Acoustic emission-based novel identification method for vertical tool center error in single-point diamond turning

During single-point diamond turning (SPDT), tool center errors can cause deviations of the tool position relative to the workpiece center, resulting in tool-above-center error (TACE) or tool-below-center error (TBCE) along the vertical direction. These errors generate residual cylindrical or conical deviations on the machined surface, thereby significantly degrading the surface quality and optical performance. To achieve online and high-precision identification of vertical tool center error (VTCE), this study proposes an online identification method that integrates time–frequency feature analysis with deep learning algorithms, thereby significantly improving the accuracy and efficiency of error identification. First, the acoustic emission (AE) signals acquired during the cutting process were analyzed using wavelet transform. The results indicate that the dominant response frequency band of the AE signal is concentrated within 187.5–218.8 KHz, while a higher proportion of high-frequency energy is observed under TBCE condition. Subsequently, time–frequency image features of the AE signals were constructed based on wavelet transform, and an improved GoogLeNet-AE model was developed to perform classification identification of VTCE, achieving a classification accuracy of 100 %. Furthermore, according to the root mean square (RMS) feature responses of AE signals at the final stage of cutting under different VTCE, convolutional neural network (CNN)-TACE and CNN-TBCE models were established to perform regression prediction for TACE and TBCE, respectively. The prediction errors of both models were maintained within 6 μm. The experimental results demonstrate that the proposed AE signal-based VTCE identification method enables online and accurate identification of VTCE in SPDT, providing an effective approach for intelligent monitoring and error control in ultra-precision machining.