The Research And Application Of Lithology Identification Method Using Well Logging Data In Hailar Basin

The Volcanic hydrocarbon reserves, as a new field of hydrocarbon exploration, has greatly attracted the attention and regard of the geologists in petroleum enterprise worldwide. Based on the exploration reality of the hydrocarbon reserves, this paper develop the method research on the lithology identification according to the emphatic domain in development and exploration in Daqing oil field recently and the difficult problem of the reservoir evaluation in Hailar Basin.Because the formation cause of hydrocarbon of pyroclastic rocks is very complex in Hailar Basin, pyroclastic rocks intergrowing with lavas and sedimentary rocks, there have more than 40 kinds of geological name. Based on the circumstances the paper concluded the pyroclastic rocks into three classification and 12 main kinds from the angle of lithology identification using well logging data. After the core data deled with in advance, using the core data graduating the well logging data, the paper drew the logging curves characteristics of lithology.In the course of establishing the method of lithology identification this paper tried to use cross-plot and spider web plot technique but the result of them is not well which only identified the classification and showed the trend of change. So the algorithm of Neural Networks and Support Vector Machin belonging to the model identification combined with programming were adapted to the automatic judgment. The effect of the two algorithms is well. The average of correct rate is over 84 percent. The result satisfied the need of exploration and development basically. The two methods can be regarded as the main identification method applied to the real production.The result of lithology identification is initially applied into reservoir evaluation of physical property and hydrocarbon potential identification. The calculated porosity is compared with core porosities. The average absolute error is 1.6 PU achieving the higher precision. The fitness rate of the fluid characters identification achieved to 79.5 percent matching the level of the real production.The two methods of model identification have achieved good effect in the lithology identification of Hailar Basin. As the development of research and the perfecting the networks structure in the course of popularization and applying, the technique will make great effect in the complex pyroclastic rocks research of the hydrocarbon in Hailar Basin and others.