ARTIFICIAL NEURAL NETWORK PATTERN RECOGNITION OF FRACTURES IN THE BEDROCK RESERVOIR OF THE QIJIA BURIED HILL
SUN Si-min1, ZHU Qing-hong2, PENG Shi-mi1
1. Faculty of Resource and Information, China University of Petroleum, Beijing 102249, China;
2. Jinzhou Oil Plant, Liaohe Oil Company, Petrochina, Jinzhou 124000, China
Based on lithologic logging interpretation, the authors identified typical fracture sections according to their logging response calibrated by core data integrating drilling mud leakage, drilling break and production data. Then the typical logging response of fracture section was employed as training samples for Artificial Neural Network Pattern Recognition (NNPR). All the 76 wells in the Qijia buried hill were processed by applying the ability of NNPR including paralleling process, distribution information storage, powerful self-study and automatic weight value adjustment. Finally, a new fracture prediction method integrating core, conventional logging, test and production data was formulated, which has been proved to be effective by drilling.
［1］ 谭廷栋.裂缝性油气藏测井解释模型与评价方法［M］.北京:石油工业出版社,1987.
［2］ Rogers S J.Determination of lithology from well logs using a neural network［J］.AAPG,1992,76(5):731.
［3］ Richard P L.An Introduction to computing with neural nets［J］.IEEE ASSP Magz,1987,(4):4.
［4］ Derek H.Comparative study of backpropagation neural network and statistical pattern recognition techniques in Identifying sandstone lithofacies.Texas A&M University,College Station,1990.
［5］ 罗利.神经网络在测井解释中的应用［J］.天然气工业,1997,17(5).
