Abstract: Physical property parameters of oil gas reservoir(i.e.reservoir parameters) are important in reservoir description. Considering both the complexity of finding reservoir parameters by seismic data and the shortages of the nonlinear mapping of BP network which needs complete sample information and is the low effciency of learning.This paper presents a method combining Complete Utilization of Sample Information (CUSI) with genetic algorithm(GA) to calculate these parameters.

Key words: function approximation, aitificial neural network, genetic algorithm, reservoir physical properties, feature optimization