Prediction of field-dependent rheological properties of magnetorheological grease using extreme learning machine method

Magnetorheological grease is seen as a promising material for replacing the magnetorheological fluid owing to its higher stability and the lesser production of leakage. As such, it is important that the rheological properties of the magnetorheological grease as a function of a composition are conduc...

Full description

Saved in:

Bibliographic Details
Published in:	Journal of intelligent material systems and structures Vol. 30; no. 11; pp. 1727 - 1742
Main Authors:	Bahiuddin, Irfan, Wahab, Nurul AA, Shapiai, Mohd I, Mazlan, Saiful A, Mohamad, Norzilawati, Imaduddin, Fitrian, Ubaidillah
Format:	Journal Article
Language:	English
Published:	London, England SAGE Publications 01-07-2019
Subjects:	rheology machine learning composition extreme learning machine Magnetorheological grease neural networks
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!

Description
Summary:	Magnetorheological grease is seen as a promising material for replacing the magnetorheological fluid owing to its higher stability and the lesser production of leakage. As such, it is important that the rheological properties of the magnetorheological grease as a function of a composition are conducted in the modeling studies of a magnetorheological grease model so that its optimum properties, as well as the time and cost reduction in the development process, can be achieved. Therefore, this article had proposed a machine learning method–based simulation model via the extreme learning machine and backpropagation artificial neural network methods for characterizing and predicting the relationship of the magnetorheological grease rheological properties with shear rate, magnetic field, and its compositional elements. The results were then evaluated and compared with a constitutive equation known as the state transition equation. Apart from the shear stress results, where it had demonstrated the extreme learning machine models as having a better performance than the other methods with R2 more than 0.950 in the training and testing data, the predicted rheological variables such as shear stress, yield stress, and apparent viscosity were also proven to have an agreeable accuracy with the experimental data.
ISSN:	1045-389X 1530-8138
DOI:	10.1177/1045389X19844007