Fractures in carbonate reservoirs are the migration channels and reservoir space of oil and gas,and fracture prediction has important guiding significance for oil and gas exploration,development and evaluation.A GWO-CS-BP algorithm is proposed to solve the problem of fracture detection in carbonate reservoirs in the study area.The algorithm combines GWO-CS(grey wolf-cuckoo search)and BP(back propagation).Coherence,curvature,dip angle,azimuth angle and configuration tensor are used as the input data of a GWO-CS-BP neural network,which is constrained by logging and geological data.An evaluation index is thereby ob-tained to evaluate and grade fractures in the study area.The detection results show that the GWO-CS-BP algorithm can integrate the characteristics of each attribute for secondary error control on fracture detection.As per the evaluation index fs obtained,frac-tures in the study area could be classified into three grades in four zones.For the fs lies in the range of 4.0 to 5.8,which indicates a moderate degree of development,area C with many high-yield wells is most conducive to oil and gas accumulation.Based on the e-valuation index fs,the modified BP neural network by a GWO-CS algorithm yields a detailed evaluation of fractures in the study area.
国家科技期刊平台
NSTL
万方数据
