機械学習の手法を導入した工具カタログからのデータマイニング

抄録

In this research, we try to construct a system to support non - expert 's cutting condition determination using machine learning method. We have constructed a system that derives cutting conditions from the shape parameters of the tool and the hardness of the workpiece by learning based on the recommended cutting conditions described in the tool catalog by the random forest method. In this section, we report on the cutting condition prediction system which can add information on the coating of the tool and can predict more accurately. In this paper, it was found that the elimination of the outlier which was the subject in the construction of the cutting condition prediction system using the random forest can not be improved by expanding the database (by the method of taking in other company's tool catalog data). It was confirmed that the cause of the outlier was not the lack of the number of data points but the lack of information contained in the database, specifically, the lack of information on the coating.