Grey-box Modeling of Small-Scale Unmanned Helicopter Based on Bayesian Technique

FANG Zhou,LI Ping,HAN Bo,HOU Xin
DOI: https://doi.org/10.3785/j.issn.1008-973x.2009.11.001
2009-01-01
Abstract:This work presented a synthesized method for identification and modeling of a small-scale unmanned helicopter in grey-box modeling framework.A nonlinear first principle model was simplified and divided into longitudinal and lateral motions using some reasonable assumptions,and then linearized around the hover point to get two parameterized models for both motion directions.The unknown parameters were explicitly expressed as prior uniform distributions on certain intervals.A Bayesian maximum A posteriori(MAP)estimation was formed and translated into a restricted nonlinear optimization problem,which was solved by a Lagrange multiplier method using a DFP-based quasi-Newton reeursive algorithm.The identification procedures were done in discrete-time domain,and a"sinch"algorithm was applied to map the direct continuous-time domain parameterization problem into the discrete-time domain.This method effectively reduces the complexity and costs of identification experiments.The acquired continuous-time state-space model shows good prediction performance and is fit for controller design.
What problem does this paper attempt to address?