Seismic Reservoir For Intelligent Information Processing Method Study

Seismic reservoir prediction is one of the main tools in reservoir description, and has become a core in reservoir geophysics. With the gradual increase of seismic attributes calculated from seismic data, the problem of seismic-attributes optimization, which is related closely to seismic reservoir prediction, has become one of the problems to be solved in oil & gas exploration and development. As far as the problem above is concerned, this paper reviews the methods of seismic reservoir prediction which plays a leading role in the industiy application, discusses the computing methods of seismic attributes and shows the geophysical implication of seismic attributes. In this paper, the author introduces Genetic Algorithm(GA) for the first time, and realizes seismic attribute optimization in seismic facies recognition with Kohonen network. The paper relates briefly the principle of seismic attribute optimization in predicting reservoir parameters with seismic data. And also, for the first time ,the author applies Complete Utilization of Sample Information (CUSD network to the seismic reservoir parameters prediction, establishes the evaluation standard on seismic attribute groups; and introduces creatively the combination GA with CUSI so as to realize seismic attribute optimization in the seismic reservoir parameters prediction with function approximation method. Aiming at different areas and different reservoirs, the paper introduces primarily the method of optimizing seismic attribute group with tactics by means of automation, or expert-automation combination. Furthermore, this paper first introduces the decision method based on Rough Set (PS) theory in seismic reservoir prediction, which provides a new method for seismic pattern recognition. It puts forward an optimization rule of seismic attributes discretization; points out creatively that the combination of PS theory with Kohonen and BP network to realize the method of optimizing seismic attribute group most sensitive (or most effective, most representative) in oil & gas prediction, creates a method of seismic attribute optimization independent of particular pattern recognition method and a set of systematic method of seismic attribute optimization adapting to different areas, different reservoirs and oil & gas fields. In addition, the paper advances a method of seismic attribute optimization combining expert knowledge with search optimization; proposes a feasible method predicting permeability based on combining function approximation with seismic attribute optimization, and makes a comparison among methods of Seismic attribute optimization. Finally, in this paper, the author introduces Gmbbs method and t-test method of mathematical statistics, distinguishes abnormal values in velocity spectrum data with combination of merits of the two methods, and then rejects the abnormal values. The author advances a 3-D inversion method of layer-by-layer interval velocity suitable the 2-D area and the large dip area. At the same time, the average velocities and RMS velocities in normal ray direction and plumb direction can be computed ~th the interval velocity. Furthermore, the error between the interval velocity inverted by the method and the velocity in well is also analyzed, and the correction method is also put forward corresponding to the error. On the basis of analyzing the causes with which there are mis-tie in cross-line and considering the law of velocity change, the author raises the method of optimum-approximation average velocity with multi-segment polygonal...