Host: The Japan Society of Mechanical Engineers
Name : [in Japanese]
Date : September 05, 2021 - September 08, 2021
Anomaly detection for predictive maintenance of cutting tools is one of the challenging problems in shop floor and tools management. A modern machine learning approach, including deep learning, has been widely studied for the last decade. This study focuses on the multimodality of various cutting time-series data for extracting features of cutting tool status and proposes a multimodal variational autoencoder (MVAE) method. As an extension of our previous work, we newly consider a time series of vibrational acceleration of a cutting tool and a main spindle motor load in addition to a cutting temperature near the cutting edge. MVAE learns a so-called generative model, which is implicit but stochastic, capable of reproducing original time series data. Euclidean distance is employed to evaluate the normality of a given cutting status on the latent space acquired by MVAE. We demonstrate the applicability of the proposed MVAE method in anomaly detection for the endmill cutting tools by comparing it with conventional machine learning methods such as Autoencoder.