A Kalman Filter Approach for the Application of Electrical Capacitance Tomography in Dynamic Processes using a State Reduction

Electrical capacitance tomography (ECT) refers to an imaging technique, which provides information about the spatial material distribution within an object. Since the image reconstruction forms a non-linear ill-posed inverse problem, prior information about the occurring material distributions is required in order to obtain stable reconstruction results. For the monitoring of dynamic industrial processes by means of ECT, recursive Bayesian estimation such as the extended Kalman filter (EKF) were shown to be suitable reconstruction algorithms. Though due to the usually high dimension of the inverse problem, computational costs are an immanent issue for the tracking of fast changes in the material distribution. In this paper we present a sample based state reduction approach for the extended Kalman filter. This approach not only allows stable reconstruction results but also involves significantly decreased reconstruction times due to the reduced state space.