Sensitivity Analysis of Cutting Force on Milling Process using Factorial Experimental Planning and Artificial Neural Networks

This paper aim to investigate and to compare the capabilities of the Artificial Neural Networks (ANN) and the Factorial Experimental Planning (FEP) to measure the most significant variables on milling process that influence the cutting force. The force data were acquired by an experimental apparatus...

Full description

Saved in:
Bibliographic Details
Published in:Revista IEEE América Latina 2016-12, Vol.14 (12), p.4811-4820
Main Authors: do Nascimento, Eriberto Oliveira, de Oliveira, Lucas Nonato
Format: Article
Language:eng ; por
Subjects:
Adaptation models
Algorithms
Artificial Neural Network
Artificial neural networks
Computational modeling
Confidence intervals
Cutting force
Cutting parameters
Cutting speed
Data acquisition
Factorial Planning
Force
IEEE transactions
Milling
Milling Force
Multilayer perceptrons
Neural networks
Planning
Profile method (forecasting)
Regression
Sensitivity analysis
Statistical analysis
Statistical inference
Citations: Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This paper aim to investigate and to compare the capabilities of the Artificial Neural Networks (ANN) and the Factorial Experimental Planning (FEP) to measure the most significant variables on milling process that influence the cutting force. The force data were acquired by an experimental apparatus and the statistical inference of the force were set by the Root Mean Squared value. The FEP used the multiple linear regression technics to evaluate the variable significance and then get a model that could predict new responses over designed experiment. On the ANN were applied the Profile Method on a supervised Multilayer Perceptron optimized with the Levenberg-Marquardt algorithm. The results showed good agreement with a confidence level of 90%, that the axial depth of cut, feed per tooth and cutting speed in this order were the most significant variables. This is in accordance with the literature and can open more applications using the ANN approach to obtain a variable significance over other systems.
ISSN:1548-0992
1548-0992
DOI:10.1109/TLA.2016.7817015