Research On Regime Identification Of Oil-Gas-Water Three-Phase Flow Pattern

Along with the rapid development of modern industry, oil-gas-water three-phase flow widely exists in fields such as petroleum, power, chemistry, medicine industries. The flow regimes are the important characteristic parameter to analyze and describe three-phase flow, the run performance of three-phase flow system and the accurate measurement of other parameters are extremely influenced too. Therefore, the objective and intelligent identification of three-phase flow regimes has long been considered as a signification research topic that is badly in need of be solved in science and engineering application.Based on a large amount of experimental data, HHT, complexity measure theory, chaotic recurrence characteristics analysis and intelligent identification model are used in flow regime identification. Intelligent three-phase flow regime identification method is discussed systematically from the aspects of theory and experiment.First of all, dynamic test were brought out by seven-electrode correlation conductance probe base signal acquisition system on the three-phase flow field simulation well experimental platform in Daqing Oilfield, and different flow regimes conductance wave data of oil-gas-liquid three-phase flow in vertical upward pipe are measured.After that, HHT, complexity measure theory and chaotic recurrence characteristics analysis are used in the characteristic extraction of measured data. First of all, EMD power entropy and IMF kurtosis coefficients are extracted by HHT,so the feature paramerers are regarded as one of the feature vectors, the change rules of different flow regimes are analyzed; and then, the randomness degree and periodic rules of different flow regimes are discussed by complexity measure theory, Lempel-Ziv complexity and approximate entropy are extracted as the feature vector; finally, on the basis of discussion of phase space reconstruction parameters calculating methods, recurrence plot is used for characterization of different flow regimes, the four kinds of recurrence quantification indicators are calculated as the feature vector and the recurrence characteristics of different flow regimes are quantized analyzed.At last, on the basis of feature extraction, training BP neutral network, traditional SVM and LS-SVM, the last identifying model is used in flow regimes identification.