The diffusive electromagnetic field can be transformed into the wave
domain by means of mathematical conversion. The transformed field can then
be interpreted with the tools in seismic data processing so that the
identification to the underground targets can be effectively improved.
However, the conversion is typically an ill-posed problem that needs to be
solved using regularization tools. Based on the conventional regularization
with smooth constraints in the L2 norm, the inversion result is of low
resolution, while that obtained using truncated singular value decomposition
(TSVD) methods is typically accurate, but has poor stability. To obtain a
stable and accurate transformed electromagnetic field value, this study
proposed to combine conventional regularization tools and singular value
decomposition algorithms by incorporating a set of weighting coefficients.
The proposed method is validated on both synthetic and observed data. The
results from the proposed method are more accurate at the early time, and at
the late time are more stable compared with the other methods. Furthermore,
the example of field data shows that the proposed method could potentially
further improve the interpretation accuracy of future mining
explorations.
Recommendation system using the k-nearest neighbors and singular value decomposition algorithms
