An analysis of models identification methods for high speed crafts

Abstract

Two different approaches of the system identification method have been proposed in order to estimate models for heave, pitch and roll dynamics of a high speed craft. Both of them resolve the identification subject as an optimization problem to fit the best model. The first approach uses genetic algorithms and nonlinear least squares with constraints methods applied in the frequency domain. The second one suggests a new parameterization which facilitates obtaining high quality starting values and avoids non-quadratic functions in the cost function. At last it is shown an example in which the two approximations are applied and compared.