A smart model for prediction of viscosity of nanofluids using deep learning
Satyasaran Changdar, Susmita Saha, Soumen De
Abstract
A novel approach is presented in this paper to predict the viscosity of nanofluids by developing a deep neural network (DNN) based smart generalized model. This study is conducted with a large experimental dataset containing Al2O3, CuO, SiO2, TiO2, Ag, and Fe2O3 nanoparticles and the DNN model is trained by Nadam optimization technique. This proposed DNN model has the learning capability of non-linearities from a training dataset automatically. The novelty of this study with the advantages of deep learning has been described in this paper. To the best of the author’s knowledge, deep learning-based model was never used to predict viscosity before. The detailed analysis of the performance of this DNN model shows that it performs better than any other existing model and overcomes all of their limitations. Also, it gives an excellent prediction on the unseen data and it takes very less amount of time to train this DNN model comparing to other traditional data-driven models. A sensitivity analysis of this smart model has also be presented. This novel DNN-based smart model can predict the viscosity of nanofluids with the highest level of accuracy with a coefficient determination 0.9999.