Pattern Classification of Spectra of Vibration Tests in a Wrist by Motor-Operated Electric Tolls Using SOM

Abstract

In this paper, we propose a pattern classification method of spectra of vibration data in a wrist by a motor-operated electric tool (MOET) using Self-Organizing Feature Map (SOM). We collect 27 data to evaluate the vibration patterns in a wrist for three kinds of MOETs. We compute spectrra of those vibration data and input them to a SOM network to find the spectral patterns. Based on the relations among vibration spectral patterns of MOETs under no load and load conditions, we show that the performance under load conditions can be evaluated by those of unloaded ones. © 2004, The Institute of Electrical Engineers of Japan. All rights reserved.