Thermo-physical properties prediction of carbon-based magnetic nanofluids based on an artificial neural network

Nanostructured magnetic suspensions have superior thermophysical properties, which have attracted widespread attention owing to their industrial applications for heat transfer enhancement and thermal management. However, experimental measurements of the thermophysical properties of magnetic-based nanofluids, especially under an external magnetic field, are significantly complicated, expensive, and time consuming. Currently, the method of predicting and summarizing material properties through machine learning has accelerated the development of materials and practical industrial applications. This study aims to predict the thermophysical properties of magnetic nanofluids by establishing an artificial neural network (ANN) using experimental data on viscosity, thermal conductivity, and specific heat. The results based on the ANN model agree with the experimental results according to the different evaluation criteria. Different previous theoretical thermophysical models are reviewed, and the ANN model is proven to be more accurate by comparing the values of the ANN model and previous thermophysical models, which can also provide a theoretical basis for explaining the heat transfer of magnetic nanofluids. In the present study, a neural network model was developed for predicting the thermophysical properties of magnetic nanofluids and using material informatics to study functional materials.

Automated summary

This study utilizes a neural network model to predict the thermophysical properties of magnetic nanofluids, demonstrating its accuracy compared to previous theoretical models and providing a theoretical basis for heat transfer.