Title :
Flow Regime Identification Based on Multi-Class Support Vector Machine of One-Against-One
Author_Institution :
Dept. of Comput. Sci. & Eng., City Coll. of Ji Lin Archit. & Civil Eng. Inst., Chang Chun, China
Abstract :
The complex flow characteristic of two-phase fluid makes the identification of flow regime the foundation of accurately measuring two-phase flow´s parameter. A method of flow regime identification based on multi-class support vector machine of one-against-one is proposed. The feature of original measured data is firstly extracted by wavelet packet analysis, then the extracted feature data is inputted into the multi-class support vector machine of one-against-one, so four two-phase flow regimes can be identified. Simulation experiment show that speed and precision of identification is higher, and the flow regime identification based on multi-class support vector machine of one-against-one is an effective method.
Keywords :
computational fluid dynamics; feature extraction; identification; support vector machines; two-phase flow; wavelet transforms; complex flow characteristic; feature extraction; flow regime identification; identification precision; identification speed; multiclass support vector machine; one-against-one; two-phase flow parameter; two-phase flow regimes; two-phase fluid; wavelet packet analysis; Data mining; Feature extraction; Support vector machines; Training; Wavelet analysis; Wavelet packets; Flow regime identification; One-against-one; Support vector machine; Wavelet packet Analysis;
Conference_Titel :
Computational and Information Sciences (ICCIS), 2013 Fifth International Conference on
Conference_Location :
Shiyang
DOI :
10.1109/ICCIS.2013.40
