Prediction of petrophysical parameters using probabilistic neural network technique

Abstract

This chapter describes the application of Probabilistic Neural Network (PNN) for prediction of petrophysical parameters in the inter well region using post-stack 3D seismic and well log data of F3 block, the Netherland. The PNN analyses the number of attributes among themselves and the best combination of attributes is selected. Further, these attributes are cross plotted with well log property which needs to be estimated in the reservoir zone and a nonlinear relationship is setup between attributes and petrophysical property. This relationship is further utilized to predict the petrophysical properties away from the borehole. To calculate the volume of porosity, density, P wave velocity, and gamma-ray, a trained PNN model is introduced and the results are compared with the results estimated using Multi-attribute regression method. © 2021 Elsevier Inc. All rights reserved.