Combining Genetic Algorithm and Generalized Least Squares for Geophysical Potential Field Data Optimized Inversion

Abstract

A genetic algorithm (GA) and generalized least squares (GLS)-based approach, hereafter called GA-GLS, is proposed to solve geophysical optimized inversion. In this method, GA is exploited to initialize nonlinear parameter estimation, and GLS is used for accurate local search. Here, we compare the results from GA, GLS, and proposed GA-GLS to invert the synthesized potential field. The results show that GA-GLS outperforms GA in terms of accuracy, as well as GLS, which needs given initial parameters. The real data are taken to verify the feasibility of implementing it in practice.