Active Quasi-LPV Modeling and Identification for a Water-Jet Propulsion USV: An Experimental Study. Issue 28 (2015)

• Record Type: Journal Article
• Authors: Xiong, Junfeng; Li, Decai; He, Yuqing; Gu, Feng; Han, Jianda
• Is Part Of: IFAC-PapersOnLine. Volume 48:Issue 28(2015)
• Page Start: 1359
• Page End: 1364
• Publication Date: 2015
• Subjects: Active Modeling -- Identification -- Quasi-Linear Parameter Varying System -- Unmanned Surface Vehicles -- Unscented Kalman Filter; Automatic control -- Periodicals; 629.805
• DOI: 10.1016/j.ifacol.2015.12.321
• Languages: English
• ISSNs: 2405-8963

Abstract

Abstract: Due to the complicated hydrodynamics and disturbances (wind, wave and current), an accurate model of Unmanned Surface Vehicles (USVs) that is rich for model-based control and simple for parameters identification is difficult to be determined. To handle this problem, an active Quasi-Linear Parameter Varying (qLPV) modeling method is proposed. The model is composed of two parts: a structured model and an online estimated additional error model. Specifically, the qLPV model is derived from the nonlinear physics of the USV to approach the real dynamics as accurate as possible. Subsequently, the UKF based active modeling technique is proposed to estimate the model error online and eliminate the impact of mismatch between the qLPV model and the real one. Finally, the new proposed model is identified and the model's priority compared to some other USV models is proved by extensive experiments.