176
S. Wu et al.
16.3 Results and Discussion
The displacement rate of the chuck was set to 0.3 mm/min in the static load test,
and the MAE signal of the specimen was collected before the stress was applied.
Then MAE signal are collected with excitation frequencies of 10 and 50 Hz, output
sinusoidal voltage amplitude of 0.5–3 V, and magnetic field direction parallel to the
axial direction.
Firstly, this paper extracts one of the most representative characteristic parameters
of the original signal, namely energy. As shown in Fig. 16.4a, when the excitation
voltage is constant, the energy increases with the increase of the excitation frequency.
With the excitation frequency unchanged, the energy increases with the increase of
excitation voltage. According to MAE generation mechanism, excitation voltage
directly affects the degree of magnetic domain structure change of ferromagnetic
metal materials. With the increase of excitation voltage, irreversible movement of
domain wall, generation and annihilation of domain wall and irreversible process of
magnetic domain magnetization vector begin to generate MAE signal, thus making
signal energy gradually increase [18]. As for the excitation frequency, it is related to
the rate of change of the magnetic domain structure of ferromagnetic metal material.
As the excitation frequency increases gradually, the energy increase rate of the MAE
signal increases relatively. Although the increase of energy increase rate is beneficial
to data analysis, the excitation frequency increases and the bimodal characteristics
of the MAE signal are weakened to a certain extent, which is not conducive to data
analysis. Therefore, when selecting the excitation frequency, the SNR and the shape
characteristics of the signal should be considered at the same time.
In addition to energy as a characteristic parameter, Fig. 16.5b, c show the comparison of amplitude and RMS voltage with static tensile force. However, it is observed
that the relationship between energy and static tensile force is not as obvious as that
in Fig. 16.5a. Since the MAE signal contains a complex noise signal, in order to
accurately extract the characteristic parameters, CEEMD algorithm will be used to
denoise the MAE signal.
It can be seen from Fig. 16.6 that it is not ideal to extract the characteristic parameters of the original signal directly. Therefore, the CEEMD algorithm needs to be
introduced to effectively improve the SNR of the signal while maintaining the characteristics of the signal. When using the CEEMD algorithm to process the MAE
signal, firstly the signal needs to be decomposed into IMF components. Secondly,
the correlation coefficients of each IMF component and the original signal are calculated, and the reconstruction signal is composed of two or three IMF components
whose correlation coefficient cumulative value is close to 1. Finally, the characteristic
parameters are extracted from the reconstructed signal.
It is found that the IMF1 and IMF4 components need to be accumulated in the
static load experiment, and the obtained results are suitable for the extraction of
characteristic parameters in the next step. Since it is easier to determine the threshold
voltage value than the original signal, the characteristic parameters of the MAE signal
can be accurately extracted. Therefore, using CEEMD algorithm to process the MAE
S. Wu et al.
16.3 Results and Discussion
The displacement rate of the chuck was set to 0.3 mm/min in the static load test,
and the MAE signal of the specimen was collected before the stress was applied.
Then MAE signal are collected with excitation frequencies of 10 and 50 Hz, output
sinusoidal voltage amplitude of 0.5–3 V, and magnetic field direction parallel to the
axial direction.
Firstly, this paper extracts one of the most representative characteristic parameters
of the original signal, namely energy. As shown in Fig. 16.4a, when the excitation
voltage is constant, the energy increases with the increase of the excitation frequency.
With the excitation frequency unchanged, the energy increases with the increase of
excitation voltage. According to MAE generation mechanism, excitation voltage
directly affects the degree of magnetic domain structure change of ferromagnetic
metal materials. With the increase of excitation voltage, irreversible movement of
domain wall, generation and annihilation of domain wall and irreversible process of
magnetic domain magnetization vector begin to generate MAE signal, thus making
signal energy gradually increase [18]. As for the excitation frequency, it is related to
the rate of change of the magnetic domain structure of ferromagnetic metal material.
As the excitation frequency increases gradually, the energy increase rate of the MAE
signal increases relatively. Although the increase of energy increase rate is beneficial
to data analysis, the excitation frequency increases and the bimodal characteristics
of the MAE signal are weakened to a certain extent, which is not conducive to data
analysis. Therefore, when selecting the excitation frequency, the SNR and the shape
characteristics of the signal should be considered at the same time.
In addition to energy as a characteristic parameter, Fig. 16.5b, c show the comparison of amplitude and RMS voltage with static tensile force. However, it is observed
that the relationship between energy and static tensile force is not as obvious as that
in Fig. 16.5a. Since the MAE signal contains a complex noise signal, in order to
accurately extract the characteristic parameters, CEEMD algorithm will be used to
denoise the MAE signal.
It can be seen from Fig. 16.6 that it is not ideal to extract the characteristic parameters of the original signal directly. Therefore, the CEEMD algorithm needs to be
introduced to effectively improve the SNR of the signal while maintaining the characteristics of the signal. When using the CEEMD algorithm to process the MAE
signal, firstly the signal needs to be decomposed into IMF components. Secondly,
the correlation coefficients of each IMF component and the original signal are calculated, and the reconstruction signal is composed of two or three IMF components
whose correlation coefficient cumulative value is close to 1. Finally, the characteristic
parameters are extracted from the reconstructed signal.
It is found that the IMF1 and IMF4 components need to be accumulated in the
static load experiment, and the obtained results are suitable for the extraction of
characteristic parameters in the next step. Since it is easier to determine the threshold
voltage value than the original signal, the characteristic parameters of the MAE signal
can be accurately extracted. Therefore, using CEEMD algorithm to process the MAE