【作者】 黄珍；

【导师】 王永骥；

【作者基本信息】 华中科技大学 ， 控制理论与控制工程， 2010， 博士

【摘要】 长江航道整治工程中,使用了大量适应施工需求的专用工程船舶如挖泥船、铺排船、抛石船等,这些工程船一般为非自航锚泊型船舶,而且大多数是由其他船舶改装而来,自动化作业程度普遍比较低。在目前市场经济的大背景下,拥有高自动化程度的工程船就意味着效益和成绩,因此研究工程船锚泊移位自动控制系统,提高工程船施工作业精度、速度,有较大的实际应用价值。本文首先研究了锚泊移位型船舶的系统建模问题。在已有文献研究的基础上,借鉴常规船舶模型分析方法,分析该类型船舶的运动机理和特点,推导了锚泊船二维和三维的线性运动模型。同时针对该类型船舶的特点和工程应用要求,基于神经网络辨识方法,提出了一种面向锚泊移位型工程船舶的系统辨识模型。以一类典型的锚泊移位工程船——软体铺排船为例,依托工程实践采集的实验数据对神经网络模型参数进行学习训练,得到铺排船纵向位移模型。然后针对建模和控制的优化问题,本文分析研究了量子粒子群算法(Quantum-behaved Paticle Swarm Optimization,QPSO)。作为QPSO算法中唯一的参数,论文从问题依赖性、种群规模等多方面对收缩扩张系数β进行了大量的实验分析,根据实验结果得到该参数选取的指导性准则。为提高收敛速度,论文提出了一种改进的量子粒子群算法。通过仿真实验对比分析,该改进算法的收敛速度相对标准的QPSO算法有显著的提高。利用改进的QPSO算法对铺排船纵向位移模型参数进行了优化训练,优化后的模型作为自动控制系统的设计和仿真的对象模型。基于模糊逻辑设计了工程船航迹保持控制系统,并利用改进的QSPO算法,对模糊控制器的模糊规则和隶属函数进行了优化设计。系统仿真实验结果表明,粒子群优化算法应用于模糊控制器的参数设计是可行和有效的,控制系统的动静态特性指标均能满足工程实践的要求。本文提出了一种基于自适应神经-模糊推理系统(Adaptive Neuro-Fuzzy Inference System, ANFIS)的工程船自适应控制器网络结构,使控制器参数可随环境因素变化以适应不同的施工任务。采用改进的QPSO算法对自适应航迹保持控制网络的前后件参数进行了优化设计。与基于Fuzzy的航迹保持控制器的对比仿真实验结果表明,基于ANFIS的自适应航迹保持控制器的控制效果在动、静态性能均有提高。本文提出基于ANFIS的航迹航向多变量自适应控制器网络模型,网络模型采用子网络并行学习模式,利用多个粒子群对应各子网络参数,使得网络优化训练过程简单、快速。结合工程船施工实际状况,选取典型实例对该自适应控制系统进行了仿真实验,实验结果显示该控制网络能够满足工程船多目标控制要求。结合实际工程项目,将基于模糊逻辑和QPSO优化算法的航迹保持控制系统应用于软体铺排船控制系统中,同时综合应用PLC控制网络技术、GPS定位技术、多传感器信息融合技术、现场总线技术等先进的技术与手段,开发研制了一套软体铺排船作业综合自动监控系统。实船应用数据充分反映了本文探讨的工程船舶锚泊移位智能控制系统的有效性,对于同型工程船舶自动化水平的提高提供了一套切实可行的实施方案。其研究成果可推广应用到类似工程船舶的自动作业系统中,对相关领域的研究也有一定的参考价值。最后,对全文进行了总结,并指出了在今后工作中需要进一步深入探讨的问题。更多还原

【Abstract】 A large number of project vessels, such as hydraulic dredge, geotextiles-laying vessel, stone dumper, are used in regulation works for Yangze River. These project vessels usually belong to a kind of non self-propelled mooring shifting vessels. Remodeled from other common ship, these vessels generally have the poor degree of automation. Under the background of market economy, high-automated project vessels will product more economic benefit. Therefore, researching automated control system of project vessels is beneficial to precision of construction operation and economic benefit.Firstly,the dissertation has studied this type vessels’ system modeling. A lot of previous literatures were analyzed in detail. A linear two and three-dimensional moving model of mooring shifting project vessel was proposed,profit from the conventional ships model analysis method to analysis this vessels’ moving mechanism and characteristic. Face to the this project vessel, the dissertation proposed a system identification plan based on the neural network. Take the geotextiles-laying vessel as example, parameters of the neural network model were training using the empirical datum, gathering from the engineering project. Then the displacement model had been established.Then in view of modelling and controllor optimized question, this dissertation Analysises the QPSO (quantum-behaved paticle swarm optimization) algorithm. The contraction expansion coefficientβis the only parameter of QPSO algorithm.Sufficient experiment is done onβfrom different aspects, including the problem dependence and the swarm size. According to the experimental result, it obtains the selection guidance criterion of the parameter. In order to increase efficiency of convergence, it proposed a new improved QPSO optimization. Simulation experiments indicate that this algorithm improves convergence speed. The proposed algorithm was used to optimize the moving model of the mooring shifting project vessel, which was the object model for the control system design and simulation.Based on fuzzy logic, a new track-keeping controller of mooring shifting project vessel was proposed. Particle swarm optimization algorithm was used to optimize fuzzy rules and membership. Simulation experiments indicate that particle swarm optimization algorithm is effectual for optimizing the parameters of fuzzy logic controller. The dynamic and the static performance indexes of the control system can meet the challenge of operating engineering.A new adaptive controller network architecture of project vessel was proposed in the dissertation and this architecture was based on ANFIS (adaptive neuro-nuzzy inference system). Therefore, the parameters of controller can vary with working environment to accomplish different tasks. The improved QPSO algorithm was used to optimization design the antecedent and consequent parameters of adaptive track keeping control network. The result of simulation experiments shows that the dynamic and static performance of adaptive track keeping controller based on ANFIS, was improved when compared with the controller based on fuzzy logic.The Multivariate adaptive controller network model was also purposed in the dissertation. The main learning pattern of this network model was sub-network parallel learning pattern. And the particle swarm was with corresponding sub-network parameters so that the network training processes would be easier and faster. Then, the simulation experiment of typical instance was chosen to analyze the effect of adaptive control system combined with the practical working situation of working ship. The experiment result shows that multivariate adaptive controller network was able to meet the multi-objective control requirement of project vessel.The track-keeping controller based on the fuzzy logic and improved QPSO has been applied in the project item. An automatic monitoring system was developed to monitor the operation situation of geotextiles-laying vessel. The programmable logic controller, global positioning system, multi-sensor information fusion, fieldbus and other technologies have been applied in this system. Therefore, the intelligent optimized strategy, automatic control and centralized monitoring of geotextiles-laying arrangement operation have been realized. The level of automation and intelligence for operation has been enhanced. The operation data of actual project vessel indicated that the mooring shifting intelligent control system of project vessel was effective, and this control system can also been applied to improve the automation of other ships. Its research results may promote apply in the similar project vessels’ automatic work system, also has certain reference value to the related domain’s research.Finally, the summary of the dissertation and the future work to be investigated are presented.更多还原