Study On Parameter Detection Of Gas-Liquid Two-phase Flow Based On Capacitively Coupled Electrical Resistance Tomography

Gas-liquid two-phase flow is very important for scientific research and industry application.It widely exists in petroleum industry,chemical industry and many other fields.Compared with single-phase fluid,flow pattern and void fraction are two special parameters of gas-liquid two-phase flow.It is of great importance to realize effective flow pattern identification and void fraction measurement.Capacitively coupled electrical resistance tomography(CCERT)is the combination of capacitively coupled contactless conductivity detection(C4D)and electrical resistance tomography(ERT).It has the advantages of contactless,simple structure,low price and good robustness,thus receiving widespread attention from researchers.However,the current CCERT technology needs to implement image reconstruction of the pipe section first,and then conduct flow pattern identification and void fraction measurement according to the images reconstructed.Its accuracy and real-time performance are greatly influenced by the image reconstruction algorithm.So the existing methods have some shortcomings.To solve the problems mentioned above,based on the previous work,an industrial CCERT system was designed according to the requirements of industrial instrumentation.Then the resistance data of the pipe section was collected by the system designed.The resistance data was directly analyzed by the data mining algorithms and then realize flow pattern identification and void fraction measurement without implementing image reconstruction.The main works and innovation points are listed as follows:(1)A prototype of industrial CCERT system was developed to realize data collection and tomography.Combined with the requirements of industrial instruments for stability and safety,the power supply module and the data acquisition and processing module of the system were designed independently.(2)A contactless gas-liquid two-phase flow pattern identification method was proposed.The K-Means and KNN algorithms were used to process the experimental data after dimensionality reduction by PCA respectively,and the static and dynamic flow pattern identification of gas-liquid two-phase flow were realized.The experimental results showed that both of the two methods had satisfactory accuracy for static and dynamic flow pattern identification,and the comprehensive effect of KNN algorithm was better than K-Means algorithm.The experimental results further verified the validity of the proposed flow pattern identification method.(3)Based on the results of flow pattern identification,the void fraction measurement models under different flow patterns were established by using ridge regression and partial least squares regression respectively,and the static and dynamic void fraction measurement of gas-liquid two-phase flow were realized.The experimental results showed that the void fraction measurement models established by ridge regression and partial least squares regression had satisfactory accuracy for the void fraction measurement of bubble flow,stratified flow and annular flow,and the comprehensive effect of partial least squares regression method was better than ridge regression method.The experimental results further verified the validity of the proposed void fraction measurement method.