Application Of BP Neural Network In Logging Interpretation

The research on the relationship in lithology, electrical, physical properties and oil content is the core of logging data processing and interpretation. It is based on the analysis and test results of various cores conbined with regular logging data and geological background. The primary research methods are interactive plot analysis, statistics and regression. As the method application of BP neural network in logging interpretation, this paper uses BP artificial neural network to predict some parameters, such as the reservoir physical parameters, lithology and oil content.First, this paper explains the overview of artificial neural network and its application in oil exploration field. Then, this paper focuses on the BP neural network in the aspects of the principles, improvement and optimization. Next, based on the actual logging and other relevant studies, conclusive data in some oil area, this paper details to use BP neural network to predict the methods of reservoir lithology, oil content and physical parameters, and gives the training methods respectively, such as modeling, data preprocessing, learning sample selection and the network; finally, shows the comparison and analysis between the predicted results and the actual results.The results show that, as a nonlinear method, artificial neural network can better simulate the actual stratum, and better establish the nonlinear relationship in four properties. The comparison of network predicted results and the analysis and test results shows, BP artificial neural network are feasible and effective in the predicted application in sand-shale reservoir parameters, lithology and oil content in conventional and unconventional conditions. Especially in the prediction of sand-shale reservoir parameters in low resistance, high radiation and other unusual conditions, BP neural network method is used for a good prediction. There is a great reference in the exploration and development of oil field.