Vibration Analysis for Turning-Milling Process Condition Monitoring using Short-Time Fourier Transform Enhanced by Empirical Mode Decomposition

Turn-milling process is widely applied in industries that provides advantages for machining large-diameter mechanical parts with high speed, reducing cutting temperature, which in turn decreases tool wear. However, it needs to monitor the turn-milling process for preventing the onset of chatter vibration during operation and the chatter induces negative effects. Vibration analysis is one of the ways for monitoring it. However, acquired vibrations should be denoised from noises, wherein the conventional signal filter may have defiance to do that. This paper presents the utilization of the empirical mode decomposition (EMD) method as an efficient and adaptive noise filter. The Short-Time Fourier Transform (STFT) improvement using EMD is then used for monitoring turn-milling process conditions in the energy-time-frequency domain. The results showed that the reconstructed signal was quite impressive compared to the raw signal and the oscillation of the filtered signal was clearer than the raw signal. The improvement of the filtered signals was proved by the kurtosis index and spectral kurtosis. The improved STFT using EMD showed a significant spectrum with high resolution compared to conventional STFT. The energy density could be observed clearly in the machining characteristic frequencies with an improvement of about 10-100 times larger. The proposed method is therefore effectively applied to monitor the turn-milling condition.