Natural resonance extraction from multiple data sets using a genetic algorithm

Abstract
The paper presents a scheme to extract the natural frequencies of a radar target from multiple data sets using a genetic algorithm technique (GAT). This technique prevents the extracted damping coefficients from being positive, as often happens with low signal-to-noise ratio data, by restricting the search parameters of the damping coefficients to be negative values. The GAT does not require an initial guess, unlike the constrained E-pulse technique (CET). The results from the GAT compare well with the CET (with much less effort required) and also with such established techniques as the pencil-of-functions method and the nonlinear least-squares technique.