Measurement Of Air Liquid Two Phase Flow Void Fraction Based On ECT

The void fraction is a major state parameter of two phase flow,which represents the proportion of each component on a cross section.The two phase flow exists widely in industrial scenes such as boiler piping,distillation towers,and nuclear reactors.The online measurement of two phase flow void fraction is of great importance to the safety monitoring and measurement of the production process.Electrical capacitance tomography(ECT)is an effective means of visualize two-phase flow or multiphase flow.The void fraction measurement of gas-liquid two phase flow based on ECT is studied in this paper.The two-phase flow void fraction measurement system is consisted of three parts: ECT system,gas-liquid two-phase flow system and void fraction measurement algorithm.Among them,the ECT system is composed of ECT sensor,data acquisition system and imaging computer.The interface limit of the data acquisition system determines the amount of the sensor plates,and the output range of the sensor is matched with the measurement range of the data acquisition system.In view of the existing data acquisition system in the laboratory,the finite element simulation of the sensor is made with COMSOL.The output value of the sensor is calibrated by an impedance analyzer.Finally,a 8 pole sensors matching with the acquisition system is designed and made.The ECT sensor,data acquisition system and imaging computer is connected to complete the assembly of the ECT system.The gas-liquid two-phase flow system is composed of oil pump,air pump and connecting pipe.The gas-liquid two phase flow system is connected with ECT system and combined debugging.The void fraction measurement algorithm based on ECT can be divided into two categories: first,the void fraction measurement algorithm based on finalization images,in which the image uses a real-time LBP(linear back projection)imaging algorithm,calculates the void fraction by calculating the proportion of the gas phase area in thebinaryzation images;the second algorithm is based on the normalized capacitance value prediction method,because the regression relationship between the normalized capacitance value and the void fraction of different flow patterns is different,the classification model of the normalized capacitance value and flow pattern identification is first set up.On this basis,the regression model of the normalized capacitance value and the void fraction of various flow patterns is established.Classification models and regression models are fitted by machine learning algorithm SVM,RandomForest,GBDT and AdaBoost.When the above four machine learning algorithms is applied to flow pattern recognition,GBDT performs best and the classification accuracy is96.63%.The simulation experiments and static experiments show that the four kinds of machine learning algorithms are used to predict void fraction and the GBDT performance is the best,and the prediction error of all types of flow patterns is less than 5%,which meets the requirements of industrial measurement.A two-phase flow void fraction measurement system with GBDT as the prediction algorithm is established,and the dynamic experiment of the two-phase flow is performed.The feasibility of the void fraction measurement method based on the ECT is proved.