Nonlinear Modeling of Miniature Helicopters - A Comparative Study Based on Flight Data

Abstract

Listen

Translate article

Mathematical models of a flight vehicle can take various forms. Physics-based phenomenological models are often preferred for aircrafts, because their parameters usually have a physical meaning, the underlying physics and aerodynamic effects are well-known, and, as the resulting structure is linked to physical phenomena, they generally offer large validity domain. During the modeling process, a trade-off has to be made between high accuracy and identifiability. This paper presents a methodology for selecting a model of adequate complexity for miniature helicopters. To this end, three models with increasing complexity, typically used for such systems, are considered. The parameters of the models are estimated from flight datasets and the models are compared in terms of accuracy and complexity, thus highlighting the relevant physical effects. The fidelity of the models on both rotational and translational dynamics is evaluated. The contribution of the paper is thus twofold: an identification procedure for a nonlinear parametric model of a miniature helicopter is first described and applied on the three aforementioned models (flight data collection and preprocessing, structural analysis, estimation technique). Secondly, prediction capability of these models is evaluated using flight data and compared, which allows the selection of a relevant model structure depending on the target application.