Well logging curve reconstruction based on kernel ridge regression

The logging curve is the most basic and important part of the petroleum industry. It plays an incomparable role in identifying of rock and oil-gas layers, and analyzing the geological structure of the formation. It also can be used to calculate porosity, permeability, and saturation. The quality of logging curve is the premise to ensure the reliability of logging interpretation results. However, in the actual application of logging data, it is often occurred that the logging curve is distorted or missing in some well segments due to tool measurement or wellbore reasons, which affects the accuracy of logging results. At present, the conventional linear fitting and statistical analysis has been difficult to satisfy the requirements for ultra-fine reservoir analysis and evaluation. Kernel ridge regression is a multivariate nonlinear regression analysis method. It combines kernel function and least square regression analysis. This method is used to reconstruct the acoustic curve of 20 wells in the study area and explain it finely. The results show that the kernel ridge regression method can be used for multivariate nonlinear regression analysis. The curve predicted by this method has high accuracy and is worthy of popularization and use in oil fields.