Features extraction of tool cutting AE signals based on Hilbert-Huang transformation

This dissertation develops a new approach of feature extraction for tool wear based on Acoustic emission, an application of Hilbert-Huang transform method to feature extract is presented. Using the empirical mode decomposition of the original times series data, the Hilbert transform is applied to each intrinsic mode function. Then our emphasis is how to calculate the Hilbert spectrum and the Hilbert marginal spectrum, and analyze the spectrum containing the characteristics of tool cutting signals. This new method provides not only a more precise definition of particular events in time-frequency space, but also more physically meaningful characteristics of the tool cutting processes. The Hilbert amplitude spectrum and marginal spectrum based on HHT can be as the features, which can be used for pattern recognition to monitor cutting processing. Such as feature extraction, the tool wear condition can be realized for identifying various tool failure states in turning operations.