Reservoir Parameter Prediction Of Application Based On Support Vector Machine

It is an important part of oil-gas exploration to make use of the earthquake material for the reservoir parameter prediction. Therefore when using the earthquake material to enhance the reservoir parameter prediction precision and the reliability, we should try our best to excavate and make full use of the earthquake and the oil well logging information for the complex reservoir oil-gas field exploration development.This article based on the research of earthquake attribute, elaborates that the earthquake attribute optimizes some basic questions. It adopts the correlative coefficient formula method of multi-variants statistics to choose optimally earthquake attribute parameter which is more sensitive with the porosity, and makes use of the principal components analysis to optimize the earthquake attribute parameter and achieve the goal of character reduction. In the meanwhile, it eliminates the redundant information caused by the correlation of the earthquake attribute. Further more, the optimal result is used for reservoir parameter prediction.Taking the Jialing River group reservoiras in the southeastern part of Sichuan province for an example, we discovered that in this area the reservoir sample data quantity is less by the scrutiny of reservoir geology characteristic. If we use traditional forecast method, it will difficult to achieve the ideal effect. This article tries to apply support vector machine model for small sample study to the reservoir parameter prediction. And the result of experiment indicates that support vector machine method based on the principal components analysis, compared with the method of traditional reservoir parameter prediction, and has obtained a better effect. According to the reservoir parameter prediction of the unknown area in the operating region with this method, and combining the surface and spatial distribution characteristics of the reservoir parameter in the operating region protracted by the Kriging estimate technique, the advantageous exploratory destination district of the reservoir beds is ensured.