STEP-compliant CNC system featuring real-time material removal simulation for online tool wear monitoring

With the advancement of intelligent manufacturing, CNC systems are increasingly expected to achieve higher levels of autonomous perception and decision-making. However, conventional CNC systems relying only on sensors have limited online analysis capability in complex machining. STEP-CNC, equipped with a simulation kernel, provides rich semantic information and enables multi-domain data fusion of “simulation + sensing”, offering a novel framework for online process analysis. Taking tool wear as a case study, this paper proposes a comprehensive online monitoring solution integrated into STEP-CNC. First, the geometric simulation is executed to calculate the Cutter Workpiece Engagement (CWE) and quantify instantaneous material removal, effectively characterizing interactions with workpiece. Second, a sparse stacked autoencoder extracts and compresses informative features from multi-sensor signals, yielding compact representations correlated with wear value. Third, an incremental prediction model tailored for online applications is developed, fusing geometric, physical, and process-domain inputs to provide precise wear increment estimates over fixed time windows. Finally, the prediction model is encapsulated as a service and integrated within the STEP-CNC framework, enabling tool wear monitoring with online feedback to the CNC controller. Experimental results demonstrate that the proposed method can accurately track tool wear progression, achieving an online monitoring accuracy exceeding 89%. The monitored wear values can further assist machining decision-making, preventing tool failures and ensuring workpiece quality. The proposed method may also serve as an actionable reference for using STEP-CNC with multi-domain data in intelligent manufacturing applications.