Applying different machine learning algorithms to predict the viscosity behavior of MWCNT–alumina/water–ethylene glycol (80:20) hybrid antifreeze
No Thumbnail Available
Date
2024
Journal Title
Journal ISSN
Volume Title
Publisher
Elsevier B.V.
Open Access Color
OpenAIRE Downloads
OpenAIRE Views
Abstract
While machine learning has become the new way of analyzing data, neutral networks form the basis of this revolutionary technology. In this work, we shall employ the power of neural networks to analyze and demystify the processes in nanofluids. By combining the precision of neural networks with the optimization capabilities of genetic algorithms, we aim to create a more accurate and efficient prediction model for MWCNT-alumina/water-ethylene glycol (80:20) hybrid antifreeze. Our approach entails using an MLP neural network and several training functions (LM, GD, BFGS, BN) with an adjustable number of neurons. The inputs of the network are φ (solid volume fraction or ϕ), temperature (T), and shear rate (γ), and the output is μnf of MWCNT-alumina/water-ethylene glycol (80:20) hybrid anti-freeze. To improve the accuracy of the final model, we use genetic optimization to make final adjustments to the parameters of the neural network. Utilizing the detailed analysis of the primary characteristics of these algorithms, we conclude that the BFGS function is the best to obtain neural network training. Steady performance achieved by this function—0.99828 of the R-value and RMSE value significantly equal to 0.213—illustrates good stability and accuracy of the suggested model. This work contributes to progressing the existing knowledge about the behavior of nanofluids and can stimulate further improvement in heat transfer and energy utilization. © 2024 The Author(s)
Description
Keywords
Hybrid antifreeze, Machine learning algorithm, MLP neural network, Viscosity behavior
Turkish CoHE Thesis Center URL
Fields of Science
Citation
0
WoS Q
N/A
Scopus Q
Q1
Source
International Journal of Thermofluids
Volume
24