Real-time chatter detection based on fast recursive variational mode decomposition

Real-time chatter detection is crucial for avoiding damage to machine tools and workpieces. Time–frequency analysis methods have been extensively adopted in the feature extraction of chatter detection, especially the variational mode decomposition (VMD) has a solid theoretical basis and outstanding decomposition performance. Nevertheless, the performance of VMD is highly dependent on the decomposition parameters (number of modes and penalty factor). In this article, a novel real-time chatter detection method based on fast recursive variational mode decomposition (FRVMD) is proposed. Unlike the optimization algorithm-based VMD, which is time-consuming, the proposed FRVMD extracts the modes one by one in a recursive framework and adaptively adjusts the penalty factor according to the iterative information. FRVMD exhibits higher computational efficiency and better decomposition performance, which is suitable for real-time chatter detection. After the adaptive extraction of chatter-sensitive component by FRVMD, two chatter indicators, namely energy ratio (ER) and dispersion entropy (DE), are introduced to characterize the machining state. Finally, a chatter identification model is established by utilizing the support vector machine (SVM). The simulation and experimental findings verify the effectiveness of the proposed method.