , September 22-26, 2009, Mall llor orca, Spain in
ETFA 2009 14th IEEE INTERNATIONAL CONFERENCE ON EMERGING TECHNOLOGIES AND FACTORY AUTOMATION September 22-26, 2009, Mallor lorca, ca, Spain C ONTROL SYSTEM AND FAULT DETECTION ALGORITHM FOR A RESTORED TEETH FATIGUE ASSAY MACHINE I. Peñarrocha, M. Orellana ipenarro@esid.uji.es Dept. d’Enginyeria de Sistemes Industrials i Disseny. Universitat Jaume I de Castelló, Spain
Recommend
More recommend
