Online tool condition monitoring in micromilling using LSTM

High-quality and cost-effective production in micro-milling involves the use of tools of diameter 50–800 μ m, at high rotational speeds, along complex tool paths. These tools are susceptible to high wear and unexpected breakage, and hence a high-precision tool condition monitoring system is required to predict the tool wear states. In this work, we propose a novel approach for high-precision tool condition monitoring in micro-milling using cutting force signals. The method correlates dominant frequency variations with the tool condition along its complete life cycle, considering both straight and circular tool paths to mimic real-life machining scenarios. Therefore, using multiple micro-milling experiments, dominant frequency was characterized using Wavelet transform and Short Time Fourier Transform, and a tool condition prognostic model was developed using LSTM networks. The model accurately predicts force signals with an RMSE less than 0.09, enabling indirect prediction of the tool condition.