Shale reservoir is characterized by physical properties of low porosity and low permeability,and fracturing treatment is often required in shale gas production.The fracability of shale reservoir can be evaluated by brittleness index(BI),and the mineral composition method is the most popular method for calculating rock BI.Based on the mineral composition method calculated core BI value,the non-linear relationship between logging parameters and shale BI has been analyzed by applying the self-learning ability of BP neural network,and then the cuckoo(CS)algorithm with global optimization ability and stability has been used to improve the prediction accuracy and stability of BP neural network,so as to establish a shale BI prediction model based on CS-BP neural network.The CS-BP prediction model has been used to predict shale BI values in wells Y1 and Y2 in the study area,which indicates that CS-BP prediction values have a basically consistent trend with core BI values,showing a good prediction result as a whole.The study results indicate that the cuckoo(CS)-BP neural network based method for rapidly calculating shale BI by using logging data has certain practical value in the study area.
关键词
可压裂性/页岩储层/BP神经网络/CS—BP算法/页岩脆性指数
Key words
fracability/shale reservoir/BP neural network/CS-BP algorithm/shale brittleness index
维普
万方数据