Genetic Algorithm and Machine Learning Based Void Fraction Measurement of Two-Phase Flow

Author

Wang Weiwei ; Zhu Xiaoqian ; Wang Ping ; Fan Shangchun ; Ren Dongshun

Author_Institution

Sch. of Inf. & Control Eng., China Univ. of Pet., Dongying, China

Volume

2

fYear

2010

Firstpage

355

Lastpage

358

Abstract

Machine learning and Genetic Algorithm based void fraction measurement method is provided in this paper. Because there are some relationships between the void fraction and the differential pressure (DP) signal acquired near the pipe wall when the two phases are flowing along the pipeline, it is possible to measure the void fraction according to the DP signal. However, the expression between the void fraction and the DP signal is complicated and is not easy to be developed because of the complexity of the characteristics of two-phase flow. In this paper, SVM is adopted to investigate the relationship between the void fraction and the DP signal. GA is used to estimate the parameters involved in SVM. The experimental results show that machine learning and genetic algorithm based void fraction measurement method provided in this paper is available.

Keywords

computational fluid dynamics; flow measurement; genetic algorithms; learning (artificial intelligence); parameter estimation; pipe flow; support vector machines; two-phase flow; SVM; differential pressure signal; genetic algorithm; machine learning; parameter estimation; pipeline; support vector machines; two-phase flow; void fraction measurement; Fluctuations; Fluid flow measurement; Genetic algorithms; Instruments; Machine learning; Phase measurement; Pipelines; Risk management; Support vector machines; Volume measurement; Genetic Algorithm; Machine Learning; SVM; Two-Phase Flow; Void Fraction;

fLanguage

English

Publisher

ieee

Conference_Titel

Measuring Technology and Mechatronics Automation (ICMTMA), 2010 International Conference on

Print_ISBN

978-1-4244-5001-5

Electronic_ISBN

978-1-4244-5739-7

Type

conf

DOI

10.1109/ICMTMA.2010.760

Filename

5459702