Quantitative Prediction of Tectonic Coal Thickness Based on FNN and Seismic Attributes

Coal and gas outburst is one of main dynamic disasters of coal mine in China. As latest studies, the development of tectonic coal is one of essential preconditions for coal and gas outburst, and an accurate thickness prediction of this coal will play a positive role in safe mining. Unfortunately, few researches on quantitative prediction of tectonic coal thickness, especially the methodology of seismic attributes, have been carried out. The technology of seismic attributes is one of the most important methodologies which can reveal the difierence between normal coal and tectonic coal from earth surface, but there are too many factors which can afiect seismic attributes. Therefore, the relationship between tectonic coal thickness and seismic attributes is non-linear. FNN (Fuzzy Neural Network) has been used in many industries for its characteristic of simple structure, strong approximation ability and non-linear processing ability. We combined FNN and seismic attributes together to quantitatively predict the thickness of tectonic coal in coal bed. In practical forecasting, flrst of all, we tested and obtained main parameters of FNN model and seismic attributes using forward synthetic data. Then, we extracted and regenerated a new training set of true seismic attributes such as 90 Hz spectral decomposition and sweetness from wells’ nearby traces. Through retraining of this new set, we built a practicable prediction model of tectonic coal thickness, and used it to predict tectonic coal thickness in research area. By compared to true values uncovered by wells and geological disciplines, the prediction’s reliability is high and the prediction’s errors meet actual requirements of coal mining.

[1]  Mark Chapman,et al.  Application of spectral decomposition to detection of dispersion anomalies associated with gas saturation , 2006 .

[2]  John P. Castagna,et al.  Layer-thickness determination and stratigraphic interpretation using spectral inversion : Theory and application , 2008 .

[3]  Li Ming Wu,et al.  Research of FWP Process Deformation Compensation Forecasting on the Basis of TS-FNN , 2011 .

[4]  Kazuo Tanaka,et al.  Stability analysis and design of fuzzy control systems , 1992 .

[5]  Michio Sugeno,et al.  Fuzzy identification of systems and its applications to modeling and control , 1985, IEEE Transactions on Systems, Man, and Cybernetics.

[6]  Zhang Zhigang.  Research on gas irradiation feature of tectonic coal , 2011 .

[7]  Zhang Hai-rong,et al.  Dynamic Monitoring on Land Use Change of Xuzhou Urban Area Using Satellite Remote Sensing Data , 2004 .

[8]  Fei Xia,et al.  Condenser Fault Diagnosis Based on FNN and Data Fusion , 2010 .

[9]  G. Partyka,et al.  Interpretational applications of spectral decomposition in reservoir characterization , 1999 .

[10]  M. Chapman,et al.  AVO analyses and spectral decomposition of seismic data from four wells west of Shetland, UK , 2008 .

[11]  M. Ahmad,et al.  Application of spectral decomposition and seismic attributes to understand the structure and distribution of sand reservoirs within Tertiary rift basins of the Gulf of Thailand , 2012 .

[12]  K. Marfurt,et al.  Narrow-band spectral analysis and thin-bed tuning , 2001 .

[13]  Andy Chadwick,et al.  Quantitative seismic analysis of a thin layer of CO2 in the Sleipner injection plume , 2012 .

[14]  Lu Xin Tang,et al.  The FNN Quilting Process Deformation Prediction Model , 2010 .

[15]  Bruce S. Hart,et al.  Channel detection in 3-D seismic data using sweetness , 2008 .

[16]  Qiang Ma,et al.  The Newly AND-OR FNN Modeling and Application , 2012 .

[17]  L. K. Rixon,et al.  Outbursts and geological structures in coal mines: A review , 1981 .