Machine learning for predicting the bubble-collapse strength as affected by physical conditions
Xiaojiao Wang, Zhi Ning, Ming Lv, Chunhua Sun
Abstract
The physical parameters are important factors affecting the bubble-collapse strength. In this paper, single factor analysis and multiple regression analysis are applied to study the relationship among the physical parameters and bubbles maximum dimensionless radius (Rmax). There is a very weak linear correlation between the density, viscosity, sound velocity, surface tension, saturated vapor pressure and Rmax, respectively. Furthermore, the regression models based on two types of machine-learning algorithms, namely random forest and neural network, are established to predict the bubble-collapse strength affected by combined physical parameters, the results show that all of them are feasible and efficient. In addition, the viscosity and surface tension have obvious influence on the collapse strength, and the influence of density and sound velocity are the least. Those results are helpful in understanding the bubble-collapse strength under different host liquids published in previous researches.