The two distinct problems currently being tackled by Smyth and his collaborators involve 1) improved vibration based health monitoring and system identification of bridges and other structures through data fusion algorithm development and data normalization, and 2) the improved modeling and simulation of the 3D rocking dynamics of objects resting on a moving base. The goal of both of these research thrusts is the improved resilience of infrastructure systems. In structural health monitoring, there are large challenges in extracting the most accurate assessment information from noisy response measurements and unknown excitation. Different types of sensors provide varying levels of fidelity in different frequency ranges. Combining this information using data fusion concepts to get "more than the sum of the parts" has been a major goal. For the rocking dynamics modeling, the goal has been to go beyond what is predominantly a 2D analysis approach and to include important nonlinear effects of interface sliding, bouncing, and base medium deformability and damping.
The research in structural health monitoring is motivated by real world challenges encountered in the monitoring of several major long-span bridges (predominantly in NY City). For example, the incorporation of differential GPS displacement sensing into accelerometer monitoring networks provides for accurate information in the low frequency range to be merged with the accelerometer measurements. This merging or fusion is done through a dual state parameter estimation framework, in which the dynamic response states are estimated based on the noisy measurements from different types of sensors (including also strain gages), and the system parameters to be identified are treated as time invariant states to be estimated. Current research has centered on the use of noncollocated heterogeneous sensing, questions of nonlinear observability and the use of Bayesian estimation approaches for highdimensional systems. For the rocking problems considered, the research approach focuses on the 3D block dynamics while introducing as much realism in the interface interactions. The models are developed analytically, but given their highly nonlinear nature, involving many switching equations associated with the contact mechanics, they are solved numerically. The interactions with the tensionless support medium have been modeled using various interaction approaches, and the development is generalizable to even more complex nonlinear interactions.
For health monitoring, we have developed tools engineers can use to incorporate dynamic measurements from different sensors to help to efficiently (and often adaptively) identify nonlinear and linear system parameters. For the rocking problem we have developed new 2D and 3D models which have demonstrated the importance of 3D modeling, and have also highlighted the importance of including sliding and bouncing phenomena in the modeling.
The structural health monitoring algorithms have been applied in real large scale bridge systems to identify bridge deflections with high accuracy. Current work on observability will help engineers plan their allocation of sensors to achieve the desired performance from a defined sensor budget. It is hoped that the very recent rocking models will provide a tool for engineers to better asses risk to critical components which may be susceptible to overturning in seismic regions or in shipping contexts.
- Chatzis M.N., Smyth A.W., "Modeling of the 3D rocking problem",
International Journal of Non-Linear Mechanics
, 47 (4), pp. 85-98 (2012).
- Chatzi, E.N., Smyth, A.W., "Particle filter scheme with mutation for the estimation of time-invariant parameters in structural health monitoring applications"
Structural Control and Health Monitoring
, in press, (2012).
- Mosquera, V., Smyth, A.W., Betti, R., "Rapid evaluation and damage assessment of instrumented highway bridges",
Earthquake Engineering and Structural Dynamics
, 41 (4), pp. 755-774, (2012).
- Wu, M. Smyth, A.W., "Application of the unscented Kalman filter for real-time nonlinear structural system identification,"
Structural Control and Health Monitoring
14 (7) , pp. 971-990 (2007)
- Smyth, A., Wu, M., "Multi-rate Kalman filtering for the data fusion of displacement and acceleration response measurements in dynamic system monitoring",
Mechanical Systems and Signal Processing
, 21 (2), pp. 706-723, (2007).
- Pei, J.-S., Wright, J.P., Smyth, A.W., "Mapping polynomial fitting into feedforward neural networks for modeling nonlinear dynamic systems and beyond",
Computer Methods in Applied Mechanics and Engineering
194 (42-44) , pp. 4481- 4505 (2005).