Based on the dielectric properties of liquid and vapor phases of cryogenic fluids

the modified linear inversion algorithms of electrical capacitance tomography (ECT) are obtained by introducing a correcting deviation into the linear approximate equation of the traditional capacitance-permittivity distribution function. The 8-electrode capacitance tomography sensor was used as the geometric model to simulate the two-phase flow in the liquid oxygen-oxygen tube. The original linear algorithm

the modified linear algorithm and the total variation 1 norm regularity were compared under four different flow patterns. The inversion results of four different flow patterns of original linear algorithms

modified linear algorithms and non-linear algorithms represented by total variation L 1-norm regularization (TV L 1-norm) are compared. The correlation coefficient and image error are taken to measure the quality of the inversion results. The results of numerical experiments show that compared with the original linear algorithms

the quality of the inversion results is greatly improved by the modified linear inversion algorithm. The TV L 1-norm takes more than 10 times as much time as the modified linear inversion algorithm. And when 3% measurement noise is applied

the non-linear algorithm shows strong sensitivity to measurement noise. The modified linear algorithms have better anti-noise ability and accuracy of inversion results.