An approach for tool health assessment using the Mahalanobis-Taguchi system based on WPT-AR

A real-time tool health assessment has a significant implication on reliable machining operations. This paper proposes a health assessment method for tools in milling machine using the Mahalanobis-Taguchi system (MTS) based on wavelet packet transformation and autoregression. In this method, the nonlinear and non-stationary vibration signal from the milling process is first decomposed using wavelet packet transforms. Then, an autoregression (AR) model is constructed for each coefficient of the reconstructed signal, and the parameters as well as variance of the remnants of each AR model are employed to form the initial feature matrix. Singular values of this feature matrix are obtained through singular value decomposition, at which point MTS is employed. In this study, MTS provides (1) a computational scheme based on the Mahalanobis distance for obtaining the health index of a tool, and (2) Taguchi methods to extract the key features and reduce the redundant ones. Finally, the performance and effectiveness of the proposed method are validated by vibration signals acquired from the milling machining process.