Research On Data Mining Method Of Well Logging Reservoir Evaluation

With the increasingly developed petroleum exploration and computer technology,a huge amount of exploration and development data has been accumulated in the major oil fields in China at a rapid pace today.In this situation,we need to dig out the potential value from the massive data and creates a new benefit.Therefore,the application of data mining technology in the oil and gas exploration field is one of the technologies needed to be realized.In this paper,we take the Nm III segment of the seven or eight fault block in the Gang Dong field as a research object.Use data mining method to establish the accurate reservoir parameters prediction model and lithology,fluid identification model.Besides,we reviewed old wells of complex reservoir for more potential reservoir exploration which can provide reference and suggestion for the oil field development in the middle and late stages of production.First of all,according to the geological survey of the study area,the relationship of the four regional reservoir characteristics is analyzed based on the core data and well logging curves.On this basis,data mining algorithm is used to establish the model to predict the reservoir physical property and oil-bearing possibility of the interests.Based on the data of core analysis,oil test and initial production data,well log interpretation data,the reservoir lithology and fluid identification data sets are used to establish their charts by rock physical method.At the same time,the applicability of each data mining algorithm for the standard data sets is analyzed.The similarity between data sets is determined by calculating the similarity between the feature vectors of the data sets.Then this paper recommends suitable algorithm for lithology and fluid identification data sets.Due to the own advantages of data mining modeling method in multidimensional data analysis and data feature classification,the accuracy and efficiency of new lithology and fluid identification model are significantly improved by the comparison with the conventional chart.Meanwhile,the data mining algorithm is used to build a model to study other wells in the study area.It has been proved that the data mining method can effectively identify the potential oil reservoirs which were not discovered before.Meantime,the recommended algorithm has a good application which provides an effective technique for complex reservoir review.