The research on the joint application of the seismic attributes optimizing based on the genetic algorithm and the neural network methods
暂无分享,去创建一个
Genetic algorithm, grey relational analysis and BP neural network can be used to get the optimization of seismic attributions by which “Oil saturation” can be predicted accurately. Parallel searching is the distinguishing feature of genetic algorithm, and it can correct the faults of contraction of BP neural network. Therefore, the combination of GA and BP does not only improve the effect of searching, but also incorporate the study and predictive function of neural network. We can search optimum solutions in short time with this method. When the data were initialized by genetic algorithm, grey relational analysis can be used to decrease the redundant and non-valid seismic attributions data. The processing velocity can be improved. Meanwhile, the time of calculation will be saved. Weights can be computed by multiple linear regression, when the real oil saturations in well were set as the objective functions and the seismic attributions were set as the constraint equations. The resolutions solved by GA-BP optimization are better than the resolution of multiple linear regression. In practical application, relative accuracy is very important and the best solution may not be the best scheme. The best ten solutions solved by GA-BP method can be selected by different destinations, which were benefit for the ultimate decision. Practical research showed that the combination of genetic algorithm, grey relational analysis and BP neural network could offset the defects of single method and develop the strong points. This method has been proved correctly in practical application.
[1] Tian Ji-dong. THE RESEARCH ON THE APPLICATION OF THE SEISMIC ATTRIBUTES OPTIMIZING BASED ON THE GENETIC ALGORITHMS AND THE NEURAL NETWORK METHODS , 2006 .
[2] Cnpc Geophysical. Application of optimum seismic attributes in development of oilfield. , 2006 .
[3] Liu Xi-yu. Applying particle swarm optimization algorithm-based artificial neural network to performance evaluation system , 2008 .