Research On Evaluation Method Of Trap Effectiveness Based On Deep Learning

Trap effectiveness evaluation has always been one of the important tasks of petroleum exploration.There are many existing trap effectiveness evaluation methods,such as fuzzy mathematics,gray system theory,expert scoring,etc.,but these methods rely on the experience of geological personnel.This paper turns the problem of trap effectiveness evaluation into a classification problem in deep learning,and proposes an intelligent trap effectiveness evaluation method based on convolutional neural network(CNN).The Luojia area of Shengli Oilfield was selected as the experimental work area to study two main types of traps in the Luojia area: structural-lithologic traps and stratigraphically unconformable traps.An evaluation system for two types of traps was established,and the distribution and accumulation characteristics of oil and gas reservoirs were studied.The two types of traps were geologically dissected,and the corresponding geological elements were extracted according to the evaluation system.A7-layer convolutional neural network was built,the two types of trap data were sorted out,a reservoir database of oil and gas reservoirs was established,a convolutional neural network was trained,and applied to the target area to evaluate the effectiveness of traps.