PSI - Issue 5

K.S.C. Kuang et al. / Procedia Structural Integrity 5 (2017) 1168–1175 Dong Yang et al. / Structural Integrity Procedia 00 (2017) 000 – 000

1172

5

Suppose the original optical signal recorded using oscilloscope by   x t , which is recorded with suitable sampling frequency. Then, the root mean square (RMS) sequence was calculated from raw signal of POFs sensor over a moving window. The resulting signal can be expressed by

1 k

=1, 2, , N i k

  

  





 

2

2

2

R i

x

x

x

(3)





 

* i w

* i w

* i w k 

1

2





where i is the number of the moving window, N is the length of the original signal,  is the length of the moving window. After that, the pre-processed RMS sequence   R i were then ready to be analysed. The obtained RMS sequence was further processed by passing it through a Hilbert transform (HT)-based filter (Chen and Wang, 2012, Feldman, 2011). As a result, the noise and other usefulness components in the signal will be removed, and the recovered bandpass sequence and envelop are ready to be analysed further. The smoothed envelop and high frequency sequence recovered using Hilbert transform-based filter are shown in Figure 4 (a) and (b) respectively. In Figure 4 (a), the continuous red line along with the increasing RMS means the crack propagates gradually. On the other hand, the initiation of crack was not expected to result in significant detectable changes in the time-domain POFs sensor signal. The proposed signal processing is, however, capable of identifying the small signal discontinuities in the signal, which were caused by the initiation or propagation of crack in the specimen. When the plot begins to show increasing amplitude of local RMS value (at approximately 4000 s ), the initiation of crack could be inferred. During 4000 s to 8000 s , the crack propagated stable. The amplitude increases dramatically at 8000 s , corresponding to further fast crack propagation. The slight change in the amplitude of the curve at time 4000 s in the plot further corroborates the method highlighting the potential of the IPOFs sensor in detecting anomaly related to crack initiation of the specimen under fatigue load. 3.2. Crack initiation and propagation identification

(a)