Analysis of response for magnetic induction tomography with internal source

Abstract

Magnetic induction tomography (MIT) can reconstruct the distribution of conductivity within the sensing field in a non-invasive and non-intrusive manner, thus receives wide attentions for its applications in industrial and biomedical detections. In some particular cases, internal signals exist in the sensing field that would affect the response of MIT. The information of the internal signals might aid the understanding of the condition of the process/object being measured. In order to understand the effects of internal sources upon the reconstruction accuracy and dynamic responses of MIT, the Izhikevich model is applied to generate different internal sources, and numerically simulated to study the responses of magnetic induction with the internal signals. The simulation results show that an internal signal causes a shifting of the object’s location in the reconstructed image, and the amplitude of the U-shape curve formed by the measurement of MIT receivers is helpful to recognize the internal sources of different signal pattern and number. Dynamic responses of the MIT to the point source model and the multi-point source model of different internal signal shapes are also numerically investigated. A modified independent component analysis (ICA) method is introduced to separate the internal signals from the measured signals. Simulation results show that the internal signals induced in MIT’s receivers can be separated by using ICA, and the kurtoses of the separated independent components are robust in identifying different internal sources. This work verifies that MIT could handle the coupling information detections, and the coupled signals could be separated effectively with ICA.