Lin, Shang-YuShang-YuLinChen, Po-HanPo-HanChenHung, Pang-HsiangPang-HsiangHungHu, Yuh‑ChungYuh‑ChungHuWEI-CHANG LIPEI-ZEN CHANG2025-12-182025-12-18202502683768https://www.scopus.com/record/display.uri?eid=2-s2.0-105021817517&origin=resultslisthttps://scholars.lib.ntu.edu.tw/handle/123456789/734765Accurate prediction of tool wear is essential for ensuring machining stability, product quality, and process efficiency in high-speed milling. However, traditional time-based wear models require extensive full-life testing under each parameter set, making them impractical for modern multi-condition machining. This study develops an enhanced and generalizable tool-wear modeling framework based on Differential Wear Signature (DWS) analysis, which formulates the wear rate as a function of cutting parameters and instantaneous wear state. By leveraging the tool runout effect, the model captures multiple effective feed rates within a single experiment, thereby improving data efficiency and reducing experimental effort. The enhanced DWS framework incorporates a broader range of cutting parameters, integrates Remaining Useful Life (RUL) estimation, and is benchmarked against conventional offline models including empirical regression, GPR, and ANN-based approaches. Experimental validation confirms that the proposed model accurately reconstructs wear trajectories and predicts RUL under both single- and multi-step cutting scenarios, achieving a root-mean-square error of 0.008 mm, compared to 0.028 mm in the prior DWS model. The results demonstrate that the proposed framework provides a robust, data-efficient, and transferable solution for offline tool-life prediction and process planning in CNC milling.falseCNC milling processDifferential wear signature analysisRemaining useful life estimationTool wear modelingTool wear rate[SDGs]SDG12journal article10.1007/s00170-025-16902-z2-s2.0-105021817517