Comparison of Different Types of Fitness Functions to Choose the Appropriate Attributes for Porosity Prediction

Abstract

Porosity is one of the most important reservoir characteristics because it indicates to fluid collection. Several techniques used to get good porosity prediction, so, in this study we employed seismic attributes and well log data in a genetic algorithm to get the best porosity prediction. The study attempt to enhance the performance of genetic algorithm for attribute selection and therefore porosity prediction by applying genetic algorithm on different types of fitness functions like average mean square error fitness, average correlation coefficients fitness and performance index fitness. Also, used two methods to represent attributes in genetic algorithm. Different witnesses applied to choose the appropriate fitness function that gives high porosity prediction.