ENAC Seminar Series by Dr A. Olivier
Event details
| Date | 23.09.2019 |
| Hour | 16:00 › 17:00 |
| Speaker | Dr Audrey Olivier |
| Location | |
| Category | Conferences - Seminars |
16:00 – 17:00 – Dr Audrey Olivier
Postdoctoral Research Scientist, Johns Hopkins University, USA
Data-assisted high-fidelity modeling for monitoring of civil systems
Increased availability of measured data has recently generated tremendous interest in the development of methods to learn from data. In parallel, engineers have a long history of building high-fidelity physics-based models that allow us to model the behavior of highly complex systems. At the intersection of these two topics, data analytics and high-fidelity modeling, lie exciting opportunities to address the challenges of modern civil engineering.
Monitoring the health of our aging or historic infrastructure and predicting its response to future events is a challenging task, in part due to the presence of various uncertainties in the inputs, measurements and the system itself. Probabilistic system identification methods such as Bayesian inference use measurements from a system to learn its equations and parameters, thus allowing detection of potential damage, while accounting for the various uncertainties. However, Bayesian techniques become computationally expensive for inference in large dimensional nonlinear systems, i.e., finite element models. We demonstrate the potential of Bayesian filtering techniques and algorithmic enhancements to reduce computational cost, and how to integrate these probabilistic learning algorithms into complex frameworks of model selection and optimal design of experiments.
Whether it relates to monitoring the health of our infrastructure, improving its resilience to natural disasters or designing the smart cities of tomorrow, data sensing and analysis is becoming an integrative part of civil engineering research and practice. Very interestingly, the combination of data-mining and physics-based modeling also finds applications in various engineering fields. In the materials sciences for instance, interesting opportunities lie in the development of scientific machine learning to speed-up materials discovery. Research in this field could thus benefit from and impact various fields of science and engineering.
Postdoctoral Research Scientist, Johns Hopkins University, USA
Data-assisted high-fidelity modeling for monitoring of civil systems
Increased availability of measured data has recently generated tremendous interest in the development of methods to learn from data. In parallel, engineers have a long history of building high-fidelity physics-based models that allow us to model the behavior of highly complex systems. At the intersection of these two topics, data analytics and high-fidelity modeling, lie exciting opportunities to address the challenges of modern civil engineering.
Monitoring the health of our aging or historic infrastructure and predicting its response to future events is a challenging task, in part due to the presence of various uncertainties in the inputs, measurements and the system itself. Probabilistic system identification methods such as Bayesian inference use measurements from a system to learn its equations and parameters, thus allowing detection of potential damage, while accounting for the various uncertainties. However, Bayesian techniques become computationally expensive for inference in large dimensional nonlinear systems, i.e., finite element models. We demonstrate the potential of Bayesian filtering techniques and algorithmic enhancements to reduce computational cost, and how to integrate these probabilistic learning algorithms into complex frameworks of model selection and optimal design of experiments.
Whether it relates to monitoring the health of our infrastructure, improving its resilience to natural disasters or designing the smart cities of tomorrow, data sensing and analysis is becoming an integrative part of civil engineering research and practice. Very interestingly, the combination of data-mining and physics-based modeling also finds applications in various engineering fields. In the materials sciences for instance, interesting opportunities lie in the development of scientific machine learning to speed-up materials discovery. Research in this field could thus benefit from and impact various fields of science and engineering.
Practical information
- General public
- Free
Organizer
- ENAC
Contact
- Cristina Perez