Silvia Vettori

Short Bio

Silvia obtained B.Sc. (2015) and M.Sc. (2018) degrees in Mechanical Engineering from University of Rome “La Sapienza”. In 2017 she moved to Belgium for a 6-months internship at Siemens Digital Industries Software (Leuven) aimed at developing new data processing techniques for aircraft Ground Vibration Testing. Since September 2018, she is involved as an Early Stage Researcher in the Marie Curie DyVirt PhD program with Siemens Digital Industries Software and ETH Zurich. Her research topic consists in the development of Virtual Sensing techniques for dynamic virtualization of structures.

Research

The working principle of Virtual Sensing techniques for structural dynamics applications consists in combining information from cost-effective simulated models and more realistic test data to calculate an estimate of the quantity of interest, e.g., excitations or responses of a system at locations that are difficult to be instrumented via physical sensors. The so-called "virtual sensor" consists in a quantity inferred at a location where a physical sensor could not be positioned, both in-service or during experiments. In a structural dynamics environment, these techniques can be exploited to reconstruct the full field response of a structure to dynamic loads, e.g. operating wind turbines, in terms of accelerations, displacements or strains. Moreover, the dynamic loads these structures are subjected to, e.g., wind and rotor dynamics in the wind turbine case, could also be inferred.

The outcome contributes to establish test-validated Digital Twins of structures, i.e., digital models featuring encoded information on the structure "as-is". The latter can help effectuate the monitoring of the performance of these systems throughout the structural life cycle. Several existing Virtual Sensing techniques are being explored by Silvia. Her focus mainly regards Kalman-type filters, such as the Augmented Kalman Filter or the Dual Kalman Filter, which are well known techniques for joint input-state estimation. Additional investigated methods are Modal Expansion techniques and further joint input-state estimation techniques.

The main objective of Silvia’s research consists in finding new methods and solutions to practical limitations of the mentioned estimators. Bayesian estimators, such as Kalman-type filters, deal with the need of building noise models that can well represent both measurement and process noise, i.e., noise that takes into account errors in the adopted system model. Moreover, when performing input estimation, a model for the input must also be included. Input modeling and noise statistics selection represent important topics in Silvia’s research.