Apply Ridge Regression Model to Predict the Lateral Velocity Difference of Tight Reservoirs

Accurately predicting the lateral wave time lag is the basis for calculating rock mechanics parameters in drilling and hydraulic fracturing construction design, but there is a large error in predicting the lateral wave time lag using empirical formula method and multiple regression method in water layer, oil layer, and porous dry layer. This error cannot meet the subsequent construction requirements. In order to improve the accuracy of predicting the lateral wave time lag, a ridge regression model for predicting the lateral wave time lag is constructed. The Pearson correlation coefficient method is used to determine the correlation between each logging curve and the lateral wave time lag and the collinearity between each logging curve. The random forest algorithm is used to rank the influence factors, and the vertical wave time lag, bulk density, and compensated neutron logging curves are selected as the input parameters of the lateral wave time lag prediction model to reduce the complexity of the model. Finally, a ridge regression algorithm is used to establish a prediction model of the lateral wave time lag based on the logging data of five wells in WQ block. The results show that the ridge regression model takes into account the influence of porous fluid on the lateral wave time lag and can effectively extract the lateral wave characteristics of oil and gas reservoirs, water layers, and porous dry layers. Compared with other fitting methods, this method has a higher prediction accuracy, with a relative error mean of 2.59%. The ridge regression model can reflect the changes in the reservoir characteristics along the vertical direction, highlighting the differences in rock mechanics parameters under different geological conditions, and providing accurate rock mechanics parameters for subsequent calculation of geostress and drilling and hydraulic fracturing construction design.