Feature analysis of tool wear states based on best wavelet packet and Hilbert-huang transform

According to the characteristics of the tool wear states, Acoustic Emission (AE) signals of different tool wear states were collected. First of all, AE signals were decomposed by wavelet packet and filtered by selecting appropriate threshold. Second, the filtered signals were reconstructed by wavelet packet. Third, they were analyzed by Hilbert-huang transform (HHT). By comparing the filtered and unfiltered energy figure of all the Intrinsic Mode Function (IMF) components which obtain from empirical mode decomposition (EMD), it shows that filtering can reduce the noise in the low frequency part. After observing HHT three dimensional time-frequency diagram of three signals which tool flank wear vb value are 0.11mm, 0.25mm and 0.35mm, signals in the high frequency part become more and more much along with the increase of vb value; Then, by contrasting marginal spectrum of three signals, result indicates that the distinct fault signal appears around 510KHZ, it proves that HHT can be used for fault diagnosis.