INVERSION OF GRAVITY DATA USING A GENETIC ALGORITHM

Abstract

The genetic algorithm for the selection of gravitational sources is considered. In the basis of the approach, the principle of selection of genes from fathers and mutations was laid down, which was adapted to form geological structures. For a two-dimensional grid model, the designation of apparent density in the blocks is selected by a choice of models of two (parental) variants, in which gravitational initial and calculated anomalies coincide better. In the quality of the object function is used a middle gradient norm of gravity fields. Generation of new models for effective density in blocks is released randomly. Theoretical models were built up for one body with one and two values of apparent densities. The theoretical sections with four layers were considered. The fitting of the model was carried out under the condition that the value of the effective density was known or a certain range was set. Each block was rectangular in shape with a square section in the plane of the gravity data profile and a limited lateral elongation. Comparison of the output and calculated anomalies of the gravitational acceleration was carried out using the average norm and the percentage error. The absence of jumps in the objective function graph ensured that an accurate model was determined. The correct geometry of a body with a homogeneous apparent density was determined at a fixed value of the effective density for four layers. The model with two values of density had some errors in determining the geometry of the bodies. The genetic algorithm, based on an evolutionary approach to certain physical parameters of blocks, performs the fitting of a gravity model rather quickly and effectively. The main factors affecting the accuracy of geometry are apparent density data. The implemented approach allows one to estimate the cross section by the grid distribution of the effective density. The development is applied for a two-dimensional interpretation of the gravity anomaly over an oil and gas field. The resulting interpretation of the shape of the anticlinal structure is consistent with geological data.