Study Of Seismic Inversion Based On Dimension Reduction

As the foremost basic energy,Coal will still be one of the most important primary energy sources in quite a long time in China.But unfortunately,most mining conditions of coal mine in our country are not good,and the average mining depth has reached over 600 meters because of the scarcity of easy exploited coal.As the increase of mining depth,the rate and energy of fatality caused by excavation grow fast and apparently will bring great damage and threat to staff and equipment underground.Seismic monitoring and inversion,an online monitoring technology which is easy to implement and has better performance in engineering practices,has received more attention from scholars.Though the technology has made some achievements,it still needs improvement in many aspects,such as number of monitoring channels,data processing capacity and reliability of seismic data inversion.Based on the status of seismic monitoring and inversion technology,we studied the key technologies of seismic monitoring and wave velocity inversion,and has completed following works:1)Aiming at the shortcomings of the existing mine seismic monitoring system with a small number of channels and inconvenient layout,we studied the key technologies of distributed seismic monitoring system,proposed a new method to reduce seismic data transfer which will greatly expand the geophone number of the system and accomplished the prototype system building and associated tests.The distributed seismic monitoring system features large capacity,high availability,standard data format and online data processing.The intelligent seismometer used in this system,which transmits seismic data by an improved way based on threshold,can save large amount of network bandwidth without data incompletion and thus significantly improved seismometer capacity of distributed seismic monitoring system.The datacenter of distributed seismic monitoring system full utilizes the latest technologies such as load balancing,message queue and server cluster to obtain the ability of high availability and online data processing.2)For the ambiguity problem of far-field source,we proposed a new method with faster locating speed which can solve two-dimensional location ambiguity problem.With the improving of seismometer capacity in distributed seismic monitoring system,the distance between seismometers decrease more than it was,the hypocenter will be more likely to locate in far field area which might lead to location ambiguity problem due to data precision of computer.For this problem,we proposed a new locating method based on model space dimension reduction,numerical experiment shows that,the new method is fast,convergent and stable while the signal-to-noise ratio of arrival time is greater than 80 dB.3)We proposed a new 2-D FFT based model space dimension reduction method for seismic wave velocity inversion to resolve convergence problems in space domain inversion,and the numerical experiment,which is conducted in 50*50 grid and with 99% model space dimension reduction,shows that the method can give steady result which is similar to theory result.Contrast experiment shows that the result gets by frequency domain inversion is much better than space domain inversion under same experimental conditions which further illustrate the advantage of the new method.4)We studied a new method to add stress constrain for wave velocity inversion,which will reduce the size of model space and thus reduce multiplicity of inversion to improve the reliability and stability of inversion.Numerical experiment shows that the wave velocity inversion with stress constrain can give wave velocity distributions before and after stress change,which is similar to theory results.Contrast experiment shows that the added stress constrain lead to 30.6% reduction of average errors between calculated results and theory results and 84.1% reduction of variance of errors between calculated results and theory results which illustrate the enormous impact of stress constrain in wave velocity inversion.