Deep Learning for Nanofluid Field Reconstruction in Experimental Analysis

Abstract

Experiment is an important method to study the thermal and flow performance. Nevertheless, only limited information, such as local temperature and pressure, can be obtained through detection machines. Based on deep learning, a general, useful and flexible reconstruction model is proposed to reconstruct global flow field in two-dimension domain with the limited information exploiting from experiments as input information. Besides, the corresponding performance parameters Nu and f are extracted from generated fields. To validate the feasible, stability and accuracy of the framework, the micro channels with nano fluids are taken as validation case. First of all, the comparison between reconstructed fields and original fields are presented. It shows that reconstructed fields are almost the same as original ones and extracted performance parameters also have high precision. Next, the effects of train size, measuring uncertainty and measuring layouts are considered in sensitivity analysis. Higher train size and smaller measuring uncertainty are advantageous to the reconstruction results. Measuring layout has little influence on reconstruction performance and at least 7 local measuring points are enough. INDEX TERMS Deep learning, microchannel, field reconstruction, nanofluid.

Thermal fluid fields reconstruction for nanofluids convection based on physics-informed deep learning

“…In flow and heat transfer problems, performance characteristics Nu and f are widely used as objective functions. Thus, the prediction performance of different traditional surrogate models including Linear Regression (LR), Polynomial Regression (PR), Supported Vector Regression (SVR), Artificial Neural Network (ANN), Gauss Process Regression (GPR), Random Forest (RF), Extreme Gradient Boosting (XGB), our previous reconstruction model constructed by the deep convolutional neural network 36 (RDCNN) and our approach in this paper for Nu and f are plotted in Fig. 8 .…”

“…Kim 35 synthesized discrete velocity fields in space and time from a set of reduced parameters. In our previous study 36 , 37 , a reconstruction model with GAN and fields gradient loss is firstly proposed to predict the physical fields of nanofluids microchannel based on design variables, limited measurement and the effect of training size, measuring uncertainty and measuring layouts are discussed in detail. The image-inspired reconstruction models can obtain the overall physical fields in one prediction at the millisecond level and capture the spatial correlations among grid points efficiently.…”

Thermal Performance Estimation of Nanofluid-Filled Finned Absorber Tube Using Deep Convolutional Neural Network

“…When it comes to nanofluid field reconstruction using deep learning, not many articles are available. Liu et al proposed a GAN network [46] and then a CNN network [47] to predict the global flow field in a two-dimensional nanofluid-filled microchannel; the results also showed the accuracy and efficiency of the proposed model.…”

Thermal Performance in Convection Flow of Nanofluids Using a Deep Convolutional Neural Network

“…Recently a few studies have considered using nanofluids physical field prediction in heat transfer problems using deep convolutional neural networks. Two nanofluid-filled microchannels with three cylindrical grooves etched [18] and two grooves [19] in the upper plate were used and studied in the same year; both studies used the limited experimental data as the input. Then, the physical fields were reconstructed and the corresponding performance parameters, the Nusselt number and the Fanning friction factor, were extracted from the generated predicted fields.…”