8 Acoustic Emission and Dual-Tree Complex Wavelet Transform …
87
Table 8.1 Specification of
sensors
Sensor model
SR 40 M
Dimension OD × H (mm)
22 × 36.8
Operating Frequency range (kHz)
15–70
Resonant frequency (kHz)
40
Peak sensitivity
>75 dB
collected by LabView acquisition system and amplified using pre-amplifiers with
40 dB gain value. 250 sets signals with 27.5 ms length were acquired at 1 MHz
sampling frequency and the operating water pressure is 2 bar. The specifications of
AE sensors stated in Table 8.1.
8.4 Results and Discussion
Five levels of DTCWT decomposition is sufficient in the work based on aforesaid
[2] since the minimum interested frequency is 15 kHz. Signals which those are lower
than 15 kHz are unreliable and had been filtered out because the minimum operating
frequency of the sensor is 15 kHz. Figure 8.9 shows the decomposed wavelet coefficients before and after de-noised by the proposed method. The figure clearly shows
that DTCWT itself is not sufficient to eliminate the noise but the proposed method can
give a smooth signal. The de-noised signal has sharpened feature with more recognized peaks. Meanwhile, comparison between ordinary wavelet de-noising and the
proposed method is shown in Fig. 8.10. It is shown that DTCWT performs better in
Fig. 8.9 Comparison of DTCWT and DTCWT-Soft threshold
87
Table 8.1 Specification of
sensors
Sensor model
SR 40 M
Dimension OD × H (mm)
22 × 36.8
Operating Frequency range (kHz)
15–70
Resonant frequency (kHz)
40
Peak sensitivity
>75 dB
collected by LabView acquisition system and amplified using pre-amplifiers with
40 dB gain value. 250 sets signals with 27.5 ms length were acquired at 1 MHz
sampling frequency and the operating water pressure is 2 bar. The specifications of
AE sensors stated in Table 8.1.
8.4 Results and Discussion
Five levels of DTCWT decomposition is sufficient in the work based on aforesaid
[2] since the minimum interested frequency is 15 kHz. Signals which those are lower
than 15 kHz are unreliable and had been filtered out because the minimum operating
frequency of the sensor is 15 kHz. Figure 8.9 shows the decomposed wavelet coefficients before and after de-noised by the proposed method. The figure clearly shows
that DTCWT itself is not sufficient to eliminate the noise but the proposed method can
give a smooth signal. The de-noised signal has sharpened feature with more recognized peaks. Meanwhile, comparison between ordinary wavelet de-noising and the
proposed method is shown in Fig. 8.10. It is shown that DTCWT performs better in
Fig. 8.9 Comparison of DTCWT and DTCWT-Soft threshold