Areas of interest: Probabilistic methods in civil engineering, risk analysis and decision-making, structural reliability, random vibrations and earthquake engineering.
Current Research Projects:
Stochastic modeling and simulation of near-fault ground motions for PBEE
A parameterized stochastic model of near-fault ground motion is being developed and it can be used to simulate an ensemble of synthetic near-fault ground motions for a specified set of earthquake source and site characteristics. Near-fault ground motions may or may not contain a forward directivity pulse and the model accounts for both pulselike and non-pulselike cases. The resulting synthetic motions are of particular interest in performance-based earthquake engineering. For example, they are being used in a simulation-based probabilistic seismic hazard analysis at a near-fault site.
Graduate Student Researcher: Mayssa Dabaghi
Funding Source: California Department of Transportation through PEER.
Tail equivalent linearization for nonlinear stochastic dynamics
The tail-equivalent linearization method (TELM), originally developed by former doctoral student K. Fujimua, is being developed in greater detail with a sound mathematical foundation. A mathematical formulation in the context of functional spaces is introduced to shed light on a comprehensive vision of the method, which includes time-domain,
frequency-domain, and orthogonal-polynomial representations. The method is further developed to accommodate multicomponent excitation and temporal and spectral non-stationarity. The effectiveness and convenience of the method to study the statistic of nonlinear systems is utilized and integrated with a decision-making framework. Integration within a Bayesian network is planned.
Graduate Student Researcher: Marco Broccardo
Funding Source: Taisei Chair in Civil Engineering.
Bayesian network methods for modeling and reliability assessment of infrastructure systems
In this research, novel compression and inference algorithms are being developed to enable the modeling of large systems as Bayesian networks (BNs). These algorithms are applied to the analysis of infrastructure systems, e.g., power networks, to perform probabilistic system assessment and reliability analysis. The objective is to create a BN framework that utilizes uncertain and evolving information to support decision making in the management of large systems.
Graduate Student Researcher: Iris Tien
Funding Source: Natural Sciences Foundation.
Bayesian network for structural health monitoring
Bayesian network is a powerful tool for probabilistic modeling, inference and decision-making, and its application in structural health monitoring is being investigated in this research project. Due to the ubiquitous uncertainty in civil structures and health monitoring systems, the performance of usual deterministic approach is not satisfying in most cases. We seek a natural combination of mature structural analysis methods and the Bayesian network for damage detection, localization, assessment, prediction and management. A dynamic Bayesian network framework for modal identification has been developed, and basing on this, a probabilistic finite element model updating strategy will be proposed.
Graduate Student Researcher: Binbin Li
Funding Source: Natural Sciences Foundation.
Infrastructure management using data sensing and computer-aided stochastic interpretation
The research focuses on data-driven engineering solutions to improve the sustainability and resilience of infrastructure. The technology is available for monitoring structures in ageing and hazard-prone cities. Despite this, data interpretation solutions for generating added value from structural health monitoring data cannot yet justify costs. My work is addressing this challenge with three research axes:
(1) Knowledge extraction for city-scale sensor networks - The objective is to develop affordable and scalable strategies for anomaly detection and post-earthquake rapid condition assessment.
(2) Autonomous data interpretation systems - The objective is to develop interpretation systems capable to detect automatically when a normal structural response signature is no longer valid and to self-adapt to changing conditions.
(3) Data-driven infrastructure management - The objective is to create engineering decision frameworks for data-driven infrastructure management (post-earthquake and in-service).
This research is providing solutions to systematically manage post-hazard structural condition assessment and constrain the rise of infrastructure maintenance costs for advanced and emerging economies.
Postdoctoral Researcher: James-A. Goulet
Funding Source: Swiss National Science Foundation and Quebec National Funds for Research in Natural Sciences and Technology.