鲁棒性的极大似然前馈神经网络及其应用研究

提出了一种新的基于极大似然法的BP神经网络算法,该算法定义了一种新的误差函数。与传统BP算法相比,其优点在于不仅考虑了噪声对网络学习的影响,而且能够从全局的角度对网络参数进行学习,从而使网络的鲁棒性增强。实验结果说明了本方法的可行性与优越性。     

Robust adaptive NN-based output feedback control for a dynamic positioning ship using DSC approach

A robust adaptive NN-based output feedback control scheme is presented for a dynamic positioning ship with uncertainties and unknown external disturbances. We tackle the problem that velocity vector of a ship is not available by employing a high-gain observer, and develop the proposed control approach by combing vectorial backstepping with dynamic surface control approach, which is simpler and easier to implement in engi- neering practice. The neural network （NN） approximation technique is used to compensate for the uncertainties and unknown external disturbances, and it removes the requirement for the prior knowledge about the vessel parameters and external disturbances. Also, it is demonstrated that the proposed control strategy can force the position and yaw angle of a dynamic positioning ship to approach the desired point while guaranteeing all singles of the designed closed-loop dynamic positioning system semi-globally uniformly ultimately bounded by means of the Lyapunov function. Simulation results of a supply ship illustrate the effectiveness of the proposed scheme.     

Synchronized tracking control for dynamic positioning vessel

This article addresses the time-synchronized control problem for a floating liquefied natural gas (FLNG) vessel and a liquefied natural gas carrier (LNGC) in side-by-side offloading operation. For such a problem, the existing finite-time control approaches available are essentially only designed to ensure that the system state-variables converge within a certain time. Specifically how each state-variable converges is typically not considered in the design process. Along this line, the approach of time-synchronized design in this paper attains the more comprehensive and desirable outcome of having the marine vessels' state-variables arrive at the sliding mode surface in a fixed time. Also additionally, the state-variables converge at the same moment. This approach thus can attain synchronous tracking of the LNGC to the FLNG vessel. All these desirable properties and outcomes of control system convergence are rigorously proved in the work here via the methods of Lyapunov-type theorems and associated stability analysis. Furthermore, in order to ensure that the approaches developed here are suitably applicable in real-world engineering conditions (with system uncertainties and external disturbances), robustness extensions are also incorporated. In the presented work, the analytical issues and the methodologies in the designed control system solution under the conditions of the disturbed systems are discussed in detail. The numerical simulations conducted verify the merits of the proposed algorithm. Comparative numerical simulations on the developed scheme, and on traditional control, are carried out to demonstrate the important and improved outcomes attained with the time-synchronized features.     

考虑事件触发输入的船舶自适应动力定位控制

为解决实际海况下全驱动船舶的动力定位控制任务存在参数不确定、模型结构不确定和通信资源限制等问题,本文提出一种具有事件触发输入的鲁棒自适应动力定位控制算法.该算法采用径向基函数神经网络对系统模型不确定进行逼近,同时针对通信带宽受限问题,设计了一种具有事件触发机制的执行器输入,降低了控制器和执行器之间的信道占用.此外,该算法还解决了状态变量与执行器增益不确定性之间的强耦合问题,并且设计了在线更新的自适应参数去补偿执行器增益不确定,以确保船舶能够稳定执行动力定位任务.利用Lyapunov稳定性理论证明了闭环控制系统中所有误差变量都满足半全局一致最终有界收敛.通过对比仿真实验验证了所提出算法的有效性.     

Robust dynamic ship positioning control system design and applications

This paper describes the robust control system design for a ship dynamic positioning system. The control design is based on an approximate linear model derived from the nonlinear hydrodynamic equations governing the horizontal motions of the ship. The nonlinear models of the ship, seawaves, current, wind and thrusters are derived and simulated for control design verification. The H_∞ control design technique is employed to design the controller. The control problem is formulated in state‐space form and the design specifications are translated into requirements on the weighting functions of the error signal and the thrusters input. A tuning procedure is proposed based on the wind and wave disturbances. The controller is initially tested on the nonlinear ship model and simulation results are presented to demonstrate the robustness of the H_∞ controller. Tank tests results are then presented to assess the controller performance. Copyright © 2001 John Wiley & Sons, Ltd.     

动力学效应的动力定位船舶模型在线辨识算法

由于船舶模型的高度非线性以及外界干扰力、推进器推力的无法测量性,导致它们的在线辨识和估计显得十分困难.本文提出一种以动力学效应为基础,应用无味卡尔曼滤波器（unscented Kalman filters,UKF）进行动力定位船舶动力学模型、外界载荷以及推进器推力在线辨识的算法.此算法能够在动力定位过程中不断求解船舶模型和其受到的载荷力,使得拥有这些参数的船舶模型和载荷所反映出的动力学效应不断逼近传感器检测到的运动反馈.基于此原理,用这些参数作为名义上的船舶模型、外界力、推进器推力就能够完成高效、自适应的定位控制.通过控制仿真,证明了此算法的有效性和正确性.     

A survey of dynamic positioning control systems

Offshore exploration and exploitation of hydrocarbons have opened up an era of dynamically positioned (DP) vessels. DP control systems maintain floating structures in fixed position or pre-determined track for marine operation purposes exclusively by means of active thrusters. There are more than 2000 DP vessels of various kind operating worldwide. This paper gives a survey of some of the major technology advances in the DP controller design having taken place during more than 30 years of research and development. In addition some perspectives for the future with corresponding research challenges will be addressed.     

一种无人船动力定位跨平台实时控制模型

为了更好地将实时控制算法应用于无人船的自动导航,针对以往跨平台实时控制的数据类型不统一、通信不可靠等问题,根据日本国立水产工学研究所进行无人船水池实验的研究模式,建立一种基于用户数据报协议(UDP)的远程控制模型,该模型在服务器端使用python编写,在客户端使用Matlab/Simulink.为验证控制的实时性,在实...     

动力定位船轨迹跟踪鲁棒自适应容错控制

本文研究了输入饱和状态下的动力定位船故障容错鲁棒自适应控制问题.该问题以动力定位船轨迹跟踪任务为目标,提出了一种新颖的鲁棒自适应控制器的设计,并且引入了二阶快速非奇异终端滑模和神经网络控制算法保证了控制器在实际任务中的执行效果.首先,介绍了三自由度动力定位船的运动模型包括了运动学模型和动力学模型以及推进器故障模型.然后,设计了二阶快速非奇异终端滑模面,提出了一种针对时变扰动和模型不确定性的鲁棒控制方案,保证系统无抖振现象的前提下实现了系统更快的收敛速度.同时运用被动容错控制思想,确保动力定位船在推进器故障发生时依然能够实现预计的跟踪性能.此外,通过Lyapunov稳定性判据分析,证明了提出的改进自适应滑模控制方法可确保系统在初始状态未知前提下,跟踪误差渐近收敛于零.最后,通过数值仿真实验结果验证了控制律的有效性.     

基于高增益观测器的船舶动力定位系统的输出反馈控制

针对动力定位船舶的速度向量不可测的问题, 考虑外部环境扰动, 将高增益观测器、动态面控制技术和矢量backstepping方法相结合, 设计仅依赖于船舶位置和艏摇角测量值的船舶动力定位系统输出反馈控制律. 动态面控制技术的引入, 使控制律结构简单, 易于工程实现. 应用Lyapunov函数证明了所设计的控制律能迫使船舶的位置和艏摇角收敛于期望值, 并保证船舶动力定位输出反馈闭环系统所有信号均一致最终有界. 基于一艘供给船的仿真研究验证了所设计的基于高增益观测器的船舶动力定位输出反馈控制律的有效性.     

Optimal setpoint chasing in dynamic positioning of deep‐water drilling and intervention vessels

Conventional controller designs for dynamic positioning of ships and floating marine structures have so far been based on the principle on automatic positioning in the horizontal‐plane about desired position and heading co‐ordinates defined by the operator. A three degrees‐of‐freedom multivariable controller either of linear or nonlinear type, normally with feedback signals from surge, sway and yaw position and velocities, has been regarded as adequate for the control objective. For floating structures with small waterplane area such as semi‐submersibles, feedback from roll and pitch angular rotation velocity may also be included to avoid thrust‐induced roll and pitch motions that are caused by the hydrodynamic and the geometrical couplings between the horizontal and vertical planes. However, for certain marine operations this control philosophy may not be the most appropriate approach ensuring safety and cost effectiveness. For drilling and work‐over operations the main positioning objective is to minimize the bending stresses along the riser and the riser angle magnitudes at the well head on the subsea structure, and at the top joint as well. A positioning control strategy solely based on manual setting of the desired position co‐ordinates may not be the most optimal solution for these applications. In this paper a new hybrid dynamic positioning controller, that also accounts for riser angle offsets and bending stresses is proposed. It is shown that a significant reduction in riser angle magnitude can be achieved. Simulations with a drilling semi‐submersible demonstrate the effect of the proposed control strategy. Copyright © 2001 John Wiley & Sons, Ltd.     

Robust adaptive fault‐tolerant control of dynamic positioning vessel with position reference system faults using backstepping design

This paper presents a solution to guarantee a fault‐tolerant control (FTC) of dynamic positioning (DP) vessel in the presence of the position reference system (PRS) faults by using the backstepping technique. The faults modes of PRS are modeled as an additive uncertainty in the feedback loop of the DP control system, and the bounds of the additive uncertainty are unknown. Through estimating online the unknown bounds on the basis of the designed adaptive mechanism, the robust adaptive fault‐tolerant controller (RAFTC), which takes the faulty PRS measurements as the inputs, is proposed rather than detecting and isolating them. Moreover, the developed RAFTC is proved by applying Lyapunov stability theory that the closed‐loop DP control system is uniformly bounded and the tracking error can be converged to zero asymptotically despite of the PRS failures. A simulation scenario on an offshore supply vessel model is provided to validate the effectiveness of the designed RAFTC, and the results show that the RAFTC has better performance and robustness compared with the ordinary fine‐tuned adaptive backstepping controller.     

动力定位船舶自适应滑模无源观测器设计

针对带有模型参数不确定性的动力定位船舶,提出一种动力定位船全速域自适应滑模无源观测器,解决了现有观测器只能应用于低速作业动力定位系统的问题.采用速度估计误差作为滑模面,设计切换自适应律估计模型不确定项上界,保证了观测器增益的有界性和系统鲁棒性.对速度估计回路的无源性进行了分析,并证明了观测器的稳定性.最后利用船舶动力定位系统半实物仿真平台,验证了算法的有效性.     

Learning optimization for CPN-based training in robot positioning control

Artificial neural net (ANN) models have been applied to the inverse kinematic problem for controlling robot positions. The selection of ANN training parameters, however, is an important yet complicated step which has to be taken before an ANN model for robot positioning control can be implemented effectively. The objective of this research is to utilize the counterpropagation network (CPN) for inverse kinematic mapping and obtain the best performance possible by systematic adjustment of network parameters. Taguchi statistical methods, efficient methods for analyzing the capability and accuracy of a system, have been used in this study. The working envelope of the robot simulated in this research is 150×150×60 mm³. The optimal accuracy and standard deviation determined by this research are 2.62 mm and 1.2 mm, respectively.     

自适应SRCKF在GPS动态单点定位中的应用

平方根容积卡尔曼滤波具有良好的数值稳定性与滤波效率,针对滤波中状态函数与实际不符带来的误差对滤波造成的影响,将基于预测残差统计量的自适应因子和最优自适应因子与平方根容积卡尔曼滤波算法相结合以降低预测信息在滤波中的权重,并将平方根容积卡尔曼滤波与带自适应因子的平方根容积卡尔曼滤波用于GPS动态单点定位数据处理,最后用航摄飞机实测GPS动态观测数据验证了算法的有效性。     

半潜式平台动力定位系统优化控制器设计

基于滚动优化机制,设计了随机波上的非线性半潜式平台动力定位系统控制器;基于最优预测机制,将每个控制周期内的半潜式平台动力定位系统控制器设计问题转化为一个带约束的非线性优化问题求解;基于系统演化方程的迭代公式,消去目标函数中状态变量,获得非线性优化问题的非线性约束表达式;给出示例说明了该控制器设计过程,仿真验证了该控制策略和控制器的有效性。     

动力定位的分数阶双环自适应终端滑模控制

为了解决具有非线性和环境干扰的船舶动力定位系统的控制问题,本文提出了一种基于分数阶双环自适应快速终端滑模控制算法,证明了闭环系统的稳定性,设计了自适应控制律,计算了滑模的收敛时间。控制系统由外环位置姿态环和内环速度环构成,外环滑模控制实现自动定位于期望位置和姿态,内环滑模控制实现对速度和角速度的定位。通过切换函数的设计,对不确定性和外加干扰具有较强的鲁棒性,避免系统出现抖振现象。对双环滑模控制器进行仿真,计算出动力定位外界环境扰动的变速运动情况下的前进位置、横荡位置、艏向角度、前进速度、横荡速度、艏向角速度、前进控制力、横荡控制力和艏向控制力矩等。对外环控制率增益λ2等参数对控制性能的影响进行了比较分析。结果表明,分数阶双环自适应终端滑模控制动态响应要稍快于传统双环控制,且超调量小,调整时间更短,所设计的控制器对有非线性和环境干扰的船舶动力定位系统都具有较强的鲁棒性。本算法为不确定性系统的变结构控制提供了新的解决方案,扩展了滑模控制算法的应用领域。     

动力定位预测控制器船舶模型试验研究

使得船舶按照特定轨迹运动或者到达指定位置,并减少或者消除到达设定位置时的超调量是动力定位控制器研究中的关键问题。针对上述问题,运用预测控制的基本理论,结合自适应方法中的最小二乘辨识,设计了动力定位系统自校正预测控制策略。在MATLAB仿真中,该预测控制器有效解决了模型失配和系统超调的问题。为了进一步验证该预测控制器的实际控制效果,利用船舶模型对算法进行了动力定位定点模式下的水池模型物理试验。在直接使用仿真参数的情况下,船模试验的控制效果并不理想;参数修改后,动力定位效果有了极大改善。试验结果表明,在实际控制中,为保证控制效果,艏向角度的控制应占最高权重;不能忽视船舶运动时的耦合现象。     

Comparison of control design models and observers for dynamic positioning of surface vessels

This paper investigates some fundamental aspects of control system design for dynamic positioning of marine surface vessels. Different control design models are compared. The impact of applying linear and nonlinear models to describe the hydrodynamic damping loads are investigated, and common observer algorithms for DP are compared. The comparison includes a linear time-varying Kalman filter, an extended Kalman filter, and an unscented Kalman filter. To make the comparisons rigorous, optimization is employed to find optimal observer gains, and high-fidelity simulation data and full-scale experimental data are used to test the performance under operations/conditions with significant transient effects.     

相关噪声下非线性滤波及在动力定位中的应用

针对实际系统状态估计具有互相关噪声的情况,研究了互相关噪声下非线性系统状态估计问题.首先基于贝叶斯理论推导出新的互相关噪声下的贝叶斯估计算法.然后使用三阶球面径向基(spherical-radial)规则计算贝叶斯估计中的非线性积分,当噪声互相关时,基于扩展卡尔曼滤波的思想分别计算状态矩阵和观测矩阵的Jacobi矩阵,可得互相关噪声下的容积卡尔曼滤波(cubature Kalman filtering with one-step auto-correlated and two-step crosscorrelated noise,CKF--CCN);当噪声不相关时,可得容积卡尔曼滤波(cubature Kalman filtering,CKF)及其平方根形式(SCKF).最后通过动力定位系统仿真实验,表明提出的CKF-CCN的估计精度要高于SCKF和仅考虑一步互相关的平方根容积卡尔曼滤波(SCKF-CN).