A Study On Reducing Man-made Electromagnetic Noise In Geomagnetic Field Observation

Geomagnetic observation data not only possess academic significance, but also a kind of indispensable strategic resource. However, as China is experiencing the period of rapid social and economic development, man-made electromagnetic noise have caused an increasingly serious disturbance to the geomagnetic observation.According to incomplete statistics,1/3 geomagnetic observatories in China Geomagetic Networks are facing such kind of disturbance. The quality of produced geomagnetic observation data declined sharply.Therefore, it is an urgent task that we should do to identify the characterises of various types of man-made electromagnetic noise, meanwhile, effective methods to reduce these noise should be developed.This dissertation firstly investigates common sources of man-made electromagnetic noise in China Geomagetic Networks and their distribution in the whole Network, as well, studies the produced mechanism and mode of these man-made electromagnetic noises respectively, then, appropriate digital signal process methods are introduced to deal with these noises.The main contributions are summarized as follows:1. A more general and simple model base on vacuum line current magnetic field model in railway transit is presented to enable quantitative estimates of this magnetic perturbation. it's effective in fast estimating magnetic perturbation value. Theoretical calculations and field measurements show a good agreement.2. A novel approach based on independent component analysis to remove electromagnetic interference generated by urban railway transit. The Independent Component Analysis (ICA), which is a statistical method for transforming an observed mixed source signals into components that are mutually as independent as possible, is provided to separate the contaminated mixture signals into physical meaning components which belongs to its source. The results analyzed by power spectrum and wavelet based time-frequency representation. A comparative analysis between signals after removal and reference signals which did not contain EM interference nearby is made.3. A method of noise reduction caused by urban railway transit based on wavelet domain mixture noise shrink model is proposed. The thresholds of affected scales are estimated using inner-scale wavelet domain model, non-noise data in midnight between operations of train is identified using inter-scale wavelet domain model. The method uses the data itself or some prior knowledge of the reference data on the wavelet coefficients to build a more accurate model, meanwhile, a variety of filtering methods are combined to obtain the best filtering result. Mulit-angle comparative analyses are introduced to evaluate the denoising result.4. In order to deal with instrument noise and impulse noise encountered in geomagnetic observations, a data processing method based on mathematical morphology filter is proposed to processing the geomagnetic signals with such noise. We use this proposed method to process the stimulated geomagnetic diurnal variation signal combined with random noise and transient impulse noise. The numerical stimulation results indicate that mathematical morphology filter is effective to eliminate the random noise and impulse noise, improve the signal-noise ratio of geomagnetic signals. It's proved this new method show a good characteristic by processing the real geomagnetic signals with instrument noise and impulse noise caused by a passing vehicle.5. Extreme characteristic scale as well as pre-whitening technology is introduced as the key parameter to geoelectric counts interference identification and reduction. Normal distributed white noise as an assistive technology is added into original geomagnetic data. Since, the input signal, noise and nonlinear system the three relationships can achieve best match, the identification of geoelectric counts interference is becoming more stability and accuracy.6. An effective and user friendly interface preprocesses system which deal with geomagnetic electromagnetic interference is implemented. The system is base on Component Object Model and Windows DNA. Algorithm of interference reduction based on MATLAB is provided via Component Object Model.MFC architecture is used to develop user interfaces. Architecture model of the system and key technologies is provided.