Oil-water Two Phase Flow Pattern Recognition Based On Electric Conductance Fluctuation Signal

Characteristic of flow and heat transfer of oil-water two-phase flow are strongly affected by its flow regime. Therefore, the study on flow regime is always an important subject of two-phase flow. There are two traditional methods to recognize flow patterns, one of which is the observation or measurement and another is transition formulas or the flow pattern maps. The traditional recognition methods of flow patterns depend largely on subjective judgment of researchers and make the on-line recognition of flow patterns impossible, so the traditional recognition methods need to be improved.The conductance fluctuation signals of the oil-water two-phase flows in vertical upward pipe are adopted and analyzed on oil-water two-phase flow test equipment. At the same time, wavelet transform, wavelet packet decomposition, EMD(empirical mode decomposition) decomposition and neural network are used in flow regime identification. Intelligent regime identification method based upon classification model is discussed systematically from the theoretical and experimental perspectives.Firstly, the conductance fluctuation signals are de-noised by Wavelet transformation. Then the conductance fluctuation signals are analyzed by wavelet packet decomposition, EMD decomposition, wavelet packet energy and IMF energy are extracted, and then use neutral network of BP(Back Propagation), RBF(Radial Basic Function), improved-RBF and Elman as oil-water two-phase flow regime recognition model. Lastly by training network of BP, RBF, improved-RBF and Elman model with different samples of different eigenvectors, these last identifying models are used as regime recognition classifier. The simulation result shows that the combination of IMF energy and improved-RBF neutral network is the best among these models. All these pave a new way for identifying the oil-water two-phase flow regimes from Theory and technology.