Study On Identification Method Of Oil-gas Lubrication Two-phase Flow Pattern

The flow characteristics of oil-air phase flow have widely been concerned by researchers and engineers around all of the word with the wide application of oil-air lubrication in industrial production.The flow pattern identification method of oil-air two-phase flow in horizontal pipeline of oil-air lubrication system is studied in detail in this thesis.The main research work and conclusions are as follows:(1)The total pressure drop expression of the oil-air two-phase flow in the oil-air lubrication level pipeline is derived.The two-dimensional fluid domain physical model of the horizontal pipeline of oil-air lubrication system is established employing Fluent simulation software.The inlet velocity of air-liquid two-phase of each flow type is determined according to the two-phase flow pattern transition curve of horizontal pipeline of oil-air lubrication system.Four types of flow patterns in horizontal pipelines of oil-air lubrication systems is simulated and the differential pressure fluctuation signals of each flow pattern are obtained.The results show that there is a corresponding relationship between the two-phase flow pattern and the differential pressure fluctuation signal in the horizontal pipeline of the oil-air lubrication system.The Fluent simulation method can be used to obtain the differential pressure fluctuation signal of the two-phase flow of the oil-air lubrication system.(2)Wigner spectrum analysis is carried out on the differential pressure fluctuation signals of each flow pattern of oil-air two-phase flow.The results show that the differential pressure fluctuation signal of oil-air two-phase flow has non-stationary characteristics,therefore,the method of statistical analysis cannot be used to analyze it.EMD decomposition and wavelet decomposition,two methods of feature extraction,are employed to calculate and analyze comparatively the correlation coefficient between IMF(intrinsic mode)component and the original signal,the correlation coefficient between wavelet component and the original signal,and the variance of each component of each flow pattern respectively in this thesis.The results show that the IMF component has a better correlation with the differential pressure fluctuation signal than the wavelet component.Therefore,the EMD decomposition is more advantageous than the wavelet decomposition when extracting the characteristic of the oil-air two-phase flow differential pressure fluctuation signal.(3)The smoothing factor ? of PNN(probabilistic neural network)is optimized by PSO(particle swarm optimization)algorithm.The IMF component obtained by EMD decomposition of each differential pressure fluctuation signal is used as the input layer of the neural network.Four types of flow patterns commonly found in lubricated horizontal pipes are identified.The results show that the optimized PNN model can better identify the two-phase flow pattern in the oil-air lubrication level pipeline.