Digital Twin Based Tool Wear State Recognition with Domain-Integrated Multi-Residual Network

Cutting tools are vital in industry, but they wear out with use and will affect the quality of the workpiece if not replaced in a timely manner, so determining the tool's wear status is very important to the cutting process. Current wear state identification models rely on large, reliable datasets, which are challenging in the real world. This study presents an approach using digital twins and domain adaptation. The digital twin model generates sufficient simulated data, supported by a domain-integrated multi-residual network (DIMRN), to extract key wear features from real and simulated data. Mapping these features through fully connected layers allows accurate tool wear state identification. The results confirm effective wear state identification and enhance digital decision-making in smart manufacturing.