Location of Underground Multi-layer Media Based on BP Neural Network and Near-field Electromagnetic Signal

 This paper includes a variety of media location data generated in the paper.According to the wideband near-field signal propagation model, sample data were generated based on four layers of media with known media types. The main differences between simulated data and real data are errors due to electromagnetic dispersion, multipath and noise, which are not currently taken into account because of different electromagnetic characteristics in inhomogeneous media. The setting of condition variables in the experiment was shown in Table I. Based on the experimental conditions, the simulation of the wideband near field electromagnetic ranging and positioning experiment is carried out. Because signals of different frequencies in the medium have different relative permittivity, ALRM, a widely used wideband mixing model, is used to calculate the relative permittivity.And the experimental data in this paper are as much as possible to simulate the signal information obtained by receiving and processing the transmitter in reality, and the noise is added to be closer to the real situation.