文献类型：期刊文献

英文题名：ESO-ECNMPC for multiple USV-AUV heterogeneous systems enclosing multiple targets without velocity measurements

作者：Liu, Haitao[1,2,3]; Zhong, Yangwei[1,2]; Tian, Xuehong[1,2,3]; Mai, Qingqun[1,2,3]; Zhang, Jing[1,2,3]

机构：[1] School of Mechanical Engineering, Guangdong Ocean University, Zhanjiang, 524088, China; [2] Shenzhen Institute of Guangdong Ocean University, Shenzhen, 518120, China; [3] Guangdong Engineering Technology Research Center of Ocean Equipment and Manufacturing, Zhanjiang, 524088, China

年份：2026

卷号：359

期号：P1

外文期刊名：Ocean Engineering

收录：EI(收录号：20261920670266)

语种：英文

外文关键词：Adaptive control systems - Data mining - Error compensation - Model predictive control - Nonlinear simulations - Nonlinear systems - Predictive control systems - System stability - Target tracking - Velocity measurement

外文摘要：In this paper, an adaptive error compensation nonlinear model predictive control (ECNMPC) scheme is proposed to solve the problem of enclosing and tracking multiple target vessels for multiple underactuated USV-AUV heterogeneous systems with unmeasurable self-velocity. First, a nonlinear extended state observer (ESO) is employed to approximate external disturbances and unmeasurable velocities. Second, in the multitarget tracking phase, two parallel target vessels are treated as a single target for enclosing and tracking. Third, upon detection of a target escape, the system is partitioned into two subsystems: one subsystem is tasked with pursuing and tracking the escaped target, while the other continues to encircle and track the remaining target. Furthermore, an adaptive error compensation method is designed to establish a prediction model compensation strategy that effectively mitigates disturbances, accelerates error convergence, and enhances system stability. Finally, numerical simulations are performed under realistic conditions to rigorously demonstrate the effectiveness and feasibility of the proposed approach. ? 2026 Elsevier Ltd. All rights are reserved, including those for text and data mining, AI training, and similar technologies.