【作者】 刘洋；

【导师】 郭晨；

【作者基本信息】 大连海事大学 ， 交通信息工程及控制， 2013， 博士

【摘要】 随着航海科学技术和船舶与海洋工程的发展,当代海洋资源开发和海上运输对于船舶动力定位系统的要求越来越高,也促进了动力定位系统技术的快速发展。研究动力定位问题具有重要的理论意义和实用价值。三自由度的水面船舶是典型意义的非线性系统,它具有强耦合、大惯性、模型参数不确定性以及工作中受到外界的风、浪和流干扰的特点,传统PID和LQG方法虽然在动力定位系统中取得了应用,但是随着人们对定位精度要求的提高,这些方法存在着一定的局限性,因此吸引了国内外广大学者的兴趣。本论文探索和系统研究船舶动力定位新的控制方法,完成了以下研究工作：(1)根据MMG模型理论建立了一个动力定位船舶的非线性数学模型,为了说明此模型的精确性,通过船舶旋回仿真实验和实船的比较研究来验证模型的有效性。为了实现船舶的航向保持控制,将一种线性自抗扰控制方法应用于船舶运动模型当中,仿真结果表明该控制算法的有效性。(2)基于简化的船舶航向Norrbin非线性模型,针对模型参数不确定和控制增益未知的非线性船舶航向控制问题,采用RBF神经网络自适应控制,提出了一种新的非线性航向保持控制器。理论上,先证明存在连续的控制律,然后通过RBF神经网络对其逼近,最后借助Lyapunov稳定性理论分析证明了船舶航向保持闭环系统的所有误差信号一致最终有界。(3)针对带有模型参数确定和外界风浪流干扰的动力定位水面船舶,采用双环滑模变结构方法设计船舶动力定位控制律,利用积分滑模来实现切换函数的设计。外环滑模控制律实现船舶位置的控制,外环控制器产生速度指令,并传送给内环系统,然后通过内环滑模控制律实现船舶实际速度对速度指令的跟踪,Lyapunov稳定性分析证明了闭环系统的所有误差信号渐近稳定。(4)针对带有模型参数不确定和外界风浪流干扰的动力定位水面船舶,提出一种动力定位船全速域RBF神经网络自适应控制器。在反步设计的过程中,采用RBF神经网络与Nussbaum函数相结合的控制策略。该方法有效地避免了控制器的奇异问题和反步设计过程中的“计算膨胀”问题,Lyapunov稳定性分析证明了闭环系统的所有误差信号一致最终有界。(5)针对带有模型参数不确定和外界干扰的动力定位水面船舶,提出一种动力定位船全速域自适应输出反馈控制器。首先应用Lyapunov直接法设计出全局指数稳定的观测器,然后采用反步设计方法设计出自适应输出反馈控制器,最后利用串级系统理论分析证明了船舶动力定位闭环系统的所有误差信号渐近稳定。(6)针对带有非线性约束条件的推力分配优化问题,对动态的等式约束进行等份离散,在传统的粒子群算法中进行了改进,加入了改进的惯性因子,改进的比较准则和改进的干扰算子,将改进后的粒子群算法应用到推力分配策略中,从仿真中可以看出,改进的粒子群算法可实现推进系统有效跟踪期望指令。更多还原

【Abstract】 With the advancement of the nautical science and technology as well as the ships and marine engineering, contemporary development of marine resources and sea transport has set increasingly higher standards for ship dynamic positioning, and also promoted its rapid development. Therefore, Studying dynamic positioning problem has important theoretical significance and practical value.Three degrees of freedom surface vessels are typical of nonlinear systems, they are characteristic of strong coupling, large inertia, uncertainties of model parameter as well as the disturbance to work by the outside wind, wave and flow. With the increasing demand on the positioning accuracy, the traditional PID and LQG methods do have some limitations in spite of their previous applications, thus arousing the interest of many scholars at home and abroad. And this thesis is to explore and systematically research the new control methods for ship dynamic positioning; the research work is as follows:(1) Based on MMG model theory, establish a nonlinear mathematical model of dynamically positioned vessels; verify the accuracy and validity of the model through ship’s turning simulation tests and comparative study of real ship. In order to achieve control of the ship’s course keeping, a linear ADRC method was applied to the ship motion model, and the effectiveness of this control algorithm has been proved by the simulation results.(2) Based on a simplified Norrbin nonlinear model of ship course, in view of the uncertainties of model parameter and the unknown control gain, using the RBF neural network adaptive control, a new nonlinear course keeping controller was proposed. Theoretically, first prove the existence of a continuous control law, then approximate through RBF neural network, and via Lyapunov stability theory, finally analyzes and illustrates that the consistency of all error signals of the closed-loop system for ship course keeping are ultimately bounded.(3) For surface vessels of dynamic positioning with parameter uncertainties and external disturbances, design the ship dynamic positioning control law by using bicyclic sliding-mode variable structure; achieve the design of the switching function by implementing the integral sliding mode. The outer sliding mode control law is to achieve control of the ship’s position, the outer ring controller generates a speed command and sends it to the inner ring system, and then through the inner sliding mode control law, achieve the actual speed’s tracking for speed command. Lyapunov stability has illustrated that all error signals of closed-loop system are asymptotically stable.(4) For surface vessels of dynamic positioning with parameter uncertainties and external disturbances, a RBF neural network based adaptive controller for the dynamic positioning vessel of all speed envelopes was proposed. In the process of backstepping design, a control strategy was adopted by combining RBF neural network and Nussbaum function. This method was effective to avoid the controller singularity problem and the calculation inflation problems in the process of backstepping design. Based on Lyapunov stability analysis, it’s proved that all error signals of vessels path following closed-loop system are uniformly ultimately bounded.(5) For surface vessel of dynamic positioning with parameter uncertainties and external disturbances, an adaptive output feedback controller for the dynamic positioning vessel of all speed envelopes was proposed. Firstly, a globally exponentially stable observer was designed by applying Lyapunov direct method, and then an adaptive output feedback controller was designed by adopting backstepping design method, and finally based on cascaded system theory, it’s proved that all error signals of closed-loop system of dynamic positioning vessel are asymptotically stable.(6) For a thrust with nonlinear constraints allocation optimization problem, equally scatter the dynamic equation constraint; modify traditional PSO by adding the improved inertia factor, comparison criteria as well as the interference operator; apply the improved PSO to thrust allocation strategy. The simulation result has shown that the improved PSO can make the propulsion system well track the desired command.更多还原