The Implementation of FEM and RBF Neural Network in EIT

• Published in: 2009 Second International Conference on Intelligent Networks and Intelligent Systems pp. 66 - 69
• Main Authors: Peng Wang, Hong-li Li, Li-li Xie, Yi-cai Sun
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.11.2009
• Subjects: Conductivity measurement; Electrical impedance tomography; Equations; Finite element method; Finite element methods; Image reconstruction; Impedance; Intelligent networks; Inverse problems; Neural networks; RBF neural network; semiconductor section resistivity; Tomography; Voltage
• ISBN: 142445557X; 9781424455577
• DOI: 10.1109/ICINIS.2009.26

With the rapid development of electronic technology, semiconductor section resistivity measurement is receiving increasing attention. This paper applies electrical impedance tomography (EIT) technology to semiconductor resistivity measurements. FEM is applied to solve the EIT forward problem. Mathematical description of partial differential equation, equivalent variation differential problem, element characteristic matrix and the assembly rule of general matrix are given for calculation. To solve the EIT inverse problem, a new method of image reconstruction algorithm based on RBF neural network is proposed. This method can well adapt to non-linear and ill-posed characteristics of EIT. The simulation experiment results indicate that the RBF algorithm can improve the reconstruction image's quality and the accuracy obviously.