Learning digital twin: a case study on chatter suppression based on a time-varying stability lobe diagram

Abstract Purpose: Chatter is a frequent problem during machining operations. It significantly influences the machining quality. A stability lobe diagram (SLD) can be utilized to avoid chatter. However, the generated SLD should be frequently modified because it is influenced by different time-varying factors, e.g., wear. In this work, we propose a new method for the generation of SLD considering its variation process. Methods: The SLD is generated by a learning digital twin (LDT). We first propose the structure of the LDT. Two algorithms based on concept drift are proposed to detect the drift of the physical machine. A new hybrid model structure is proposed and used to simulate the behavior of the physical machine. The modified SLD is then generated by the evaluation of the simulated behavior of the hybrid model. Results: The results show that the change in SLD can be correctly detected. The modification of the SLD also receives high accuracy. Conclusion: We conclude that the LDT can be utilized for the generation of SLD and remains high accuracy in the life time of the physical machine.