| Summary: | 碩士 === 國立臺北科技大學 === 製造科技研究所 === 94 === In the recent decades, the great progresses have been made on the measurement techniques for tiny magnetic fields. This creates more possibilities in the applications of high-resolution magnetic-flux-density maps. However, to figure out the source current distribution according to a magnetic-field map is an ill-posed problem, which does not have a unique solution. Some appropriate assumptions on the source are necessary in order to obtain an approximate solution of the inverse problem.
In this work, two inversion algorithms, including the Fourier transform (FT) and the minimum norm estimate (MNE), are compared in their spatial resolutions and image qualities achievable. The magnetic-flux-density map are measured with the giant magnetoresistive magnetometer, of which the detectable range is 2 mT to 2 nT (120 dB) and the flat bandwidth is 20 kHz. It was found that the major factors affecting the image quality of source currents are the source-to-field distance, the current continuity, and the magnetometer noise. The MNE method is slightly superior to the FT in image quality and spatial resolution, but the FT algorithm is much faster. With the optimized filtering, one can satisfactorily retrieve the source-current distribution. For the MNE algorithm, the optimal filtering parameter of singular values is about 0.01 to 0.001, which is independent of the source-to-field distance. For the FT algorithm, the optimal cutoff spatial frequency is about 2.0 mm-1 to 2.5 mm-1 with the source-to-field distance of 3
|
|---|
