Research On 3D Forward Modeling And Inversion Method Of Surface-tunnel NMR For Water-bearing Structures In Coal Mine

Coal is the main energy in China.For a long time in the future,coal will still occupy the dominant position in energy production and consumption structure.With the deepening of mining depth,the risk and frequency of dynamic disasters including water inrush also increase.This poses a great threat to the safe and efficient production of coal mines.The main reasons for the frequent occurrence of water inrush accidents are the unclear understanding of the disaster mechanism,the unclear identification of the disaster source and the insufficient detection effectiveness.Therefore,it is of great significance to improve the level of mine water hazard prevention and control to innovate detection methods,thereby improving the ability of accurate detection and early identification of disaster water sources.At present,NMR is the only geophysical method to directly find water.Compared with the conventional electromagnetic method,it has the advantage of quantitatively interpreting the information of underground water content.Among the NMR detection methods,the Surface-tunnel NMR method（in which the Tx loop is on the ground and the Rx loop is in the tunnel）is a new detection method proposed in recent years.This method can break through the detection boundary of the traditional NMR method,and enhance the capability of NMR excitation and the geological information.However,STNMR is still in its infancy at this stage,and there are many problems that need to be solved,such as the imperfect theoretical system of forwarding and inversion,and the unclear response characteristics of complex water-bearing structures,which greatly limit the practical application of this method.In view of the above deficiencies,this dissertation focuses on the actual engineering needs of the STNMR method to detect the disaster-causing water-bearing structures in coal mines,and carries out the research by combining theoretical analysis and numerical simulation.The performance analysis,optimization method,response law,forward and inverse method of STNMR were studied for the problems.The main research contents and achievements include:1.Improvement and perfection of STNMR 3D forward modelling method.Forward modelling is not only the basis for studying the response of the water-bearing structures in the coal mine,but also the premise of inversion methods.Based on the calculation method of separated loop configuration and the finite element method of the total magnetic field,this research realizes the 3D forward modelling of STNMR.The method can freely define the parameters of the model and the loops,so as to calculate the multi-component NMR signal of any 3D geological body,which lays a foundation for the study of the response law and the inversion method.2.Research and analysis of STNMR detection capability.In order to effectively detect the target aquifer,it is necessary to understand the detection range and resolution of STNMR.In this research,several models of aquifers at different depths were established,and the maximum initial amplitude of the signal received at different depths was calculated using the power parameter of the GMR instrument.The comparison with the SNMR signal shows that the detection depth of STNMR is increased by more than 93.33%,and the vertical resolution at 100 m underground is improved by8.32%～12.31%.3.Design and optimization of STNMR observation methods.Due to the space limitation of the mine tunnel,the size of the Rx loop must be reduced,which leads to the reduction of SNR.In this research,the underground Rx loop was designed according to the geometric characteristics of the tunnel,and the STNMR responses with different configurations of the loops were calculated.The calculation results show that the amplitude of the signal is proportional to the Rx loop area,and changing the loop orientation may increase the signal strength or reduce the pulse moment required to detect the target.Based on the above conclusions,two optimization methods for the underground Rx loop were proposed:（1）using an"unequal loop",that is,increasing the side length of the Rx loop along the direction of the tunnel;（2）carrying out multi-component detection,that is,change the direction of the receiving loops to receive the signal.4.Study on the response law of STNMR signal of coal mine water bodies.In this research,models were established based on the characteristics of several typical water-bearing structures in coal mines,and their three-component signal responses were calculated.And the changes of the response during the dynamic evolution of separated water were compared.The calculation results show that the STNMR signal reflects the spatial location and geometric structure of the water-bearing body.Analyzing its response characteristics is helpful to determine the type of water-bearing structure in coal mines.The comparison results show that the STNMR signal can effectively distinguish the separated water structures at different development stages,and it is feasible to use the STNMR method to dynamically monitor the water-bearing structures of coal mines.5.Implementation of STNMR inversion method.In this study,based on the Occam method and the deep learning,the quantitative interpretation of water content and ₂T^*relaxation time was realized,and the SNMR-STNMR joint inversion was also realized through deep learning.The inversion results show that the average RMSE of the deep learning method on the test models is 0.2057.Compared with Occam inversion,the resolution of deep learning inversion is also higher.The SNMR-STNMR joint inversion method further improves the accuracy of prediction.Compared with the SNMR and STNMR separate inversion,the RMSE on the validation set for joint inversion decreased by 1.03%and 27.52%,respectively.Using the combination of theoretical analysis and numerical simulation,this dissertation analyzed the detection capability of STNMR,summarized the signal response law of STNMR of typical water-bearing bodies in the coal mine,put forward the corresponding observation method,and realized the inversion interpretation of STNMR.This research has laid a foundation for the theoretical method,technology development,instrument development,and application promotion of STNMR.