Host: The Japan Society of Mechanical Engineers
Name : [in Japanese]
Date : March 07, 2022 - March 08, 2022
Tool changes in the cutting process are mainly constant changes based on the results of experiments conducted in advance. However, due to the variation in the wear progress of different tools, tools that can be used sufficiently are resharpened or replaced, resulting in waste. Therefore, monitoring during the cutting process has been actively researched for the purpose of avoiding wasteful tool changes. In this study, the relationship between sound signals generated during drilling of SUS630 and tool damage was experimentally investigated in order to construct a tool damage detection system using cutting sounds. In the experiment, the cutting noise during drilling was captured by a camera installed inside the machine tool. Data processing such as Fast Fourier Transform was performed on the obtained cutting sound data, and the correlation between the characteristics of the cutting sound and tool damage was investigated. As a result, it was confirmed that the component around 1000[Hz] increased as the wear progressed, while the component around 1750[Hz] was dominant for new tools. In addition, machine learning was conducted using the cutting sounds of the new and worn tools as supervisory data, and it was examined whether the damage state of the drill could be determined.
