Structural Health Monitoring Using Smart Sensors - ideals ...
Structural Health Monitoring Using Smart Sensors - ideals ...
Structural Health Monitoring Using Smart Sensors - ideals ...
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Data processing<br />
Data acquisition<br />
Figure 2.2. Centralized data acquisition approach.<br />
Data acquisition & processing Processing<br />
Figure 2.3. Independent data processing approach.<br />
Figure 2.2). Centralized SHM algorithms are then applied to this data. This approach<br />
allows for application of a wealth of traditional SHM algorithms reviewed in 2.3<br />
<strong>Structural</strong> <strong>Health</strong> <strong>Monitoring</strong>. As the number of smart sensors increases, however, the<br />
measurement data to be centrally collected exceeds the network bandwidth, whether<br />
homerun or hopping communication is adopted. The lack of scalability is a serious<br />
deficiency of this approach. One approach toward a scalable solution is to have a tiered<br />
network. Chintalapudi et al. (2006) utilized lower tier nodes and powerful upper tier<br />
nodes. Assuming upper tier nodes have sufficient power, power consumption at lower tier<br />
nodes is moderated. The tiered network approach is applicable only when installation of<br />
powerful nodes and power supply to these nodes are practical.<br />
The second group, on the other hand, assumes that each smart sensor measures and<br />
processes data independently without sharing information among the neighboring nodes<br />
as illustrated in Figure 2.3 (Lynch et al., 2005; Nitta et al., 2005; Sohn et al., 2002).<br />
Because only the data processing outputs are sent back to the base station, the required<br />
communication can be quite small.<br />
Consequently, this approach is scalable to a large number of smart sensors. However,<br />
the independent sensor node approach does not utilize available information from<br />
neighboring nodes; all spatial information is neglected. For example, information<br />
regarding mode shapes cannot be obtained and used in this approach. Information from<br />
25