F-1027 ニューラルネットワークを用いた空気圧電磁弁の音響診断(G10-3 ニューラルネット等)(G10 機械力学・計測制御部門一般講演)

概要

論文の詳細を見る
This paper is concerned with a fault diagnosis of an electromagnetic valve using both neural networks and spectrum analysis. Recently, a gas leakege accidents generated on a large scale have been the big social problem. It is very difficult to diagnose a gas leakege, early. In this research, the proposed fault diagnosis system is based on the spectrum of vibrations or sounds obtained from the electromagnetic valve, using neural network, the time series data of vibrations or sounds are complicated and they include noises. The difference between normal data and abnormal one becomes clear by comparing with spectrum data. As a fault diagnosis, there are three cases assummed. Through these results, the effectiveness of proposed a fault diagnasis system is verified.
一般社団法人日本機械学会の論文
2001-08-22