Tool Health Classification in Metallic Milling Process Using Acoustic Emission and Long Short-Term Memory Networks: A Deep Learning Approach

Abstract

The manufacturing industry has experienced remarkable progress as a result of integrating automated and intelligent production processes fueled by technological innovation, leading to substantial advancements. These advanced processes involve the use of flexible and high-performance machines, tackling complex and sophisticated processing problems with ease. However, the processing performance can deteriorate due to tool damage or malfunction, which can lead to the discarding of workpieces. Hence, it carries immense importance to have a close look over the condition of the tool throughout the processing to proactively address any potential issues and minimize the possibility of significant tool failures, particularly when manufacturing intricate and costly machine components. Many researchers have looked at the use of machine learning and deep learning approaches for monitoring tool condition. In this study, we are incorporating time series sequential data for which LSTM is the best opted technique. It further explains, how deep learning using Long Short-Term Memory Networks (LSTM) and the acoustic data acquired through a microphone during the metallic milling process has a potential and achieved impressive results. In our study, the accuracy of the model was assessed for different workpiece materials, including Aluminum, Mild Steel, and Brass, and demonstrated the model’s ability to make highly accurate predictions. Specifically, the model achieved an average test accuracy of 99.03% for Aluminum workpieces, while achieving very good test accuracies of 97.16% and 97.83% for Mild Steel and Brass workpieces, respectively. These results were benchmarked against previous work in the same domain, confirming the efficacy of the model.

Keywords

• Long short-term memory networks (LSTM)
• model test accuracy
• tool condition monitoring