关键词: 磁记忆信号, 小波变换, 能量, 奇异性指数, 梯度峰值, 特征提取

Abstract: In order to identify the way how the magnetic memory signals of ferromagnetic materials change during the fatigue process and to make quantitative assessment of fatigue damage,an axial tensile fatigue test with holed-cracked Q235 plate specimen as objects is designed.It goes like this:The magnetic memory signals can be collected steadily with the aid of three dimensional motion stage,and the noises accompanying are reduced by a multi-scale analysis tool of wavelet transform signal.Then such magnetic signals are decomposed and reconstructed with wavelet packet to extract the wavelet packet energy,singularity indexes,maximum gradient values Extraction and characteristics.Therefore,how the relative energy distributes in bands can be detected and how the singularity indexes change is also discovered.The obtained wavelet packet energy,singularity indexes and signal gradient value are combined as multi-characteristic vectors,serving to assess fatigue damage.The results indicate that the wavelet packet energy,singularity indexes and maximum gradient values Extraction change obviously amid the fatigue damage process.Such energy is ever increasing in the lower frequency band and decreasing in the higher.Total energy distribution starts shifting to focus on the lower one.Meanwhile,singularity indexes begin to decrease,yet the gradient peak values in different stages increase.With these multi-characteristic vectors that can make up the inherent deficiency of single characteristic adopted previously,the result promises to provide technique support for evaluating the fatigue damage of metal components.

Key words: magnetic signals, wavelet transform, energy, singularity indexes, maximum gradient values, feature extraction