College of Earth Sciences, Jilin University, Changchun 130061, China

To improve the accuracy of lithology identification on well logs, the data of cores and well logs are used to summarize the logging response of the glutenite in Wuxia area. Five well logs including AC, CNL, DEN, GR and RXO are selected for the testing. The SVM lithology identification model is built by optimizing the SVM kernel parameter σ and penalty factor C with genetic algorithm. The result shows that the actual data forecasting coincidence rate of the model is 81.6% generally, and the accuracy of GA-SVM is obviously improved, compared with traditional identification methods of glutenite.