Research Of Seismic Facies Analysis Technology Based On Three-dimensional Seismism

Along with object complex degree of oil-gas exploratory development increasing and seismic interpretation technology being more and more mature day by day, seismic oil-gas forecasting technology develops toward fine and the practical direction. In order to enhance the rate of accuracy of the oil-gas forecasting, division of the seismic facies is especially one important step, accuracy of division of the seismic facies influences the reliability of result of oil-gas exploration.This thesis has study the seismic attribute extraction and seismic attribute optimization, seismic attribute property is withdrawn from the three dimensional seismic data, attribute optimization is carried on view of these complex seismic attribute property, the useful attribute property information is selected.In order to solve the question of many excessive input variables in three dimensional seismic data,the principal component analysis method is used in this thesis.After the principal component analysis method transforms the originally bigger group of input variables by using linear transformation,it can obtain a group of new irrelvant input variables,which include majority of information of the original group of input variables. Then these few new input variables are used as the data input of the BP neural network and the SOFM neural network.The method of division of seismic facies is the key point in this thesis.This thesis has studied emphatically the method of BP neural network algorithm using the waveform analysis which divides the seismic facies and the method of SOFM neural network algorithm which divides the seismic facies, And some improvements to the BP neural network algorithm using the waveform analysis and the SOFM network algorithm are made. The experiment of seismic data has proven that the improving algorithm of BP and the improving algorithm of SOFM which divide seismic facies in this thesis, which not only has a faster speed of classification ,but also has a higher precision of prediction.The detailed steps of two algorithms are given in this thesis,the two algorithms` processes of training and forcasting are explained by two experiments separately.The extraction of seismic attribute property and two methods of neural network which divide seismic facies are taken as the base of theory, VC++ 6.0, MatLab as well as OpenGL are taken as the development kit,then a suit of system of analysis of seismic facies based on three-dimensional seismism has been achieved,which conforms to the conventionality of 3D seismic data interpretation,it realizes the display of two-dimensional seismic facies and two-dimensional seismic facies.The system is divided into several modules by using object-oriented programming technology as a guide during the process of development.The function of every module are implemented, and imitative results are presented. What is more, shortcomings of the thesis are analyzed, and some requests are put forward in the future.