Both phase fraction and flow velocity are two key parameters in the detection of multiphase flows. Electrical Resistance Tomography (ERT) is an advanced measurement technique that has been increasingly addressed in recent years owe to visualization, non-intrusiveness, real-time performance, environmental protection, etc. However, due to the complex measurement process in dredging engineering, the ERT measurements are easily affected by noisy data and system faults, resulting in inaccurate estimation of phase fraction and flow velocity. Therefore, the accurate control and optimization of dredging operations are not guaranteed. In this paper, we recover the main factors that lead to inaccurate measurements for phase fraction and flow velocity. According to their various categories, the criteria to distinguish normal and abnormal measurements are proposed. And the method to overcome the effect of abnormal data is built so that the ERT measurements can normally remain correctly being used even though partial faults and noisy data are encountered. Simulations and experiments in actual dredging engineering are used to verify the effectiveness of the proposed method.