This research presents a unique method for estimating nanofluid viscosity by building a smart generalized model on top of a deep neural network (DNN). The DNN model was trained using the Nadam optimization approach on a large experimental dataset that contained Alumina (Al₂O₃) nanoparticles. Nonlinearities may be automatically learned by the proposed deep neural network model from a training dataset. This paper details the innovative aspects of this investigation and how they combine with the benefits of deep learning. To the author’s knowledge, no prior attempt was made to predict viscosity using a model based on deep learning. The comprehensive investigation of this DNN model’s efficiency demonstrates that it outperforms all competing models while also avoiding their pitfalls. Additionally, our DNN model provides remarkably accurate predictions on unseen data and can be trained in a fraction of the time mandatory by conventional data-driven models. This intelligent model has also been subjected to a sensitivity study. With a coefficient of determination of 0.9999, our unique DNN-based smart model is the best at predicting the viscosity of nanofluids.

PDF (Download Full text)

Download Citation (XML)