Deep Learning Based Tool Wear Estimation Considering Cutting Conditions
Zongshuo Li, Markus Meurer, Thomas Bergs
TL;DR
The study addresses the challenge of estimating tool wear under varying cutting conditions with the need for zero-shot transfer in industrial settings. It introduces a CNN that conditions on cutting parameters by fusing a parameter sequence with multi-sensor time-series inputs, aiming to eliminate the need for retraining when parameters change. Results show substantial improvements over a baseline model that omits cutting conditions, with RMSE reductions of $22.3\%$ (fixed parameters) and $11.4\%$ (variable parameters) and $R^2$ gains of $3.8\%$ and $4.1\%$, respectively, demonstrating the feasibility of zero-shot transfer in practice, though rapid wear phases can still yield larger errors. The approach has practical impact for industrial milling, enabling direct deployment of wear-monitoring systems across diverse cutting settings using limited training data.
Abstract
Tool wear conditions impact the final quality of the workpiece. In this study, we propose a deep learning approach based on a convolutional neural network that incorporates cutting conditions as extra model inputs, aiming to improve tool wear estimation accuracy and fulfill industrial demands for zero-shot transferability. Through a series of milling experiments under various cutting parameters, we evaluate the model's performance in terms of tool wear estimation accuracy and its transferability to new fixed or variable cutting parameters. The results consistently highlight our approach's advantage over conventional models that omit cutting conditions, maintaining superior performance irrespective of the stability of the wear development or the limitation of the training dataset. This finding underscores its potential applicability in industrial scenarios.