Adaptive neural network asymptotic path-following control of underactuated ships with stochastic sea loads. (15th December 2022)

• Record Type: Journal Article
• Authors: Deng, Yingjie; Zhang, Xianku; Zhao, Dingxuan; Ni, Tao; Gong, Mingde; Zhang, Zhuxin
• Is Part Of: Ocean engineering. Volume 266(2022)Part 5
• Publication Date: 2022-12-15
• Subjects: Path following -- Underactuated ships -- Stochastic loads -- RBF NNs -- Asymptotic tracking; Ocean engineering -- Periodicals; Ocean engineering; Periodicals; 620.4162
• DOI: 10.1016/j.oceaneng.2022.113147
• Languages: English
• ISSNs: 0029-8018

Abstract

Abstract: This paper investigates the high-precision path following of underactuated ships suffering from stochastic sea loads, where a novel adaptive neural network asymptotic control framework is fabricated. Based on the nonlinear 3-degree-of-freedom (3-DOF) motion model of the underactuated ship, the stochastic ship motion model is established by involving both unknown deterministic and stochastic disturbances, which is chosen as the control objective. To transform path following to tracking control, a novel virtual ship guidance approach is fabricated with generating smooth reference signals. Through the control scheme, the radial-basis-function neural networks (RBF NNs) are employed to approximate the total marine uncertainties. By introducing integral-bounded functions to control laws and adaptive laws, the asymptotic convergence of positional tracking errors is guaranteed with the aid of adaptive backstepping control. To ensure the computational simplicity, the adaptive laws are fabricated with the minimum learning parameters (MLPs). Via the important inequality of the neural basis function, the "algebraic loop" problem is successfully solved, such that the control logicality is guaranteed. Finally, the numerical experiments validate the high-precision tracking performance of the proposed scheme. Highlights: The uncertainties including the stochastic loads are offset in the underactuated ship. High-precision path following is guaranteed through the asymptotic tracking control design. A virtual ship guidance approach is developed to ensure its compatibility with control.