Grey Support Vector Machines for Porosity Prediction in Carbonate Reservoir

Abstract

This paper introduce the basic theory of grey incidence degree and support vector machine(SVM), then propose an prediction method based on degree of grey incidence and SVM, that is grey support vector machine(GSVM). Firstly, we test the interpolation and extrapolation ability of GSVM through the sine function. The results show that GSVM has good interpolation ability, but poor extrapolation ability. Secondly, in the analysis of seismic rock physics model and basic characteristics of carbonate reservoirs, we build rock physics model of carbonate reservoir and calculate the seismic attributes of synthetic seismograms. Then we calculate the degree of grey incidence between seismic attributes and porosity in carbonate reservoir, and use the GSVM to learn and train. The change trend between the prediction and actual value is basically the same, the maximum relative error is only 5%. The application of GSVM in the carbonate reservoir model shows that the method can be used to predict the carbonate reservoir porosity, and the result is reasonable and effective.