A machine learning model for predicting the mass transfer performance of rotating packed beds based on a least squares support vector machine approach

Abstract

Rotating packed beds (RPBs) have been widely noted due to its superior gas-liquid mass transfer performance compared to conventional packed bed for CO2 absorption. The overall volumetric gas-side mass transfer coefficient (KGa) is selected as one of the key parameters for the screening and evaluation of RPBs. Existing theoretical and semi-empirical models for the KGa are easy to be used but have a poor accuracy and generalization ability. In this paper, a machine learning model based on least squares support vector machine (LSSVM) is developed to predict the KGa more accurately for CO2-NaOH chemical absorption system in different types of RPBs. Unlike the conventional prediction models, the input parameters are selected by multiple correlation analysis in the model establishment. Then, the proposed model is comprehensively evaluated by using four evaluation indicators, including determination coefficient, mean relative error, Root mean square error and standard deviations. The results show that the proposed model has the prediction performance with R2 = 0.9808 and RMSE = 0.0055 for testing set. In addition, the model performance is compared with the multiple nonlinear regression and artificial neutral networks. The results show that the proposed model has a superior performance for predicting the KGa of CO2 absorption in RPBs.