关键词: 火山岩, 岩性识别, 元素俘获谱测井, 电成像, 神经网络

Abstract: Deep volcanic reservoir in Xushen Gas Field is featured with complex and multivariant lithology and fluid composition, low crystallization degree and similar acidic rock composition. Effect of rock framework on resistivity exceeds that of reservoir fluids. Therefore, there is lots of difficulty in fine logging evaluation of volcanic reservoir, while precise identification of volcanic lithology is the basis and keystone for further research. Aiming at the difficulty of lithology identification in this area, it makes full use of advantages of Elemental Capture Spectroscopy (ECS), electric imaging, nuclear magnetic logging data on lithology identification, produces the knowledge of lithology identification by combining composition with structure, and determines the classification system of volcanic rock. Through utilizing TAS graph, graph mode and neural network methods, well logging lithology identification of volcanic rocks is realized, and it lays a firm foundation for fined logging evaluation of volcanic reservoir.

Key words: volcanic reservoir, lithology identification, Elemental Capture Spectroscopy, electric imaging, neural network