EXPONENTIAL MODEL NORMALIZATION OF ELECTRICAL CAPACITANCE TOMOGRAPHY
M. R. Baidillah and M. Takei
Graduate School of Engineering, Div. Artificial System Science, Dept. Mechanical Eng.
Chiba University, 1-33, Inage-ku, Chiba-shi, Chiba, 263-0022, Japan
To overcome the ill-determined condition, commonly, the inverse problem of Electrical Capacitance Tomography needs to constrain the predicted permittivity value of object interest by normalizing the measured capacitance in-between of two known-medium. Mathematically, a widely used normalization is based on a linear approach, either series or parallel model. This is based on the assumption in equivalent circuit that background medium and inhomogeneity medium are in parallel or series composition. However, due to the nonlinear relationship between permittivity and capacitance, linear approximation becomes a limitation and a new issue as it will eliminate information and affect to error in reconstructed image. In this study, we propose a new normalization technique based on a nonlinear model, i.e. exponential model, which can be applied for a wide range of dielectric permittivity prediction and particularly for a relatively complex distribution. The exponential is proposed based on the relationship between the capacitance change and permittivity change is in an exponential form. A numerical study was conducted to analyze the nonlinearity of capacitance measurement and to test the performance of the proposed technique. The results showed the correlation coefficient of exponential model’s result are higher compared to the linear model’s result.
Keywords Normalization technique, Exponential Model, Nonlinear Model, Inverse problem ECT
Copyright © International Society for Industrial Process Tomography, 2016. All rights reserved.