Data Acquisition, Detection and Estimation for Structural Health Monitoring

Missouri University of Science & Technology

Maggie Cheng
Associate Professor
Missouri University of Science and Technology
325 Computer Science
Rolla, MO 65409
(573) 341-7501
chengm@mst.edu

One graduate student

Although using sensor networks for SHM (structural health monitoring) is not a new concept, very few projects have investigated the problems of detection (of defects) and estimation (of damage location) using network-acquired data. In statistics detection and estimation theory were established by assuming the measurement data come with reliable statistics, for instance, the probability of a particular observation. However, such statistics often requires large amount of observations. In wireless sensor networks, data acquisition is a costly operation since wireless sensor networks are both bandwidth and power limited. The amount of measurement data that can be reported to the base station is therefore very limited. Data acquisition from sensor networks has been treated as a trivial subject and often is performed by using fixed-interval sensing and reporting. In this project, we will provide a thorough treatment of sampling, detection and estimation for using sensor network data. Specifically, (a) We will investigate the fundamental sampling issue, particularly, for each type of physical measurement, what is the best sampling rate and whether adaptive sampling is more suitable than uniform sampling. Based on the sampling discipline, the sensing and communication protocols are developed; (2) for structural defect detection, we propose to use the likelihood ratio test method with Bayes criterion and compare it with the basic LRT method; through the detector, we narrow the scope of the defect to be within the spatial interval of some sampling points; (3) once it is concluded that a defect exists, the maximum likelihood estimator is used to further estimate the location of the defect. The algorithms will be validated thorough test bed experiments or simulations.

Relationship to other Research/Projects

This is the continuation of the previous year’s NUTC project. Previous year’s work focused on structural health monitoring in the presence of data anomaly and network fault. So far we have developed a good
understanding of the spatial and temporal dependence structure of measurement data.
The new project will focus on the energy and bandwidth efficiency of the data acquisition system and its fitness in a wireless network environment, and most importantly, improving the timeliness of detection and the accuracy of estimation of the damage location. The knowledge acquired from the previous project will greatly reduce the learning curve of the new project.

Transportation-Related Keywords

Technology Transfer Activities

The findings and results will be summarized and presented at the annual UTC conference and other professional conferences as well as the regular CIES meetings.

Project Deliverables

Anticipated Benefits

The project will result in a low-cost and yet highly-effective way for structural health monitoring. Without direct inspection of the structure, the damage to the structure can be detected timely. Moreover, the
deployment of wireless network does not interfere with the normal operation of the bridge. Finally, wireless sensor network is more cost-effective than the wired counterpart.

Milestones