Professional Documents
Culture Documents
ML Poster
ML Poster
net/publication/337822022
Machine Learning Approach for Predicting Angle of Repose in Powder Spreading Process
CITATIONS READS
0 563
3 authors, including:
All content following this page was uploaded by Kuldeep Mandloi on 07 December 2019.
Introduction
o Selective laser melting (SLM) process is one of the most
Widely Used Additive Manufacturing (AM) Process.
o Powder spreading is done to feed into the bed. High power
laser beam is used to melt and fuse the selected area of
powder materials to manufacture desired part.
o Conducting experiments to understand the effect of process
parameter is costly. Hence, in this work, the machine
learning-based predictive models, Artificial neural network
(ANN) is used for the same.
o The neural network modeling was performed using Python
the training algorithm. The optimal neural network
architecture along with the activation function that produced
Selective laser melting- Process setup Angle of repose
the least error in prediction was identified.
Methodology
Data collection for building Neural network is done by varying the following
input parameters. A total of 36 simulations are run to get angle of repose as output
for each. 80% of data is used for training and 20% for testing.
Result
o Number of hidden layers are varied from 2 to 3 and number of
neurons are varied from 2 to 15. The architecture with 3-9-7-1
gives the best result.
o ANN model is trained to predict the angle of response within a
fair accuracy.
o Relu activation function gives the least error.
REFERENCES
1. Körner, Carolin, Elham Attar, and Peter Heinl. "Mesoscopic simulation of selective beam
melting processes." Journal of Materials Processing Technology 211.6 (2011): 978-987.
2. EDEM 2.6 Theory Reference Guide
3. Pontes, Fabricio J., et al. "Artificial neural networks for machining processes surface
roughness modeling." The International Journal of Advanced Manufacturing
Technology 49.9-12 (2010): 879-902.
4. Mekarthy, Sai Manish Reddy. A Combined Finite Element and Machine Learning
Approach for Predicting Specific Cutting Forces and Maximum Tool Temperatures during
Orthogonal Machining. Diss. The University of North Carolina at Charlotte, 2019.