| The hydrogeological conditions of coal mines in China are complex,and there are many kinds of water-conducting channels(collapse columns,goafs,faults and fissures,etc.),which lead to the occurrences of coal mine water hazards frequently.As one of the most widely distributed hidden hazards in Coal mines,collapse columns have the characteristics of strong concealment,high sudden occurrence and large water inrush,especially serious for coal mine production.Therefore,it is of great significance to accurately identifying the collapse column and its water conductivity for ensuring the safety of coal mine production and prevention of water hazard in coal mines.At present,seismic exploration and electromagnetic method are the main methods of the collapse column detection at home and abroad.Seismic exploration,especially 3D seismic exploration,has high detection accuracy for geological structure,and can accurately identify the spatial position and shape of collapse column.However,this method cannot accurately determine the water-richness of the collapse column;while electromagnetic method is sensitive to low resistance anomalies and can effectively detect water-richness of collapse column.But the method has a low resolution and cannot accurately delineate the boundary of the collapse column.That is to say,the use of either seismic exploration or electromagnetic methods alone can not detect the location and water abundance of collapse column precisely.Therefore,it is necessary to carry out joint inversion by effective parameter coupling method,overcome the limitation of single method,reduce the multi-solution of inversion,and realize the precise location of collapse column.Firstly,the finite difference form of staggered grid is derived based on the first-order velocity-stress elastic wave equation.Then,based on the geological conditions of the coal mine in northern China,the model of-the collapse column is established.The seismic wave field response characteristics of different models are analyzed.The results show that:(1)with the decrease of the diameter of the collapse column and the increase of the burial depth,the amplitude of the reflection wave at the top and bottom of the collapse column becomes weaker.When the diameter of the collapse column is small,the reflected P-wave appears as a "smoke ring" in seismic records.(2)For a 250m depth collapse column studied in this paper,when the ratio of diameter to depth is less than 8%,three-dimensional seismic exploration method can effectively identify the top interface,but it is almost impossible to identify the bottom interface.Secondly,the forward of 2.5D and 3D controlled source audio magnetotelluric(CSAMT)is realized by using finite element method.The correctness and accuracy of the program are verified by solving the layered model and the subsurface anomalous body model with low resistance,and comparing them with the analytical solution and the integral equation method.Then the numerical simulation of the collapse column model in the coal-bearing strata is carried out.The characteristics of CSAMT response of collapse column model with different diameter,different buried depth and different resistivity are analyzed emphatically.The results show that:the abnormal response of collapse column decreases rapidly with the decrease of its diameter and the increase of its burial depth,and the resolution ability of CSAMT decreases gradually with the decrease of its diameter and the increase of its burial depth.The nonlinear joint inversion algorithm of seismic and CSAMT is studied based on generalized cross-gradient.The basic principles of artificial bee colony(ABC)algorithm are studied,including the basic behavior,algorithm flow and main parameters of ABC algorithm.Aiming at the problems of excessive randomness and easy to fall into local optimum,the improved ABC algorithm is put forward in the aspects of randomness,the formula of profitability and the formula of new food source.By using the standard test functions,the optimization results of the improved ABC algorithm are compared with those of the particle swarm optiruization(PSO)algorithm and the standard ABC algorithm,and the optimization ability is evaluated.In order to solve the problem of parameter coupling in different geophysical methods,the cross-gradient function is added to the joint inversion objective function.And the definition,discretization and properties of the cross-gradient function are studied.In view of the problem that the cross-gradient can not be calculated due to the lack of quantitative information such as seismic velocity or impedance in seismic interpretation profiles,the cross-gradient method is improved by using the idea of clustering.The modified cross-gradient method is proposed and the algorithm flow is given.The correctness and feasibility of the method are verified by comparing the traditional cross-gradient method with the modified cross-gradient method for two-dimensional velocity and resistivity model.On the basis of seismic and CSAMT forward modeling,the results of individual inversion of CSAMT,joint inversion based on accurate seismic interpretation,joint inversion based on "inaccurate" seismic interpretation and joint inversion based on muti-cluster seismic interpretation are compared and analyzed for collapse column model.The results show that the joint inversion is significantly better than the individual inversion,which can effectively improve the inversion accuracy.In the end,combining with the detection example of a suspected collapse column in a coal line in Shandong Province,the 2D inversion of CSAMT,2.5D inversion of CSAMT and joint inversion are compared.The results show that the joint inversion can make the low resistance anomaly at the location of collapse column show obvious difference with surrounding rock,and improve the lateral positioning accuracy.Joint inversion results show more prominent performance in detail,indicating that the joint inversion algorithm proposed in this paper has good applicability. |
PDF Full Text Request