Prediction of the Coefficient of Friction during Toggle Pin Bearing Tests by Training an Artificial Neural Network with Extracted Vibrational Features

碩士 === 崑山科技大學 === 機械工程研究所 === 106 === Plain bearings are mechanical elements commonly found in machines. Its sole purpose is to reduce the friction between a rotating shaft (journal) and a stationary fixture. Plain bearings are particularly important in the plastic injection moulding industry as it...

Full description

Bibliographic Details
Main Authors: Jamin Jamir Escalante, 賈米爾
Other Authors: Huann-Ming Chou
Format: Others
Language:en_US
Published: 2018
Online Access:http://ndltd.ncl.edu.tw/handle/9z47gv
Description
Summary:碩士 === 崑山科技大學 === 機械工程研究所 === 106 === Plain bearings are mechanical elements commonly found in machines. Its sole purpose is to reduce the friction between a rotating shaft (journal) and a stationary fixture. Plain bearings are particularly important in the plastic injection moulding industry as it plays a critical role in facilitating the opening and closing of the mould via the toggle pin. The objective of this research is to use vibrational analysis paired with an artificial neural network to monitor the changes in the coefficient of friction during the toggle pin test. For convenience, the test of the toggle pin bearing was conducted using a bearing test rig. The vibration produced during a 16-hour toggle pin bearing test was collected and analyzed. Valuable vibrational signatures such as Root Mean Squared, Kurtosis, Crest Factor, Frequency Noise Floor and Driving Shaft Frequency Analysis (120 Hz Magnitude) were all extracted for various bearing types and lubricating conditions. These vibrational features were then used as inputs to train an Artificial Neural Network to predict the changes in coefficient of friction. This paper used this approach to predict the coefficient of friction for various toggle pin-lubricant fixtures. It was successful in predicting the coefficient of friction for a given toggle pin-lubricant fixture within a mean error of 1.93%.