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.
7.1.4 Complex matrix inverse<br />
A complex matrix inverse function is developed based on the Gauss-Jordan matrix<br />
inverse method described in Numerical Recipes in C (Press et al., 1992). The accuracy of<br />
the algorithm is examined by calculating the matrix product of the inverse and original<br />
matrices. The identity matrix was obtained as the output.<br />
The complex matrix inverse algorithm for an n n complex matrix needs<br />
n + n B of memory. n B is first used to store the original matrix and then<br />
replaced with the matrix inverse. n B is internally used. The necessary memory space,<br />
n + n , exceeds 256 kB when n = .<br />
7.1.5 Sort<br />
A quick sort algorithm is developed based on Numerical Recipes in C (Press et al.,<br />
1992). One application example requiring two vectors be sorted simultaneously is the<br />
ordering of natural frequencies in ascending order and the corresponding reordering of<br />
multiple modal parameters. This function is implemented using floating point operations.<br />
The memory size required to run this function is approximately n B, where n is the<br />
length of the vector to be sorted; the size of floating type variable is 4 B. Sorting of more<br />
than 30,000 data points can be achieved with 256 kB memory. This function is not<br />
considered to be expensive in terms of memory usage.<br />
7.2 DCS implementation<br />
Components of DCS (i.e. Sensing, NExT, ERA, DLV methods, and DCS logic) are<br />
implemented on the Imote2s. The outputs of each step of the DCS algorithm on Imote2s is<br />
compared with reference values calculated on the PC using MATLAB. Sensing is<br />
examined by comparing signals from the Imote2s with those from conventional reference<br />
accelerometers. To check the validity of the various DCS components, acceleration data is<br />
injected from the PC, instead of acceleration measured on the Imote2, and used for the<br />
numerical evaluation. The Imote2 is shown to perform DCS calculations as designed.<br />
7.2.1 Sensing<br />
The middleware service developed in section 5.3 is implemented on the Imote2, and<br />
the sensing capability investigated. The sensing capability is first calibrated against<br />
reference sensors. In anticipation of the damage detection experiment with Imote2s in the<br />
next chapter, the Imote2s and reference sensors are placed on the three-dimensional truss<br />
structure, and the measured acceleration signals are examined. Synchronization of the<br />
measured signals is considered by sensing acceleration responses of the truss and<br />
comparing the phase characteristics of the signals.<br />
108