Estimation of AUV Hydrodynamic Coefficients Using Analytical and System Identification Approaches

Abstract

This article discusses a challenge of the hydrodynamic coefficient identification in autonomous underwater vehicles (AUVs). The considered coefficients correspond to a nonlinear mathematical model of an AUV maneuvering in the horizontal plane. The dynamics of underwater vehicles is intrinsically nonlinear. Hence, to reproduce different types of maneuvers, such as straight line, zigzag, or turning in circles, a nonlinear mathematical model is important. The proposed identification approach combines an analytical and semiempirical estimation method (ASE), based on the hydrodynamics and geometric characteristics of submarine vehicles, with a parameter estimator based on the extended Kalman filter. The identification starts with ASE estimations, and then the estimator based on the extended Kalman filter adjusts the hydrodynamic coefficient estimates based on the experimental data. The experimental data were obtained with the sensors of the Pirajuba AUV during sea trials. This identification method is a cyclical estimation process, where the less reliable parameters are adjusted to the experimental data while the most reliable parameters are kept fixed. Therefore, in the next identification iteration, the originally fixed model parameters are adjusted to the experimental data, keeping the already identified values. Finally, the identified model is used to simulate the vehicle's movement, and the movement variables are compared to the experimental data, thus validating the identified model.