Tool Condition Diagnostics Flanked by Force and Acceleration Correlation in Machining

Monitoring tool wear is critical for ensuring productivity and reliability in automated CNC machining. This study presents an intelligent diagnostic system for continuous tool wear monitoring during the dry milling of SCM440 alloy steel. Utilizing a multi-sensor fusion system consisting of a wireless sensory tool holder and a dynamometer, the research introduces a signal feature: the ratio between the dynamic cutting force at the tooth-passing frequency (160 Hz) and the mean cutting load. To ensure data reliability, a coherence function was implemented to assess the linear relationship between forces and vibrations, which had a median of 0.92 at the tooth-passing frequency along the feature axis. A predictive model was developed using a cubic polynomial to link the extracted feature to measured flank wear. By constraining the model’s leading coefficient to align with the physical progression of tool wear, the system achieved a root mean square error (RMSE) of 0.153 mm on held-out data (0.115 mm excluding the first three measurements of each run, attributed to early process instability). Error showed no significant trend with wear level (p = 0.099), indicating stable performance across the measured range. The results demonstrate that physically constrained mathematical models, combined with multi-sensor integration and targeted feature extraction, provide a promising framework for autonomous tool health monitoring, validated in this case for dry milling SCM440 steel under a limited range of cutting conditions. Further studies including broader validation across materials and process parameters would be needed before generalizing to Industry 4.0 environments.