ConvLSTM deep learning signal prediction for forecasting bending moment for tool condition monitoring

Abstract

This paper presents a new method using a Convolutional Long Short-Term Memory (ConvLSTM) network to predict the future sensor signals for cutting tool bending moment in milling. Tool bending moment is directly correlated to cutting forces and can be used for tool condition monitoring. Bending moment from a sensory tool holder was obtained and transformed into uniform sized time-frequency domain data to be used as frame inputs to the network. Future frames of sensor signals are predicted and validated by comparing the predicted frames with the ground truth. The investigations show the potential of the proposed method for predicting future signals which can be used for tool condition monitoring.