Research Axes

  /main page/research axes

Bayesian Dynamic Linear Models for Structural Health Monitoring

(BDLM-SHM)





Shervin Khazaeli
Ph.D. | 2018 - 2023

Bhargob Deka
Ph.D. | 2018 - 2022

Dai Vuong Van
Ph.D. | 2019 - 2023

Luong Ha Nguyen (D)
Postdoc | 2019 - 2021 
Ph.D. | 2017 - 2019
M.Sc. | 2015 - 2016

Ianis Gaudot (D)
Postdoc | 2017 - 2019


Catherine Paquin (D)
M.ing. | 2016 - 2018


Simon Brousseau (D)
M.ing. | 2017 - 2018



Summary: This research axis aims at developing new probabilistic methods for interpreting in real-time the data recorded on structures. The goal of data interpretation is to detect anomalies in the behaviour of structures in order to trigger preventive actions such as visual inspections and maintenance.

Keywords: Bayesian Dynamic Linear Models; State-Space Models; Continual Learning; Decision Making; Reinforcement Learning; Bayesian Neural Networks; Multivariate Time-Series; Structural Health Monitoring; Preventive Maintenance; TAGI 

Funding & Industrial Partners:
Hydro-Qubec/IREQ/NSERC

MTQ
PJCCI

Infrastructure Degradation




Zachary Hamida
Postdoc | 2021 -
Ph.D. | 2017 - 2021

Blanche Laurent
M.Sc. | 2020 - 2022

Ali Fakhri
M.Sc. | 2021 - 2023


Summary: This research axis aims at developing probabilistic models capable of forecasting the degradation of bridge structures along with planning tools for supporting decision making. This research builds upon the information contained in safety inspection reports from a population of structures. 

Keywords: State-Space Models; Network Analysis; Decision Making; Reinforcement Learning; Bridge Degradation; Visual Inspections; TAGI

Funding & Industrial Partners:
Quebec Transportation Ministry (MTQ)


Material Behavior modeling





Lucie Tabor (D)
M.Sc. | 2015 - 2017


Summary: This research axis aims at creating probabilistic models from empirical data. The goal is to capture the behaviour and intrinsic variability observed in the laboratory for advanced materials.

Keywords: Gaussian Process Regression; Heteroscedasticity; Bayesian Methods; Fiber Reinforced Concrete (UHPFRC)

Soil Contamination Caracterization

Alyssa Quach (D)
M.ing. | 2015 - 2016

Summary: This research axis aims at developing an artificial intelligence capable of assisting engineers during the characterization of contaminated sites. The goal is to enable the selection of an optimal soil sampling sequence and locations in order to minimize site characterization costs.

Keywords: Artificial Intelligence; Pre-Posterior Analyses; Gaussian Process Regression; Optimization; Soil Contamination

Industrial Partner: WSP | Parsons Brinckerhoff