FLOW REGIME IDENTIFICATION OF GAS-WATER TWO-PHASE FLOW USING SVM

摘要

The automatic and objective identification of gas-water two-phase regime has important industrial background and scientific significance. In this paper, the gas-water two-phase flow regime in horizontal tube was taken as experimental object. A novel powerful learning method called Support Vector Machine (SVM) is applied to identification of flow regimes. Support Vector Machine is an effective and general method for representing complex function in high dimensional space which is suitable for pattern recognition and function regression et al. According to the design theory presented in the paper, the experiment was carried out in the multifunctional flow equipment in Institute of Automation and instrument of Zhejiang University. The pressure difference signals on Venturi meter of measured tube segment were taken as experimental signals. Then autocovariance function (r(0),r(2)) and partial autocorrelation function of the pressure difference signals were taken as feature parameters and trained . RBF is adopted as kernel function. Then the two-phase flow regime in horizontal tube was identified by the method of pattern recognition. The result shows that this method has high precision for familiar flow regime identification, for example bubble flow, slug flow and churn flow et al.