EIT image reconstruction for two-phase flow monitoring using a sub-domain based regularization method

Electrical impedance tomography (EIT) has been employed for two-phase flow monitoring because of high temporal resolution characteristics to monitor fast transient processes. EIT inverse problem is highly non-linear and ill-posed. Therefore, the regularization methods are used as an additional penalty term to stabilize the inverse problem for having meaningful images. Depending on the type of regularization method, the reconstruction performance is different. A common regularization method is the generalized Tikhonov regularization using difference-type regularization matrix. This method assumes that the change in resistivity is equal and isotropic, so it causes blurring regions between anomalies and background and often underestimates resistivity of anomalies. Therefore, in this paper, a sub-domain based regularization method with adaptive thresholding is used to improve spatial resolution for two-phase flow monitoring. In the proposed regularization method, a part of whole computational domain is considered as a sub-domain, and it is incorporated into the regularization method as prior information. The regularization matrix is asymmetrically modified to include the sub-domain, and regularization parameter is set to different weight for the sub-domain. The performance of the proposed regularization method is evaluated with numerical simulations and phantom experiments. The results show better reconstruction performance with the proposed regularization method.