Structural Health Monitoring Using Smart Sensors - ideals ...

Initialization: within local sensor communities
Inject parameters
• NodeID
• Measurement direction
•# of samples
• NExT/ERA parameters
• Modal parameters from the previous
step
Time synchronization
• Send beacon signal
• Estimate local clock offset
• Estimate clock drift
Sensing
• Update local clock offset
• Start sensing
• Copy acquired data to global
variables
• Acquire timestamp of each block of
data
• resample and synchronize the
measured data
Report to the base station
Correlation function estimation
• broadcast data
• cross-spectral density estimation
•ifft
•averaging
yes
Need more
averaging ?
no
Report to the cluster heads and base
station
ERA
Report to the base station
Inject parameters
•NodeID
• Measurement direction
•# of samples
• NExT/ERA parameters
• Modal parameters from the previous
step
• mass normalization constants
Time synchronization
• Send beacon signal
• Estimate local clock offset
• Estimate clock drift
Sensing
• Update local clock offset
• Start sensing
• Copy acquired data to global
variables
• Acquire timestamp of each block of
data
• resample and synchronize the
measured data
Report to the base station
Correlation function estimation
• broadcast data
• cross-spectral density estimation
• ifft
•averaging
yes
Need more
averaging ?
Report to the cluster heads and base
station
ERA/DLV
Monitoring
no
Report to the base station
Figure 7.25. The block diagram of monitoring in a sensor community.
Synchronized sensing is performed in the manner described in section 5.3. The
number of sensing samples is set to be 11,264. The raw data is sent back to the base station
for debugging purposes. This reporting allows numerical operations on Imote2s to be
reproduced on the PC; the validity of data processing on the Imote2 can also be examined.
Centrally collecting all of the data takes a substantial amount of time and can be
eliminated once this DCS system is fully developed.
Correlation functions are then estimated in a distributed manner. The model-based
data aggregation approach explained in section 5.1 is employed. If the number of averages
in correlation function estimation, n d
in Eq. (5.1), is smaller than the predetermined value,
sensing, reporting to the base station, and correlation function estimation are repeated (see
Figures 7.24 and 7.25).
128