Thermographic Detection of Tool Wear: A Review on Deep Learning-Based Approaches

Abstract

In machining processes, cutting tool wear is a critical problem affecting key performance parameters such as surface quality, production time, and cost. High temperatures cause deformation and wear in the tool material and shorten tool life. This reduces production efficiency and negatively affects the sustainability of quality-oriented goals such as zerodefect manufacturing. In this review study, the current literature on the prediction of cutting tool wear in machining processes using thermal imaging and deep learning techniques is analyzed. With advancing manufacturing technologies, unattended machining capability and the proliferation of approaches such as Zero Defect Manufacturing (ZDM), realtime and accurate monitoring of cutting tool conditions is a strategic requirement for production efficiency and quality control. In this context, the possibilities offered by indirect wear detection methods based on temperature data and the applications of deep learning-based models in this field are evaluated. Significant success has been achieved in tool life prediction by processing temperature maps obtained with thermal cameras, thermocouples, and infrared (IR) imaging systems using algorithms such as CNN and ANN. The methods in the literature are discussed comparatively and their advantages and limitations are discussed. This paper systematically reviews the existing approaches in the literature for tool wear monitoring using temperature-based methods and provides a comprehensive review for future research in this area. © 2025 IEEE.