| KEYWORD |
Area Engineering
Evaluation of measurement uncertainty in Digital Twin by Bayesian Statistics
keywords BAYESIAN STATISTICS, CALIBRATION, DIGITAL TWIN, DYNAMIC CONTROL, MEASUREMENT SYSTEMS, MEASUREMENT UNCERTAINTY
Reference persons MAURIZIO GALETTO, GIANFRANCO GENTA, GIACOMO MACULOTTI
Research Groups Ingegneria della qualità
Thesis type APPLIED RESEARCH
Description A Digital Twin (DT) is a virtual model of a physical object/system. It runs the lifecycle of the object/system and uses real-time data sent from sensors on the object/system to simulate behavior and monitor operations. Furthermore, DTs allow real-time control of the physical objects/systems to which they refer. The control is aimed at correcting and compensating for errors and, through simulation, predicting and preventing damage to the physical object/system itself by implementing appropriate control strategies. The measurement uncertainty of the DT is essential for associating a confidence interval with the prediction on which to base controlling decisions. The measurement uncertainty depends on several factors, foremost among which are the sensors and actuators of the system as well as errors in modeling, diagnosis, and prognosis (fault prediction). The measurement uncertainty of the DT system (physical plus virtual entities) depends on the downstream control, where the measured data are used to predict the system conditions at subsequent time points. The measurement uncertainty is therefore also affected by this closed-loop control. Dynamic Bayesian Networks are typically used to evaluate the model parameters. The thesis activity will assess the effect of control measurement uncertainty on the estimation of parameters through Bayesian Statistics. The thesis will then test the modeling on measurement systems at the Metrological Room of Mind4Lab at the Department of Management and Production Engineering (DIGEP). The thesis is part of the EU-funded ViDiT project.
Required skills Data analysis, simulation and statistical analysis, fundamentals of Bayesian Statistics, fundamentals of industrial metrology.
Deadline 31/12/2024
PROPONI LA TUA CANDIDATURA