Research Method of Fuzzy Neural Network of Reservoir Heterogeneity

Abstract

For analysis of the reservoir, due to complexity of the reservoir heterogeneity, the relationship between the logging response value and the non-homogenous reservoir is complex and fuzzy. The use of fuzzy clustering method of neural network combines the adaptivity and the fault tolerance of the neural network organic manner with the fuzzy comprehensive discrimination in the human thinking simulation of fuzzy logic, so as to recognize the reservoir heterogeneity through multi-factor fuzzy comprehensive judgment and reasoning. This method has been used to treat the data about the 135 wells in Xujiahe Group in central Sichuan, and has shown good results, with a coincidence rate as high as 87.6%. The result demonstrates that this method enjoys good adaptivity against the problem of automatic recognition of the reservoir heterogeneity, and could improve the precision of automatic recognition of the reservoir heterogeneity.