Analysis of Identification Methods Applied to Free Model Tests of the Pirajuba AUV

Abstract

System identification based methods represent a compromise between accuracy and cost in the estimation of a marine manoeuvring dynamics. However, the use of this kind of approach requires the adoption of a methodology capable of identify satisfactorily the vehicle dynamics, filter away the perturbations that occur during the experiments, and validate the identified model. This paper presents the results of a proposed identification methodology applied to the free model tests of the AUV Pirajuba. In this methodology, deterministic and stochastic identification models are applied to the AUV yaw rate signal of zigzag manoeuvres of several amplitudes, so that the influence of nonlinear effects could be assessed. The estimated models are validated against experimental results from zig-zag manoeuvres and a number of turning manoeuvres. The identified models provide an envelope which includes the vehicle's behavior. Such result can be applied to robust identification and control design.