Liquid holdup prediction model based on big data approach

Author of the article:ZHENG Lin, LIU Yun

Author's Workplace：Petroleum Engineering College, Yangtze University, Wuhan, Hubei, 430100, China

Key Words：Gas-liquid two-phase flow; Liquid holdup; BP neural network; Grey Relational Entropy; Genetic Algorithm

Abstract：In order to solve the issue of liquid holdup of gas-liquid two-phase flow in a pipeline, a new genetic algorithm based on big data approach was established through software programming to optimize the neural network model weighted by Grey Relational Entropy (Genetic Algorithm Optimizes Neural Network Model Weighted by Grey Relational Entropy, GA-GRE-BP). The basic BP neural network model and the traditional model were selected to predict the liquid holdup respectively and used for analysing the accuracy and feasibility of the new model. The results show that compared with the basic BP neural network model, the new one not only converges faster, but also has a significant improvement in prediction accuracy. When compared with the traditional model, the GA-GRE-BP neural network model has a wider range of applications and is simple to apply. This shows that the GA-GRE-BP neural network model is accurate and feasible for predicting the liquid holdup of gas-liquid two-phase flow.