Optimal design of ECT sensors using prior knowledge

Abstract

Electrical capacitance tomography (ECT) is a noninvasive imaging modality aiming on the reconstruction of the dielectric material properties within a region of interest. The estimation of the permittivity distribution forms a nonlinear inverse problem, which is marked by a severe ill-posed nature. As a consequence, reconstruction algorithms rely on the use of prior knowledge or regularization techniques to overcome this difficulty. In this paper we want to make use of prior knowledge to optimize the design of ECT sensors. Typical ECT sensors maintain a number between 8 to 16 electrodes, which are mounted in an equidistant scheme on the circumference of the region of interest (e.g. a pipe). This is a suitable approach for sensors with a large number of electrodes. In this work we consider sensors with a considerable smaller number of electrodes and investigate the design of such sensors using prior knowledge. We discuss the incorporation of all modeling aspects and formulate a stochastic optimization function which will lead to optimized sensor designs. We present a simulation study and discuss the design results with respect to the physical properties of the sensing scheme.