ZASTOSOWANIE SZTUCZNYCH SIECI NEURONOWYCH DO WYZNACZANIA PRZEPUSZCZALNOŚCI SKAŁ NA PODSTAWIE DANYCH OTWOROWYCH Z REJONU DZIKÓW–WOLA OBSZAŃSKA W PÓŁNOCNO-WSCHODNIEJ CZĘŚCI ZAPADLISKA PRZEDKARPACKIEGO

Abstract

ARTIFICIAL NEURAL NETWORKS APPLYING FOR DETERMINING THE ABSOLUTE ROCK PERMEABILITY ON THE BASIS OF DATA FROM BOREHOLES SITUATED ON THE DZIKÓW–WOLA OBSZAŃSKA AREA (NORTHEASTERN PART OF THE CARPATHIAN FOREDEEP BASIN)

Abstract. The absolute rock permeability was determinated with the use of artificial neural networks (ANN). Authors checked up ANN ability to determine permeability on the data from five borehole locked in northeastern part of the Carpathian Foredeep: Dzików 16, 17, 20 and Wola Obszańska 10 and 15. Neural models were built on the basis of results from laboratory tests, well logs data and the results of the comprehensive interpretation. ANN provided good results in estimating laboratory permeability. The best neural network was applied on similar data set from DZ17 borehole to show that complicated links between input variable and absolute permeability can be used for prediction of permeability from another data. It is hard to find deft deterministic model for permeability estimation so neural model gained in training process is an alternative method.