Digital twin-based dynamic prediction of thermomechanical coupling for skiving process

Skiving has potential for gear machining, but its cutting force fluctuates greatly, its cutting temperature is high during processing, and it has the characteristics of thermomechanical coupling. It is difficult for traditional methods to dynamically predict the thermomechanical coupling during the skiving process, whose efficiency and stability cannot be guaranteed. Aiming to solve these problems, a digital twin (DT)-based dynamic method is proposed to predict thermomechanical coupling in the skiving process. Considering the time-varying characteristics and coupling of the cutting force and temperature, a multi-physical modeling method dual-driven by mechanism and data is proposed to establish a thermomechanical coupling DT (TMDT) model of the skiving process. The dynamic consistency of the skiving process between the digital and physical spaces is realized. Principal component analysis (PCA) and an extreme learning machine (ELM) are used to reduce the order of the TMDT, the reduced-order model is trained using the skiving big data, and a relationship mapping model of the cutting parameters and the cutting force and temperature is established to realize the dynamic prediction of the cutting force and temperature. The effectiveness of the proposed method is verified through gear skiving experiments. This research has important theoretical guiding significance to realize efficient and stable skiving processing.