neural network training data seelction using memory reduced cluster analysis for field model development tank

摘要

Abstract: the present invention relates to a method and predict the needed data by selecting a data channel training samples input data of geophysical geophysical multidimensional. The input data is a set of data generated from a sounding logging neutron pulse tool at multiple points and depths in a well Baltksyt cased processor. The data may be required in response to a probe logging tool open wells.The inputted data are divided into groups (20-24, XVI). The required data of actual well directed training well clusters to the cluster are analyzed to find differences. Clear 34), etc.. The results are used. Fuzzy inference) to select a part of Lklmjmwt (36a) to be set to extend training. The group used shorthand to model, for example artificialneural network lines of artificial neural network.Driven model could be used to produce measurements of Sbraftradyt synthetic logs aperture hole exposed in response to input data of the sounding hole cased Baltksyt processor.