Tool wear feature extraction in BTA deep hole drilling process based on maximum probability multi-synchrosqueezing transform of spindle current signal

Abstract

Multi-synchrosqueezing transform (MSST) enhances the concentration of the time–frequency representation (TFR) of time-varying signals by introducing multi-instantaneous frequency (IF) estimation operator. However, the existence of estimation bias may lead to uncertainty in the discrete values of IF during computer computations, causing issues with non-reassigned points. In this paper, a maximum probability multi-synchrosqueezing transform (MPMSST) is proposed, which incorporates distribution probability during the discretization process of the IF, taking the frequency corresponding to the maximum distribution probability as the determined value of IF. Through verification with multi-component time-varying simulation signals, MPMSST not only effectively eliminates non-reassigned points but also accurately estimates the theoretical time–frequency curve. By applying MPMSST to analyze spindle motor current signals during deep hole machining, TFRs under different tool wear states are obtained. The singular value decomposition (SVD) method is utilized to extract the principal component of the TFR, and a tool wear state feature index is constructed based on the sum of squares of the first 10 singular values. The correlation coefficient and covariance between the proposed feature index and the tool wear curve are 0.835 and 5.704 respectively, indicating a good consistency of the index with tool wear changes, and a high sensitivity to variations in tool wear state.