Neural Network Application for Chatter Detection
Hae-Deuk Yu, Do-Hun Chin, Jong-Do Kim, Moon-Chul Yoon · Journal of the korean society of manufacturing technology engineers · 2018
The end-milling chatter behavior is very complex and is closely related to a non-periodic dynamic property and the end-milling force; therefore, it is very difficult to detect and diagnose chatter using the end-milling force. This paper presents a novel method for detecting chatter in end milling using neural network, such as Generalized regression neural network (GRNN), Radial basis neural network(RBNN) and perceptron regardless of periodic and non-periodic forces. As a pattern criterion variable for target data, stochastic variance and kurtosis are used for the neural network configuration. By comparing the end-milling force histories with stochastic variables in the fundamental end-milling property, the time domain chatter characteristics are well reviewed, and separated and patterned well for chatter detection. These neural network results using stochastic variables show the reliability of chatter detection; furthermore, it can detect the malfunction property in end-milling and can be applied for determining the existence of chatter.