Study of MultiResolution Classification Accuracy On Numerical Model Data

James Garrett, Civil and Environmental Engineering

Structural Health Monitoring (SHM) is a major concern for Bridge Authorities. In the aftermath of the 2007 I-35W Mississippi River bridge collapse in Minneapolis, a 2010 U.S. Department of Transportation report rated 26% of the country's nearly 600,000 bridges either structurally deficient or "functionally obsolete". Since then, the search for reliable structural health monitoring (SHM) technologies able to prevent catastrophic failures or to flag early signs of early deterioration became a top priority.

To date, besides mandatory visual inspection, the SHM of bridges is conducted using direct approaches where sensors are deployed in the structure and measurements are streamed to a centralized unit for data cleansing, feature extraction and classification, to provide diagnostics. This approach faces several practical problems and great direct and maintenance costs

We have been working on a paradigm shifting approach for indirect monitoring. This approach has the advantages of decentralizing the monitoring apparatus to fleets of vehicles that can continuously store or send data. The objective is cost-effective and sustainable assessments of a large population of bridges.

 Project description: 1D oscillator-beam interaction and 3D vehicle bridge interaction models have been developed. They enable the simulation of the on-going oscillator or vehicle dynamic motion as it travels over the supporting structure. Making use of a MultiResolution classification algorithm we have determined that variations in the supporting structure affects the response of the passing vehicle in such a way that the algorithm is able to accurately classify among variations of locations and severity of inertia reduction. Making used of the tools already developed and modifying them, further studies should incorporate uncertainties such as roadway roughness and measurements and noise. They should also incorporate other damage types as frozen bearings, localized cracks and damping change.

Objective: To determine the influence of uncertainties such as roadway roughness and measurements and noise in the classification accuracy of vehicle’s response when traveling under different damage conditions such as frozen bearings, localized cracks and damping change.

If interested, please contact Professors Jim Garrett (garrett@cmu.edu) and
Jacobo Bielak in CEE to get more details.