Identification Of Gas-liquid Two-phase Flow Regime Based On Symbolic Dynamic Filtering

The distribution pattern of different media in two-phase flow is called flow pattern.The flow pattern of gas-liquid two-phase flow system is one of the important parameters to characterize its flow characteristics.Gas liquid two phase flow pattern recognition plays an important role in the safety of industrial production such as petroleum and chemical industry.At present,flow pattern recognition technology based on mathematical models has become the main trend of development.A new soft sensing method based on symbolic dynamic filtering is proposed in this paper,which provides a new idea for identifying the flow pattern of gas-liquid two-phase flow.The main contents of this paper are as follows:(1)A vertical transparent ultrasonic measurement pipeline was designed and made.A set of vertical rising gas-liquid two-phase flow experimental device was set up to meet the requirements of measurement.The experiments were carried out for four flow patterns of pure water,bubbly flow,elastic flow and annular flow;(2)This paper focuses on the theory of symbolic dynamic filtering and introduces the analytical steps of the symbolic dynamic filtering method.The experimental data of the four flow patterns are processed and analyzed.The similarity measurement of the same flow pattern and the reference state transfer matrix corresponding to the different flow patterns is carried out to characterize the difference in the fluctuation of the two-phase flow under different flow patterns.The state transition matrix of the same flow pattern is similar,and the similarity of the state transfer matrix under the different flow patterns is lower,which provides the basis for the flow pattern recognition;(3)According to the analysis of the recognition model,the size of the symbol set is 4,5 and 6 under different thresholds.When the threshold is 0.2 and the corresponding symbol set is 5,the correct rate of flow pattern discrimination is the highest.Meanwhile,in order to improve the accuracy of flow pattern recognition,the number of training sets is increased,and the recognition rate of flow pattern reaches 92%.At the same time,power spectrum density analysis and symbolic dynamic filtering method are applied to compare the performance of flow pattern recognition,which proves that the accuracy of the latter is higher than that of the former.