Ultrasonic identification method of oil-gas-water multiphase flow pattern based on K-means clustering algorithm

Oil-gas-water multiphase flow pattern identification is widely existed in aerospace, petrochemical and other fields. Existing identification technologies may have problems such as interference, high cost, and environmental requirements, we carry out the transient liquid phase distribution simulation experiments of stabilized fluid in horizontal pipeline and related researches by acoustic method. Based on the finite element method, design a multi-transceiver distributed ultrasonic test system. we carry out simulation experiments on typical flow patterns of oil-water two-phase flow and oil-gas-water three-phase flow. In order to recognize the flow pattern in oil-water two-phase flow, the average instantaneous acoustic pressure difference is extracted as the input feature vector, the K-means neural network algorithm is adopted to recognize the flow pattern, and the overall recognition accuracy of the five oil-water two-phase flow patterns is 90%. For the identification of liquid phase distribution in oil-gas-water three-phase flow, we analyze the attenuation law of the liquid phase discrete phase fraction rate and ultrasonic transmission, as well as the average acoustic pressure, reflected echo intensity and echo duration characteristics in different flow types, to identify the oil-gas-water multiphase flow pattern. The experimental results show that the combination of ultrasonic transmission attenuation and ultrasonic reflection echo phase distribution identification method can identify the oil-water two-phase flow pattern and the liquid phase distribution law in the oil-gas-water three-phase flow, which can provide a theoretical basis for the identification of gas-liquid-liquid multiphase flow pattern.