適応フィルタとニューラルネットワークを用いた切削異常状態の検知

概要

論文の詳細を見る
The present paper deals with a method of detecting tool flank wear, tool fracture and chatter vibration, continuously in process of a cutting operation. A detecting system made by combining an adaptive filter and a neural network is proposed and is verified its effectiveness by using acceleration signals recorded during turning operations. An autoregressive model is fitted to an acceleration signal by applying an adaptive filter and then the autoregressive parameters are recognized by using a neural network. An autoregressive model in front of the neural network makes the neural network compact and a learning convergence faster. Acceleration signals are recorded during turning of medium carbon steel (JIS : S45C) with a piezoelectric sensor mounted at the top of a tool holder. As results of simulations on a computer using the actual acceleration signals, which correspond to tool flank wear, tool fracture, chatter vibration, useful detections of abnormal conditions are achieved and an effectiveness of the method is confirmed.
公益社団法人精密工学会の論文
1991-10-05