Extraction Of Artificial-source SLF/ELF Electromagnetic Signal And Data Processing

The signal of SLF/ELF electromagnetic wave from an artificial source is characterized by stability, large intensity, and high precision. And its frequency, amplitude and phase are easy to be controlled, which can cover a large area. Thus it is quite suitable to monitoring of earthquakes and volcanoes which requires highly accurate observations. Although the artificial-source SLF/ELF electromagnetic signal is powerful, it will become very weak after propagation over a long distance and usually submerges in strong background noise. In order to improve the prospecting accuracy and efficiency of this method, based on achievements of the predecessors, this thesis makes a study of the extraction methods for artificial-source SLF/ELF electromagnetic signal. Firstly, the principles and algorithms of both the correlation detection technique and the adaptive filter technique are presented, and the similarities and differences of the two techniques are summarized. Especially, the key technique of the adaptive filter technique is analyzed. Secondily, theoretically arithmetic simulation of two methods and a case application are carried out by using MATLAB ,both in the cases that beginning and end times are available and unknown, respectively. Then the filter effects of two methods are compared. In the end , the program is written to compute the apparent resistivity and phase. Comparing with the results of the MT algorithm, the possible factors that affect the result are analyzed. The major results of this thesis are summarized below.(1) Comparison study of the two signal extraction methods for the artificial-source SLF/ELF electromagnetic technique, including their similarities and differences. The correlation detection technique and adaptive filter are both based on the cross correlation function to extract the signal. They use the correlation between the periodic signals and the non-correlation between the random signals to distinguish them, but their algorithms are different. The correlation detection technique is to directly estimate the values of amplitude and phase, and construct the object signal. While adaptive filter makes estimation of the object signal by using numerical simulation under the criterion of mean square error, where there is an iterative process. This is reason that the adaptive filter can achieve a good result when the beginning time of emission is unknown. Based on further analysis, this thesis has improved the iteration-based algorithm of the adaptive filter method, suggesting a two-step way to enhance the signal extracted.(2)Correcting the phase estimation formula of the correlation detection technique. By using MATLAB, extraction and analysis of the simulated artificial signal are performed. It is found that when the correlative function between the mixed signal (artificial-source signal and noise) and the sine signal in two reference signals is negative, the correlation detection result differs much from the theoretical amplitude. Through analysis and many times of tests, two formulas for phase estimation are derived which improve the previous algorithm. Comparison indicates that the result of signal extraction by the adaptive filter is a little better than that of the correlation detection in the case that the begin and end times of artificial-source emission are known. While in the case that these times are unknown, the extraction result of the correlation detection has a large error while that of the adaptive filter remains pretty good. The simulation of this work shows that both the methods can extract well very weak sine signals from the noisy background, and the adaptive filter has a broader range of application. In general, the error of the signal extraction by adaptive filter is smaller than that by the correlation detection.(3) Extracting real artificial-source signal containing intrinsic frequency and displaying variations of the amplitude of this signal before and after emission using the adaptive filter method. In this work the adaptive filter method is used to process the real data containing artificial-source signal. The filtering result shows that the field amplitude of the time section with artificial-source signal is considerably larger than that of the section without the signal. So it displays variation of the signal amplitude at this frequency before and after the emission of the artificial-source signal. When the start and end times of the emission are unknown, the filtering can be performed first by the adaptive filter method. Then based on variation of the signal amplitude after filtering, the signal emission time section is roughly estimated. In this way the artificial-source signal in the emission time section can be further extracted by using the correlation detection technique. Meanwhile, the adaptive filter method, which can trace signal of intrinsic frequency, enables it possible to monitor variation of amplitude of the electromagnetic signal at a certain frequency in the time domain.(4)Accomplishing the work from data reading to calculation of apparent resistivity and phase and comparing it with the results of the MT method. First ,the apparent resistivity and phase are computed by the software Mapros and the program written by the author. The comparison of the results shows that the curve forms are almost consistent. This proves that the program written by the author is correct. Then the real data is processed by the correlation detection technique and adaptive filter, respectively. The results demonstrate that the apparent resistivity curves are nearly consistent in the form after filtering and processing by the software Mapros, but the phase curves are quite different at low frequency. The primary reason may be big errors caused by the two filter methods, and this needs to be verified by further studies.