Wyznaczanie przepuszczalności skał na podstawie różnorodnych parametrów petrofizycznych z wykorzystaniem sieci neuronowych

Artificial Neural Networks have been applied for determination of permeability of sandstones. The set of data published in [Sen et al, 1990] were used for testing different configuration of parameters. Poro sity, bulk density, m-factor and longitudinal constant of relaxation NMR-T1 were enclosed into computations. Simple feed-forward networks, trained with back-propagation method have been used. ANN proved to be more efficient tool than statistical approach for determination of the permeability reservoir parameter owing features of fuzzy data. The OS factor rate of statistical deviation of error of output data computation (permeability) to statistical deviation of output data worked as a quality of solution measure.