随机海浪干扰下的无人水面艇编队分布式协同控制

Surge, swaying stochastic forces and yaw, rolling stochastic moments can be produced as seawave stochastic disturbances working on the formation of unmanned surface vessels . The stochastic forces and moments, which are coupled by nonlinear system models of the vessel formation, increases nonlinear characteristics and stochastic disturbances quantity of the formation system. Meanwhile, rolling stochastic moments rock the vessels, which leads to severe stochastic measurement noises in the measured vessel states. Therefore, the stochastic disturbances not noly deteriorate formation control accuracy, but also increase collision hazard of the vessel formation. In the face of these problems, the objective of this project is to design distributed cooperative control for the surface vessel formation under stochastic disturbances. The research mainly includes: (1) To construct stochastic system models for the surface vessel formation, and to integrate Kalman filters with neural networks to enhance estimation accuracy of vessels states, which can decrease the effect on states estimation from stochastic seawave disturbances and measurement noises; (2) To propose Kalman consenus filters for virtual leder consensus estimation, such that the accuracy and speed of consensus estimation can be enhanced under coloured measurement noises; (3) To design distributed stochastic predictive self-tuning control by neural network-based feedforward compensation, such that formation control accuracy can by improved by both the feedforward compensation and statistic characteristics of stochastic disturbances, and collision hazard can be decreased by virtual of both prior consideration of constraints and receding horizon optimization in stochastic model predictive control.

随机海浪干扰作用于无人水面艇编队，产生了纵荡、横荡随机干扰力及艏摇、横摇随机干扰力矩。这些干扰力和力矩与编队水面艇的非线性模型相耦合，增大了系统的非线性特性和随机干扰量；而横摇干扰力矩使得水面艇处于摇晃状态，导致水面艇状态被随机量测噪声严重污染。因此，随机海浪干扰不仅恶化水面艇编队控制精度，而且增加编队碰障风险。针对以上问题，本项目研究随机干扰下水面艇编队分布式协同控制，包括：(1)建立符合海况的水面艇编队随机系统模型，并联合卡尔曼滤波和神经网络提高水面艇状态估计的精度，降低随机海浪和量测噪声对状态的影响；(2)提出对虚拟领航者的卡尔曼一致滤波估计，提高量测有色噪声下一致估计的精度和速度；(3)设计神经网络前馈补偿的分布式随机预测自校正控制， 利用神经网络的前馈补偿和随机扰动的统计特性，提高随机干扰下编队控制精度；利用随机预测控制事先考虑约束和滚动优化的优点，降低随机干扰下编队的碰障风险。

随机海浪干扰作用于无人水面艇编队，产生了纵荡、横荡随机干扰力及艏摇、横摇随机干扰力矩。这些干扰力和力矩与编队水面艇的非线性模型相耦合，增大了系统的非线性特性和随机干扰量；而横摇干扰力矩使得水面艇处于摇晃状态，导致水面艇状态被噪声严重污染且增加了倾覆的可能。因此，随机海浪干扰不仅恶化水面艇编队控制精度，而且增加编队碰障风险和倾覆的风险。针对以上问题，该项目联合卡尔曼滤波和扰动观测器提高了水面艇状态估计的精度，降低了随机海浪和量测噪声对状态的影响；建立了自适应卡尔曼一致滤波估计虚拟领航者的状态，提高了随机干扰下领航者的估计精度；根据虚拟结构法将编队控制问题转化为带有约束的轨迹跟踪问题；基于预测控制、障碍李雅普诺夫函数等设计了存在状态和控制输入约束的无人水面艇控制，极大降低了碰障风险和倾覆可能且完成了期望轨迹的跟踪，进而实现了随机干扰下无人水面艇的编队。在该项目资助下，共计发表国内外权威期刊和会议论文17篇，其中SCI论文9篇。圆满完成了项目预期指标。本项目的研究促进了水面艇控制理论的发展，并为水面艇的应用提供了重要技术支持。