The mathematical model of ship motion is the core of autonomous navigation systems for unmanned vessels. Obtaining accurate model parameters enables more precise control of ships. This paper proposes a Multi-Innovation Unscented Kalman Filter (MIUKF) and introduces forgetting factors to reduce the correction weight of historical innovation, aiming to improve the identification accuracy of model parameters. Additionally, an indepth investigation is conducted on the influence of different innovation lengths on the identification accuracy. The MIUKF is applied to identify the parameters of an improved ship 2-order nonlinear response model, which includes the time constant of the rudder. This model is obtained by multiplying a 2-order linear response frequency domain model with the frequency domain model of the rudder servo system and introducing nonlinear terms after Laplace inverse transformation. The model parameters are identified using simulation data from a Ziazag test maneuver, and the identified parameters are validated for their generalization capability. The results demonstrate that the identification accuracy of the MIUKF algorithm gradually improves as the innovation length increases. Additionally, it verifies the excellent generalization capability of the identified parameters.
Parametric Identification of Ship Maneuvering Model Based on Multi-Innovation Unscented Kalman Filtering
2023-10-28
1152859 byte
Conference paper
Electronic Resource
English
Parametric Identification for Nonlinear Ship Maneuvering
Online Contents | 2006
|An Unscented Kalman Filter based wave filtering algorithm for dynamic ship positioning
Tema Archive | 2011
|Adaptively robust unscented kalman filter for tracking a maneuvering vehicle
Online Contents | 2014
|