Using deep learning algorithm to predict flow boiling pressure drop under ultrasound fields in mini-channels

Abstract

Ultrasound is widely acknowledged as an effective method for enhancing active heat transfer, thereby significantly augmenting the heat dissipation capacity of mini/micro-channels. However, there is lack of accurate methods for predicting two-phase pressure drop under ultrasound field. The study proposes a novel deep learning-based approach, namely the Multi-Scale Convolutional Neural Network (MSCNN) combined with Time-Frequency Representation (TFR), for predicting flow boiling pressure drop in mini-channels under ultrasound field. Time-series pressure drop fluctuation data of gas-liquid two-phase were transformed into useful TFRs data by wavelet transform. TFR could effectively reveal the pressure drop fluctuation of gas-liquid two-phase in mini-channel under ultrasound field, but the TFRs data scale was the high dimensionality and consumes computer resources. Therefore, bilinear interpolation was used to reduce TFRs data scale and as input of deep learning model. Compared with the traditional convolutional neural network (CNN) model structure, MSCNN model structure has global and local information synchronization. The prominent features are helpful to predict flow boiling pressure drop in mini-channel under ultrasound field and to the automatic learning of MSCNN. Experiments showed that the prediction performance of MSCNN model has been greatly improved than the traditional data-driven.