Porosity Prediction fromWireline Logs Using Artificial Neural Networks: a Case Study in north-east of Iran

The objective of this study is a modeling in artificial neural networks (ANN) and its generalization to predict reliable porosity values from log data obtained from three wells in Khangiran gas field located in northeast of Iran. We used a back-propagation ANN method (BP-ANN) to predict porosity. The ANN for porosity is a simple three-layer network which uses sonic, density and resistivity logs for input. Porosity predictions were then compared with log porosity which had been derived from density and neutron logs. The results confirmed the capability of using ANN.