Research On The Method Of Applying Logging Data To Analyse The Micropore Structure

With the oil-field entering high water cut stage of late development, the integrated water cut of oil-field content is more than 90% at most cases .The remaining oil distribution is more fragmented and more complex. While clearing the distribution of remaining oil is one of the main topics to maintain high and stable yield for oil-field. The micro-pore structure of the reservoir layers is fundamental reason in various factors that affect the remaining oil distribution.The study of Micro-pore structure are mainly restricted to the high cost of experimental methods at present, and core data is very limited, so that many results of research limited to theoretical research, and can not meet the research of the oil field scope. While logging data is the most readily available and relatively wide, and can reflect the continuous nature of stratum. If log data can be used to identify micro-pore structure of reservoir, then results of laboratory studies can be extended to the actual study area. A significant change will implement from theory to practice. It will be a low-cost, functional, rapid evaluation method. It is of great significance for the latter part of the oil field development to maintain stable.In this thesis, a method for identifying the type of pore structure is proposed based on the dealing with the log data by neural network technology. Neural network model can realize the transform of complexity function through a combination of simple nonlinear function, and the learning function, adaptive capacity, memory capacity, association memory and the way dealing with information of neural network model can be used as a kind of method for log interpretation. In the process of log interpretation, it does not need setting up prior logging response equation or providing empirical formula for the neural network model. So, it avoids the human factors when selecting parameters for the log interpretation. It is a new way for log interpretation.It is the search for a nonlinear mapping or fitting between logging information and pore structure parameters, that using neural network to predict the type of pore structure. That is, getting an interpretation model from a study of given sets of training samples, then predicting the pore structure type of unknown samples. The practical application proved that it will have a good result that using BP neural network to predict the pore structure types of reservoir.