论文 / Publications

竞技帆船与机器人帆船遵循相同的物理规律，却因目标取向与研究社群的分野而长期缺乏系统比较。本文沿部件性能、整船建模、运动控制与气象航路四级技术结构，采用五指标框架评估方法可迁移性，归纳二者的协同点、结构性分歧与共同挑战，为风力推进领域的交叉发展提供参照。

Although racing and robotic sailing are governed by the same physical principles, divergent objectives and research communities have long limited systematic comparison. This perspective review examines the two domains across four hierarchical dimensions—component performance, vessel modelling, motion control, and weather-aware routing—evaluates methodological transferability through a five-metric framework, and synthesises synergies, structural divergences, and shared challenges for the broader wind-propulsion community.

针对非均匀、时变风流场条件下无人帆船的区域覆盖路径规划问题，本文提出一种时空优化框架：在空间域引入拓扑形态约束以保持覆盖区域的紧凑性与连续性，在时间域采用预报感知的前瞻规划以应对环境演化及局部方向不可达，从而在传统往复扫掠策略难以适用的场景中生成可行且高效的覆盖路径。

This paper addresses coverage path planning (CPP) for autonomous sailboats in inhomogeneous and time-varying wind–current fields by proposing a spatiotemporal optimisation framework. Topology-based morphological constraints in the spatial domain promote compact and continuous coverage, while forecast-aware look-ahead planning in the temporal domain anticipates environmental evolution and partial directional inaccessibility, thereby generating feasible and efficient paths in scenarios where conventional boustrophedon sweeping is often inadequate.

针对波浪浮标类比（WBA）在非平稳海况下因固定时间窗平稳假设所引入的谱估计误差，本文定义半平稳海况，并提出基于小波的自适应分段算法，用以识别船舶运动响应中的海况转折点，抑制病态反演中的扰动放大，从而提高实时海况估计的可靠性。

Addressing spectral estimation errors induced by fixed-window stationarity assumptions in the wave buoy analogy (WBA), this paper defines semi-stationary sea states and proposes a wavelet-based adaptive segmentation algorithm that identifies transitions in ship-motion response spectra, mitigates disturbance amplification in the ill-posed inverse problem, and thereby improves the reliability of real-time sea-state estimation.

针对翼帆无人单体帆船总体构型（帆、船体与龙骨尺寸匹配）设计空间广阔且缺乏可靠性能先验的问题，本文提出融合贝叶斯优化与知识迁移的仿真驱动设计框架，在“Seagull”原型上实现显著的航速提升，并相对传统离线代理方法明显降低计算成本。

Addressing the challenge of a large design space and scarce prior performance estimates for the general configuration of wing-sailed autonomous sailing monohulls (size matching among sail, hull, and keel), this paper proposes a simulation-driven design framework that combines Bayesian optimisation with knowledge transfer. Applied to the “Seagull” prototype, the framework yields substantial speed improvements while markedly reducing computational cost relative to conventional offline surrogate-based methods.

针对任务定制且形态差异显著的无人单体帆船，本文建立通用速度预报程序（VPP）。增强遗传算法求解器无需性能先验即可实现收敛，邻域信息优化（NIBO）策略利用相邻风态信息加速求解；数值仿真与海试结果表明，该方法适用于概念设计阶段的性能评估。

For autonomous sailing monohulls characterised by mission-specific and morphologically diverse designs, this paper develops a generalised velocity prediction program (VPP). An enhanced genetic-algorithm solver converges without prior performance estimates, and a neighbourhood information-based optimisation (NIBO) strategy accelerates the solution by exploiting adjacent wind states. Numerical simulations and sea trials indicate that the method is suitable for performance evaluation at the conceptual design stage.

本文从海洋科学任务所关注的结构耐久性、能源自持能力、抗倾覆能力与航速四项关键性能出发，系统评述无人帆船设计的现状与瓶颈，并讨论场景定制设计、自适应机构与仿真驱动优化等发展方向，以期提升其作为海洋观测平台的能力。

This review systematically examines autonomous sailboat design from four performance perspectives central to marine science missions—structural durability, energy self-sufficiency, overturning resistance, and sailing speed—identifies present limitations, and discusses future directions including scenario-customised design, adaptive mechanisms, and simulation-based optimisation to enhance their utility as ocean observation platforms.

本文提出面向翼帆无人帆船的高保真 MATLAB/Simulink 过程对象模型。该模型在六自由度多体动力学框架下，融合 WAMIT 频域水动力系数与基于 CFD 系数图的 Morison 方程及面元分布载荷描述，并通过静力标定与波浪响应分析加以验证，可用于设计优化与控制算法评估。

This paper presents a high-fidelity MATLAB/Simulink process plant model for wing-sailed unmanned sailboats. Within a six-degree-of-freedom multi-body dynamics framework, the model integrates WAMIT frequency-domain hydrodynamic coefficients with Morison-equation and panel-based distributed loads derived from CFD coefficient maps, and is validated through static force characterisation and wave-response analysis for design optimisation and controller assessment.

针对动态扩展水污染区域的监测需求，本文提出一种多无人水面艇（USV）分布式协同覆盖方法。各艇实时共享覆盖地图与状态信息，在运动学约束下优化期望速度与航向，并利用相互疏离机制降低碰撞风险与冗余覆盖。

To address the monitoring of dynamically expanding water-pollution regions, this paper proposes a distributed coordinated coverage scheme for multiple unmanned surface vehicles (USVs). The vehicles exchange coverage maps and state information in real time, optimise desired velocity and heading under kinematic constraints, and employ mutual separation to reduce collision risk and redundant coverage.

针对浮式海上风电机组“双体船运载预组装件并实施海上对接”的一体化安装方案，本文建立反向传播神经网络模型，依据相对运动的统计特征预测对接点极值相对位移与速度，为对接作业风险预警与现场决策提供数据驱动依据。

For an integrated floating offshore wind turbine (FOWT) installation scheme in which a catamaran transports onshore preassemblies for offshore mating, this paper develops a backpropagation neural network that predicts extreme relative displacements and velocities at the mating points from statistical motion features, thereby supporting hazard warning and operational decision-making.

针对大尺度海域内离散多目标点的海底复测任务，本文提出两阶段贪心路径规划方法：首先通过迭代 K-means 聚类确定支撑船锚点位置，继而采用蚁群算法规划各锚点邻域内的访问顺序，以在续航约束下降低多点测量的总体代价。

For multipoint seabed resurvey missions with discrete targets distributed over large sea areas, this paper proposes a two-stage greedy path-planning method. Iterative K-means clustering first determines the locations of support-vessel anchor points; an ant colony algorithm then optimises the visit order within the neighbourhood of each anchor, thereby reducing the overall survey cost under endurance constraints.

本论文围绕无人帆船航速性能不足这一核心问题，系统分析航速与抗倾覆能力之间的矛盾，并沿三条技术路线展开研究：基于计算流体力学与优化求解的速度预报程序（VPP）评价方法、融合贝叶斯优化与知识迁移的仿真驱动总体设计方法，以及被动水翼增稳以实现间接提速的设计方案，并在“Seagull”等平台上予以验证。

This doctoral dissertation addresses the limited sailing speed of autonomous sailboats by systematically analysing the conflict between speed and capsize resistance, and develops three complementary approaches: a velocity prediction program (VPP) for performance evaluation based on computational fluid dynamics and optimisation, a simulation-driven configuration design method combining Bayesian optimisation with knowledge transfer, and a passive-hydrofoil stabilisation scheme for indirect speed improvement. The methods are validated on platforms including the “Seagull” prototype.