| Biography | |
|---|---|
 Prof. Stefano Mariani Department of Civil and Environmental Engineering of the Polytechnic University of Milan, Italy |
|
| Title: Multi-scale deep learning model for uncertainty quantification at the microscale | |
| Abstract: Smart, bio-inspired devices can be characterized by a hierarchy of length- and time-scales, and by different physical phenomena affecting their properties. Data-driven formulations can then result helpful to deal with the complexity of the multi-physics governing their response to the external stimuli. Referring to a single-axis Lorentz force micro-magnetometer adopted for navigation purposes, if an alternating current with an ad-hoc set frequency is let to flow longitudinally in a slender beam, the system is driven into resonance and the sensitivity to the magnetic field to be sensed may get enhanced. In former activities, a reduced-order physical model was developed for the movable parts of the device, to feed a multi-physics and multi-objective topology optimization procedure. This model-based approach did not account for stochastic effects, which are responsible of the scattering in the experimental data at this micrometric length-scale. A recently proposed formulation is here discussed to allow for such stochastic effects through a multi-scale deep learning model featuring: at the material scale, a deep neural network adopted to learn the scattering in the mechanical properties of polysilicon induced by its morphology; at the device scale, a further deep neural network adopted to learn the most important geometric features of the movable parts that affect the overall performance of the magnetometer. Some preliminary results are discussed, along with a proposal to frame the approach as a kind of multi-fidelity, uncertainty quantification procedure. | |
| Biography: Dr. Stefano Mariani received an M.S. degree (cum laude) in civil engineering in 1995 and a Ph.D. degree in structural engineering in 1999, both from the Polytechnic University of Milan.He is currently an associate professor at the Department of Civil and Environmental Engineering of the Polytechnic University of Milan. He was a research scholar at the Danish Technical University in 1997, an adjunct professor at Penn State University in 2007, and a visiting professor at the Polytechnic Institute of New York University in 2009.He is a member of the Editorial Boards of Algorithms, International Journal on Advances in Systems and Measurements, Inventions, Machines, Micro and Nanosystems, Micromachines, and Sensors. He has been a recipient of the Associazione Carlo Maddalena Prize for graduate students (1996), and of the Fondazione Confalonieri Prize for PhD students (2000). His main research interests are: reliability of MEMS that are subject to shocks and drops; structural health monitoring of composite structures through MEMS sensors; numerical simulations of ductile fracture in metals and of quasi-brittle fracture in heterogeneous and functionally graded materials; extended finite element methods; calibration of constitutive models via extended and sigma-point Kalman filters; multi-scale solution methods for dynamic delamination in layered composites; smart materials and structures. | |
