Deeply-towed underwater vehicle systems : a verified analytical procedure for creating parameterized dynamic models

Abstract

The dynamics of deeply-towed cable/vehicle systems are governed by nonlinear partial differential equations and as a result, trajectory control is generally difficult using the available techniques. This work examines the possibility of utilizing parametric dynamic models in differential equation form, to present a far more tractable controls problem. A learning-model method for generating accurate approximations of this type is used, and the identification process is unique in that an analytically-based model provides the primary data sets, allowing for a priori characterization of system responses without using any real data. The performances of the parametric forms are then verified through comparison of model output against actual sea data obtained during recent cruises in the Caribbean and Mediterranean Seas. The respective merits and limitations of several different model structures are discussed, with respect to both pure performance and identification efficiency.