【作者】 续爱民；

【导师】 金烨；

【作者基本信息】 上海交通大学 ， 机械制造及其自动化， 2008， 博士

【摘要】 大型船舶船体装配过程是船舶制造中的重要环节,它不同于作业车间和流水车间的制造过程,具有设计制造并行、小批量、高度依赖船台资源、制造过程受多种外界干扰等不同于传统制造系统的特殊性和复杂性,并且船舶的生产完全是面向订单的生产模式而不同于面向库存的生产模式。在这种形势下,大型船舶装配过程处于复杂而又具有变化的环境中,其调度控制过程也变得复杂和特殊,建立在传统的生产运作管理理论与运筹学基础上的生产调度与控制方法,已经不能满足其调度控制要求。本文的研究针对这一问题,从建模、调度技术和仿真等方面对大型船舶船体吊装过程的调度控制进行研究。主要包括下述主要工作:(1)为了实现对船台吊装过程的规范化描述、准确表示吊装过程,提出了基于共享合成着色时间Petri网(Colored Timed Petri Net, CTPN)的船台吊装过程建模方法。在建立吊装过程典型操作组件模型的基础上,根据共享合成原理构建出吊装过程的模型,所建CTPN模型能够包含吊装中的技术约束、资源约束和时间等信息,支持对船舶吊装过程调度和控制策略的仿真分析。该建模方法给出了吊装过程的便于计算机表示和操作的数学描述形式。针对所建船舶吊装过程CTPN模型,给出了一种利用极大加法代数求解模型中分段吊装时间的算法。对所建模型的性能分析方法进行了说明,讨论了结合简化规则和可达图方法的定性分析策略。(2)为了在既有的吊装设备、场地等资源条件约束下,寻找可能的最短船台周期并且尽量提前某些重要时间节点,提出了一种基于改进遗传算法的调度算法。根据吊装过程调度问题的特点,对传统遗传算法进行改进,给出了一种适合这类问题的交叉算子,通过避免不可行子个体的产生从而大大提高了搜索的效率和解的质量。某74,500DWT散货船吊装过程调度优化实例说明这种遗传算法对于解决时间安排、设备使用分配等吊装过程调度问题是有效的。为了探讨这类整数编码遗传算法的机理,获得改进和分析此类遗传算法的理论基础,采用Markov链对基于整数编码的遗传算法进行了收敛性分析。提出了一种整数编码遗传算法种群多样性量化定义方法,求解出了特定“基因对”顺序变换概率的公式表示,说明了遗传算法收敛性与遗传算法参数间的关系。(3)降低分段堆放成本是船舶企业的重要课题,从吊装过程调度安排的角度出发提出了一种以分段堆放成本最小为目标的拉格朗日松弛算法。将吊装过程调度问题归结为一种新的包含任务先序约束、机器能力约束以及某些任务同时占用多台机器情况的平行机排序问题。采用拉格朗日算法对船体装配过程中的先序约束、起重设备能力约束进行松弛,然后采用次梯度法求解拉格朗日对偶问题,再通过一定的启发式规则获得可行解。实际集装箱船例子和仿真数据的计算结果表明,这种方法能够满足实际工程生产的求解速度需要和精度要求。(4)吊装过程受多种因素影响,为了提高吊装生产效率和稳定性,必须更加准确地确定吊装时间,根据干扰的影响及时进行有效的重调度。在对分段大小、重量、船型、分段所在区域以及建造季节等吊装时间影响因素分析的基础上,提出了利用模糊神经网络(FNN)确定吊装所需时间的方法。提出了船舶吊装过程重调度策略选择评价指标,给出了基于分支定界思想的重调度优化算法,并结合船舶吊装生产实例,对以上重调度控制技术进行了验证。(5)在仿真和工程实例方面,提出了基于高层体系架构(HLA)的协同仿真平台,对其逻辑结构、功能模块划分、数据以及时间管理等关键技术进行了分析和定义;初步建立了协同仿真平台的各功能模块。建立了大型船舶装配过程仿真模型和调度控制策略之间的桥梁,实现了仿真环境下的调度策略性能分析。在吊装过程三维动态仿真模块中,提出了一种能够增强虚拟吊装过程沉浸感的查看路径规划方法。根据样条小波多分辨率表示曲线原理,并结合查看通道的宽度情况生成查看路径。给出了路径规划的具体实现步骤并展示了在吊装过程三维虚拟场景中进行查看的例子。更多还原

【Abstract】 Erection operation scheduling on dock (EOS), differing from both job-shop and flow-shop, is a kind of new type one, whose special and complex characteristics include large scale, existing occupying multiple machines simultaneously, unbalance load of machine. And its production mode is order oriented style. Under the complex and change chaos circumstance, the control system of EOS is nonlinear. The production scheduling and control methods, based on traditional production operation theory, can’t meet the need of the control of EOS. In order to solve this problem, supported by the Key Item Program of S&T Committee, Shanghai City, China project‘Research on material flow in shipbuilding enterprise’(Grant No.04DZ11004), the thesis’topic is research on the intelligent coordinated scheduling technologies of EOS and its key technologies. The paper discussed from different points, and expected gain breakthrough in theory, method, supporting technologies and engineering application. The following are the main works in this paper:(1) Block Erection in Shipyard is a complex process. A method based on sharing synthesis of timed colored Petri net for modeling of block erection process in shipyard is proposed. A min-max algebra algorithm is used to calculate the time. The information of erection constraints, facilities, space and time consuming are involved and expressed clearly in this model. The approach can be employed to model the erection process of single vessel or multi-vessel. Finally, a case study of the modeling and optimization of erection process of two 74,500 DWT bulk carriers is given and the result shows that this modeling method is effective.(2) A method based on modified genetic algorithm for optimization of block erection process is proposed. Firstly, the constraints of erection network and the production ability of block are analyzed and the objective function to minimize the makespan on dock is given. Then, a modified genetic algorithm is used to optimize this problem, and the coding approach and a crossover operator are introduced in details. Finally, a case study of the optimization of erection process of a 74500 DWT bulk carrier is given and the result shows that this optimization method is effective. The convergence rate of modified genetic algorithm with integer coding was analyzed. Firstly, a diversity of population of genetic algorithm is defined which can formally express the degree of premature for this kind of genetic algorithm. Then the formalized equation is given about the relationship between diversity and crossover rate, mutation rate and population size. Finally, a simulation experiment about block assembly sequence scheduling in a shipyard is introduced to verify the above opinions.(3) A new Lagrangian Relaxation approach is developed to schedule the erection operations to minimize the inventory cost of blocks. It corresponds to a parallel machine problem with the precedence constraints of tasks, the capacity constraints of machines and the situation that some tasks must occupy multiple machines simultaneously. The objective is to minimize the weighted total completion time. A mathematical formulation of the problem is presented firstly. Then the task precedence constraints rather than crane machine capacity constraints are relaxed by Lagrangian relaxation method and the subgradient method is used to solve the Lagrangian dual problem. Finally the results are modified by some heuristic operation. A practical example of container vessel that includes 65 blocks is solved by the proposed approach, and the results show that the Lagrangian relaxation method is able to solve industrial-dimensioned problems within reasonable time and accuracy.(4) The disturbance types of erection time were analyzed and a rescheduling method was proposed. A novel fuzzy neural network based algorithm for analyzing disturbance of erection time was proposed. The learning samples for FNN were setup by simulation evaluation. The relation between the input parameters of block and erection time was built by FNN. The branch and bound algorithm is used to rescheduling the erection operations. The construction of feasible tree and the branch fathom rules were given. The nodes-including rule and lower-bound rule were used to cut the invalid branches.(5) A simulation platform based on HLA was proposed. This platform can help user to simulation their scheduling results dynamically. The architecture and function modules of this platform were analyzed and defined. The data and time management methods were discussed and the prototype of this platform was developed. In 3D simulation module, a method about navigation path multi-resolution planning based on wavelets was proposed to eliminate the vision dither phenomenon and enhance the immerse feeling in virtual environment of shipbuilding. The theory and realization of constructing navigation path were introduced in details and an example about block erection is given.更多还原