Research On Prediction Of Coal And Gas Outburst Risk Area Based On Seismic Forward Modeling And SVM

Coal and gas outburst refers to a complex kinetic phenomenon that gas is given off in great amount from coalbeds in the mining process in underground workplace of coal mine. The outburst has a direct impact on each and every aspect of the production of coal mining, and poses a huge threat to safe production and life of the miners. Quantitative ascertaining and predicting of coal and gas outburst area is a critical problem that needs to be addressed at present time in the process of coal mining, and it is also an important concern and goal for building digitalized mines.Supported by such programs as the National Natural Science Funds and taking into consideration the application requirement of coal mines, this dissertation is centered on the discussion of quantitative prediction goal of coal and gas outburst risk area, analyzes the evolving properties of such information as waves (mainly seismic waves) and fields (data of gas content, coal thickness, depth of burial, and gas pressure) of coal and gas outburst risk areas from the temporal and spatial perspectives, and finally reveals the coupling relationship between seismic properties, geological data variation and gas outburst risk areas. Adopting the methodology of seismic modeling data and building the forward seismic section of coal-and-gas geological model, the current dissertation extracts, optimizes and reduces seismic attributes, on the strength of which the non-linear relationship between reductions set of seismic attributes and gas bearing capacity is studied quantitatively by using Support Vector Machine (SVM) algorithm. Hence, the prediction is made of the coal and gas outburst risk area, and feasibility and validity of predicting coal and gas outburst risk area is approached by referring to post-stack seismic data. The research findings achieved in this dissertation are mainly as follows:Several typical geological and geographical models on gaseous coalbeds are constructed. On the basis of geological and drilling data, this research employs Effective Medium Theory respectively by Backus and Hudson to compute the physical property parameters of different coalbeds respectively with regards to their structures. Hence, the typical geological and geophysical model is constructed under the condition of gas concentration.Based on seismic forward modeling, seismic property technology and Support Vector Machine, research methodologies of predicting coal and gas outburst risk areas are proposed and successfully put to practical prediction. First, the finite difference algorithm is used to construct several typical forward seismic sections; the property of reflective waves on the seismic section coalbed is then analyzed, thus the corresponding seismic properties are obtained. Second, the seismic properties are reduced by using rough set algorithm, and the main seismic property set is determined of influencing gas outburst. Finally, the non-linear relationship between reductions set of seismic properties and gas bearing capacity is studied quantitatively by using Support Vector Machine (SVM) algorithm; by using field data from One Coal Mine of Jincheng Mining Area, the proposed methodologies are verified. The findings indicate that the methodologies proposed in this current dissertation are highly reliable and practicable, and provide a new approach to predict gas outburst risk area by having recourse to post-stack seismic data.The positivistic research is undertaken on the predictoin of coal and gas outburst risk areas by taking the deep mining area in Huainan Mining Area as an example. To start with, the geological and geophysical model of the research area is built, and the seismic forward section is then generated by utilizing finite difference algorithm. Through the analysis of seismic properties and seismic spectral decomposition, and the grouping of seismic properties and geological properties associated with seismic properties, three predicting models are established on gas risk areas to test and compare their efficiency. The result shows that the model for predicting coal-and-gas outburst risk area, which is constructed by combining the two geological parameters of coalbed thickness and burial depth with seismic forward simulated property data, is the more practical, time-effective and operable. Finally, the GIS platform is used to present the achievements for predicting coal and gas outburst risk area.This dissertation has 71 diagrams and graphs, 28 tables, and 102 reference entries.