Abstract
Knowing machine tool dynamics and cutting force is critical to machining process optimization, tool life prediction, and in-process monitoring. Identifying system dynamics and estimating dynamic cutting forces often require dedicated procedures and extensive execution effort. This article presents a Fourier neural operator (FNO)-inspired end-to-end architecture for the rapid estimation of dynamic forces by inferring system characteristics through interpretable operator learning in the frequency domain using system excitation and response data. This machine learning method learns to approximate the system frequency response function (FRF) as an intermediate step and subsequently produces a functional mapping from acceleration to dynamic force. For validation, both a numerical study with a theoretical two degrees-of-freedom model and a field experiment on an actual machine tool are conducted. Results demonstrate that this FNO-based method predicts dynamic forces with over 90% accuracy in terms of R2 value for both validation cases, with the approximated FRF offering insights into the underlying machine tool dynamic behavior. Model training considerations, limitations, and practicality of this method, including the approximate nature of the inferred system characteristics, are also discussed in this article.