Metalworking Fluid Classification Based on Acoustic Emission Signals and Convolutional Neural Network



Published Jun 29, 2021
Xiao Wei Anna Lena Demmerling Dirk Söffker


Acoustic emission (AE) which describes the transient stress
waves generated by the rapid release of energy from solid
sources has been widely used in nondestructive testing
(NDT) of materials and structures especially in health
monitoring. As a class of deep neural networks,
convolutional neural network (CNN) has applications in
many fields. Several investigations have been conducted on
the application of CNN in feature learning and fault
diagnosis and prognosis. Metalworking fluids (MWF) play a
significant role in manufacturing processes. By reducing
friction between tool and workpiece, the heat generation in
metalworking process is affected. Thread forming is a
transformative manufacturing process for generating threads
in ductile materials. As the thread geometry is manufactured
by cold forming of the material, lubricating properties of the
MWF strongly effect tool wear and workpiece quality. Up
to now, there are only a few papers on MWF classification
using the process variables like torque or released AE. In
this contribution, a novel approach combining AE signals
and CNN is raised for MWF classification. A tribometer is
used to carry out thread forming trials under well-controlled
experimental conditions. AE measurements are conducted in
context of thread forming. The AE signals are divided into
suitable samples and CNN is applied as classifier. The
results of MWF classification show that the new approach
could distinguish different types of MWF.

How to Cite

Wei, X., Demmerling, A. L. ., & Söffker, D. . (2021). Metalworking Fluid Classification Based on Acoustic Emission Signals and Convolutional Neural Network. PHM Society European Conference, 6(1), 6.
Abstract 15 | PDF Downloads 33



metalworking fluid, acoustic emission, convolutional neural network, classification

Technical Papers