Porosity prediction based on LS-SVR: A case study in Shuangcheng depression of the Northern Songliao Basin

Porosity is a key indicator for evaluating reservoir quality. Porosity analysis identifies the type, structure, and distribution of reservoir pores, essential for evaluating oil and gas accumulation and reserves. Accurately predicting porosity is crucial for petroleum exploration and engineering in the Shuangcheng Depression, Northern Songliao Basin. Traditional methods, such as core sampling, are often limited by high costs, time constraints, and the need for discrete samples that may not fully represent the reservoir, thereby hindering accurate porosity prediction. Therefore, this study assesses the accuracy of porosity prediction using the Least Squares Support Vector Regression (LS-SVR) model, selected for its effectiveness in handling small datasets and capturing nonlinear relationships. LS-SVR also mitigates computational challenges associated with traditional Support Vector Regression (SVR). The model utilizes geological and geophysical data from the Shuangcheng Depression in the southeastern fault zone of the Northern Songliao Basin to predict reservoir porosity. Nine well-logging data are used as input features, with porosity values obtained from core samples serving as the target label. This study develops an optimal porosity prediction model by training it with a sigmoid function, optimizing the penalty factor C and kernel parameter γ via grid search, and selecting the best parameters through 5-fold cross-validation. To ensure the model's performance, statistical metrics are used to evaluate the model. Evaluation results show that the model achieves an R2 of 0.90 on the test set, explaining 90% of the variance in the target variable. Compared to traditional methods, the LS-SVR model demonstrates a significant improvement in porosity prediction. The remaining metrics include MAE (0.55), MSE (0.40), and RMSE (0.63). The results indicate that the LS-SVR method significantly improves the prediction of reservoir porosity in the Shuangcheng Depression of the Northern Songliao Basin. It is crucial for reservoir evaluation and petroleum engineering decision-making, providing valuable references for further research and practical applications.