Chatter detection based on probability distribution of wavelet modulus maxima

This paper presents a statistical chatter detection method. The methodology is based on the study of discrete wavelet transform (DWT) scheme and statistical analysis of wavelet transform modulus maxima (WTMM). Wavelet transform modulus maxima is used to describe any point where wavelet transform of a signal is locally maximal at corresponding time location. Meanwhile, due to the noisy machining environment, a wavelet-based de-noising method including a hybrid thresholding function and a level-dependent universal threshold rule is proposed. A non-dimensional chatter index varying between 0 and 1 is designed based on the statistical analysis of the WTMM. The main advantages of the proposed chatter index include that: (a) its variation range is independent of process parameters and machining systems, and (b) its threshold value is much less susceptible to cutting condition changes since its value is in relative term. As a result, the chatter index could be used for different machining processes without the time-consuming recalibration process.