Databases Reference
In-Depth Information
FIGURE 9.9: Conceptual system architecture for the proposed integrated
grid-sensor network framework for the structural engineering life cycle.
9.2.2 Structural Analysis, Design, and Monitoring Applica-
tions
There are three specific applications for the integrated grid-sensor network
framework in a structural engineering life cycle: structural pattern matching
within a WSN, parallel adaptive FEA, and grid-enabled pattern recognition
for the feedback and data integrity analysis. These applications are incorpo-
rated in a single grid workflow, as shown in Figure 9.10.
9.2.2.1
One-Shot Structural Pattern Matching in a Sensor Net-
work
Adaptive FEA [99] provides the means to predict the behavior of a range
of electromechanical and structural systems that are under the influence of
anticipated load conditions accurately. These analyses are heavily relied on
in complex engineering designs. Meaningful information can be derived by
correlating the input patterns, gathered in situ by a WSN, with the patterns
calculated by an FEA using the latent associative memory of the network. Be-
cause the associative memory must be implemented in a network that has very
limited computational resources, the governing algorithm must be modified to
suit the limitations of the operating environment. This would generally entail
the replacement of complex sequential algorithms with parallel/decentralized
algorithms. Moreover, some drop in the accuracy of pattern matching might
occur, which could be offset by further processing at the remote system's end
(WSN base station). To provide an accurate pattern matching scheme and
simultaneously reduce the computational requirements in resource-restricted
networks while provide responses in real time, a GN algorithm is considered.
The GN approach (see Section 3.1) models the parallelism available in nat-
urally occurring associative memory (AM) systems and bypasses the defi-
ciencies present in some contemporary approaches. The GN is implemented
