Study On The Propagation Law Of Electromagnetic Wave In Coal Seam And Its Full Waveform Probabilistic Inversion Method

Electromagnetic based Radio image method(RIM) is one of the most promising geophysical methods to identify geological structures in advance of coal mining, which is widely used for locating geological anomalies. Current imaging accuracy can not fit the requirement of coal mine production on account of inadequate observing direction and the usage of only amplitude parameter for tomographic image reconstruction. Therefore more capable method should be developed to improve solution as the coal seam becoming more geologically complex. This paper is focus on finding more sensitive observing parameters for better RIM image reconstruction. The propagation law of sensitive parameters including phase shift are studied by using numerical simulation and physical experiment. A full waveform probobilitic inversion method was constructed to improve the inverse resolution. The following are the main archievements.(1)A finite difference time domain(FDTD) approach is used to compute the total electric and magnetic fields within the coal seam. The amplitude and phase distribution of sinusoidal wave source in coal seam are archieved by imaginary and real componenent detection. Comparative sensitivity analysis between amplitude and phase parameters is implemented, and researches show that phase parameter has a good indication to the fault strike and the boundary of anomaly which is smaller than Fresel radius. Researches also show that phase parameter also has a good discrimination of coexistence of multiple anomalies.(2)Reconstructed results using simultaneous iterative reconstruction method reveal that phase data seem more amenable to recovering geometry information than amplitude data especially in lateral direction. Laboratory experiment based on vector network analyzer and wide angle coverage attena designed by ourselves also verified the feasibility of phase imaging.(3)Based on the superiority of phase imaging, this paper proposed a full waveform probilistic inversion method based on geostatistical simulation. Make use of Bayesian theorem combined with Markov-chain Monte-Carlo(MCMC) sampling, samples from the solution to the inverse problem can be obtained using the extended Metropolis algorithm in conjunction with priori information. The methodology provides a means of resolution analysis with posteriori samples from solution to the inverse problem.(4) Inversion for synthetic data reveals that full waveform inversion method dramatically improves the resolution of RIM method because the diffraction and scattering are taken into account suffiently. Bayesian-MCMC methods can incorporate any information that can be expressed in terms of probabilities as priori information so as to reduce the multisolution phenomena of inverse problem.(5) Probilistic modeling can provide more precise model parameter even with an arbitrary initial model. Noise sensitivity can be reduced by using data uncertainty model to describe data noise.The research results of this paper indicate a new way for RIM instrument designing based on new parameter observation.