Adaptive detection of tool-workpiece contact for nanoscale tool setting based on multi-scale decomposition of force signal

Abstract

Due to tool wear under long-time and long-distance cutting, it is inevitable to change the tool to achieve relay machining. However, the existing tool setting methods based on visual inspection and trial cutting have the problems of low precision and complicated process, resulting in low precision and poor consistency of surface quality in relay machining. In order to solve the above problems, this paper proposes a method to detect the contact status between the tool and the workpiece with the discrete wavelet transform (DWT) of force signal to realize the nanoscale tool setting. Firstly, a weighted smoothness evaluation metric is proposed to guide the selection of wavelet basis function and decomposition scale of DWT to achieve noise suppression and impact signal enhancement. Then, the force signal processed by DWT under the optimal parameters is input to density-based spatial clustering of applications with noise (DBSCAN) to detect the contact status between the tool and the workpiece. The parameters selection of DBSCAN is adaptively realized by differential evolution (DE) algorithm so as to improve the accuracy of the contact detection. Finally, ultra-precision tool setting experiments were carried out on the single-point diamond turning tools with a radius of 10um and workpieces with various materials such as brass, copper and red copper. The results show that the wavelet basis function and decomposition scale with the minimum value of the proposed metric are optimal for the task. In analysis of force signals with different signal-to-noise ratios (1 dB, 5 dB, 12.5 dB, 20 dB), the proposed metric all works well. In terms of the contact detection between the tool and workpiece, the proposed method achieves 0 % false positive rate compared to other detection methods, and the corresponding tool setting accuracy is about 3 nm, which is much higher than the traditional tool setting method.