【作者】 赵俊；

【导师】 苏义鑫；

【作者基本信息】 武汉理工大学 ， 交通信息工程及控制， 2016， 博士

【摘要】 随着世界人口的增多、科技的进步和经济的增长,人们逐渐开始对深海资源进行勘探和开采。在深海中应用常规的锚泊系统对船舶进行系泊,难以满足工程的实际需求,导致了船舶动力定位技术的产生和快速发展。船舶动力定位技术具有不受水深限制的特点,具有极大的优势,在海上补给、援潜救生、海上打捞和深海油气开采等工程领域有着广阔的应用前景。但船舶动力定位系统是一个非线性、强耦合、时变、工作环境恶劣的系统,常常受到强烈的扰动作用,且检测信号也经常受到干扰,给控制带来了挑战。本文从模型预测控制出发,研究船舶动力定位系统控制。主要工作与成果如下:(1)针对风、浪、流扰动下的船舶动力定位约束控制,提出了一种带有约束调整策略的船舶动力定位广义预测控制方法。分析风、浪、流扰动的特性,对于可测扰动,根据其测量值实现前馈控制和对船舶推力约束的调整,并将该约束调整策略引入到广义预测控制的滚动优化中。应用所提出的方法为一艘水面供应船设计动力定位约束控制器,并对控制器的有效性进行了仿真验证。(2)针对随机扰动下的船舶动力定位控制,提出了一种基于非线性连续滤波器估计船舶运动状态的非切换解析模型预测控制方法。应用无迹卡尔曼布西滤波算法,获取船舶运动状态的估计值;根据船舶非线性运动模型,应用非切换解析模型预测控制方法设计动力定位控制器。仿真结果表明滤波后信号的波动明显减小,所设计的控制器的控制量连续,且波动较小。(3)提出了一种带有扩展卡尔曼滤波(extended Kalman filtering,EKF)滚动时域估计的动力定位预测控制器设计方法和一种带有无迹卡尔曼滤波(unscented Kalman filtering,UKF)滚动时域估计的动力定位预测控制器设计方法。由于无迹卡尔曼布西滤波算法较多依赖扰动的先验信息,为了减少对先验信息的依赖,分别应用基于EKF的滚动时域估计方法和基于UKF的滚动时域估计方法设计滤波器,将这两种滤波器分别与非切换解析模型预测控制相结合,设计动力定位控制器。通过MATLAB仿真,将这两种控制器分别与前文提出的带有无迹卡尔曼布西滤波的船舶动力定位预测控制器进行了比较,指出了不同滤波器的优缺点。(4)考虑船舶时变参数对动力定位的影响,将径向基神经网络补偿器与预测控制方法相结合提出了一种带有径向基神经网络补偿器的动力定位预测控制器设计方法。通过非切换解析模型预测控制器实现船舶动力定位系统的闭环控制,进一步通过带自调整功能的径向基神经网络补偿器产生补偿量,减小模型不准确对动力定位控制效果的影响。通过MATLAB仿真,将所提出的带有神经网络补偿器的动力定位预测控制器与带有神经网络补偿器的动力定位动态逆控制器进行了比较,结果表明所提出的预测控制器的超调量小、鲁棒性好。本文针对船舶动力定位系统及其工作环境的特点,深入研究其运动状态估计方法和模型预测控制方法,提出了多种控制器设计方法,通过MATLAB仿真验证了所提出方法的有效性。本文的成果对船舶运动控制系统的研发有一定的指导意义。更多还原

【Abstract】 With the increase of world population,progress of science and technology and growth of economy,people gradually began to explore and exploit the resources of the deep sea.The application of traditional mooring positioning system is unable to satisfy the practical requirements in the deep sea,which leads to the emergence and rapid development of marine dynamic positioning technology.The mooring of ships are not restricted by water depth when marine dynamic positioning technology is used,so the technology has great advantages.The technology has broad application prospects in the seaborne resupply,submarine rescue,marine salvage and deep-sea oil and gas extraction.But the marine dynamic positioning system is very difficult to control because of several factors.The first one is that the system is a nonlinear,strong coupling,time varying and poor working conditions system,the second one is that the system is often subjected to strong disturbance,and the last factor is that the detection signal is often disturbed.In this dissertation,based on the model predictive control,the control of marine dynamic positioning system is studied.The main work and results are as follows:(1)For the dynamic positioning constrained control of ship under wind,wave and current disturbances,a generalized predictive control(GPC)method with a constraint adjustment strategy is proposed.Based on the analysis of the wind,wave and current disturbances,the feedforward control is realized according to the measurements of the disturbances and the thrust constraints of ship are adjusted according to the measurements of the disturbances.Then the constraint adjustment strategy is introduced into the rolling optimization of GPC.The proposed method is applied to design a dynamic positioning constraint controller for a surface supply vessel,and the effectiveness of the controller is verified by simulation.(2)For the dynamic positioning control of ship under random disturbances,a non-switching analytic model predictive control(NSAMPC)method with a nonlinear continuous filter is proposed.Unscented Kalman-Bucy filtering(UKBF)is used to get the state estimates of the ship motion.According to the nonlinear ship motion model,NSAMPC is applied to design a dynamic positioning controller.The simulation results show that the fluctuation of the filtered signals is obviously reduced,and the control values of the designed controller are continuous and the fluctuation of them is small.(3)A dynamic positioning predictive controller with extended Kalman filtering(EKF)moving horizon estimation filter and a dynamic positioning predictive controller with unscented Kalman filtering(UKF)moving horizon estimation filter are proposed.The UKBF algorithm is heavily dependent on the prior information of the disturbances.In order to reduce the dependence on the prior information,an EKF based moving horizon estimation method and a UKF based moving horizon estimation method are applied to design filters,respectively.These two filters are respectively combined with the NSAMPC control to design dynamic positioning controllers.By MATLAB simulation,these two controllers are respectively compared with the previously proposed marine dynamic positioning predictive controller with UKBF,and the advantages and disadvantages of different filters are pointed out.(4)Considering the influence of time varying parameters of ship on dynamic positioning,a design method of dynamic positioning model predictive controller with a radial basis function(RBF)neural network compensator is proposed.The controller is the combination of RBF neural network compensator and model predictive control.The closed loop control of the marine dynamic positioning system is realized by a NSAMPC controller.The compensation values are generated by a self-tuning RBF neural network compensator.The values are used to reduce the impact of model inaccuracy on the control performance.By MATLAB simulation,the proposed dynamic positioning model predictive controller with an RBF compensator is compared with the dynamic positioning dynamic inversion controller with an RBF compensator.The results show that the overshoot of the designed predictive controller is small and the controller has strong robustness.In this dissertation,according to the characteristics of marine dynamic positioning control system and its working environment,the state estimation methods and model predictive control methods are deeply studied.A variety of controller design methods are proposed,and the effectiveness of the methods are verified by MATLAB simulations.The results of this dissertation have some instructive significance for the research and development of ship motion control system.更多还原