Two-Phase Flow Parameter Measurement Using Multi-Task Learning and Electrical Capacitance Tomography
YANG Haichao,HU Hongli,LU Chengcheng,LU Jiayu,WANG Maojie
DOI: https://doi.org/10.7652/xjtuxb202303019
2023-01-01
Abstract:A two-phase flow parameter measurement method, which combines multi-task learning and electrical capacitance tomography, is proposed to solve the problem of the interference in parameter measurement caused by the low accuracy of the images reconstructed by the reconstruction algorithm in terms of electrical capacitance tomography. Firstly, the residual module is used as the feature extraction module to extract the features of the images reconstructed by the Landweber algorithm, so as to eliminate the interference of artifacts of reconstructed images on parameter measurement. Secondly, a dual output multi-task neural network with hard parameter sharing is built to finish the task of initial measurement of the phase fraction in order to further improve the anti-noise ability of the model. Then, a method of fusing neural network and classification models, such as extreme random forests, is proposed to achieve flow identification. Finally, regression models such as extreme random forests are used to obtain the final phase fraction by fusing the results of flow pattern identification and initial results of the phase fraction in order to further improve the accuracy of phase fraction measurement. The experimental results show that the proposed algorithm can improve the accuracy of phase fraction measurement when compared with the method of only using the neural network. In the actual measurement, the relative error of phase fraction measured by the proposed algorithm is within 5%, and the accuracy of flow identification is beyond 98%, which shows a good anti-noise ability of the model and can meet the demands of practical industrial applications.