Abstract: In the southern part of Songliao Basin, Chaganhua Area, the lithology of the Huoshiling Formation volcanic rocks is complex and variable. Traditional methods such as two-dimensional intersection and step-by-step classification based on conventional well logging data are difficult to accurately identify the lithology of volcanic rocks. To address the issues, a proposal is developed to use machine learning algorithms for intelligent identification of volcanic rock lithology. By observing sample cores and thin section analysis, the lithology of volcanic rocks in the sampled section is determined. The logging data set of the coring section is divided into training set and test set. The training set is used to match the object function, and the test set is brought into the model to predict results, and use integrate models with ensemble learning to conduct blind well prediction.The fusion model establishes a quantitative mathematical relationship between the characteristics of each well log curve, integrates the characteristics of multiple machine learning, and improves the learning efficiency of the model based on accurate lithology data set labels.The results show that the prediction accuracy of the integrate model for blind wells achieves 95.10%. The model has wide applicability, which can accurately identify and predict the lithology of volcanic rocks. This study can provide support for the intelligent exploration of volcanic rock oil and gas.

Key words: volcanic rock, lithology, machine learning, integrated learning, GBDT gradient decision tree, Songliao Basin