Position estimation based on Electrical Impedance Tomography
S. Vergara 1*, D. Sbarbaro 1
1 Department of Electrical Engineering, Universidad de Concepción, Chile
This work deals with position estimation of known shape anomalies in a conducting medium without solving the EIT inverse problem for the resistivity field. Optimization based method and data driven approaches are proposed and compared using simulated and experimental data. Optimization based method uses the full electrode model and information of the anomaly shape to find its position by means of minimize a cost function. On the other hand, data driven approaches use models obtained from a set of data points (training set), which directly map the measured voltages into the position estimates. It has been found that the estimates obtained by optimization approach are sensitive to the model characteristics such finite element mesh size. The accuracy of data driven approaches is more sensitive to the training set and the model structure used for modelling the relationship between measured voltages and position. For some applications the use of model based or data based approaches can be an attractive alternative for estimating low dimensional features compared to the full reconstruction of the resistivity field. These approaches do not require solving high dimensional optimization problems reducing in this way the computation times and enabling their use in real-time applications.
Keywords Optimization, position estimation, modelling.
Copyright © International Society for Industrial Process Tomography, 2016. All rights reserved.