【作者】 王维佳；

【导师】 宗群；

【作者基本信息】 天津大学 ， 控制理论与控制工程， 2007， 硕士

【摘要】 不确定优化问题具有重要的研究意义,而鲁棒优化作为研究不确定优化问题的一种新方法,近年来得到了各国学者的广泛关注。论文主要研究了鲁棒离散优化的建模、模型转化及模型求解理论。并将鲁棒离散优化应用于电梯群控调度问题的实际研究中。论文的具体研究内容如下:论文首先介绍了不确定优化理论的主要分析方法,说明了鲁棒优化的研究意义,并对鲁棒优化理论的研究现状进行了分析。针对不确定系统的鲁棒优化理论,介绍了鲁棒优化的基本原理,然后分析了Bertsimas的鲁棒优化理论框架。重点阐述了鲁棒离散优化理论,给出了鲁棒离散优化建模,模型转化的方法,并研究了约束违背概率边界理论。针对电梯群控调度实际应用问题,研究了鲁棒离散优化建模与求解问题。首先对电梯群控调度问题进行简要介绍,总结分析了现有群控调度方法,并对交通流不确定性这一关键问题加以分析。在此基础上,建立起基于Bertsimas理论的电梯群控调度鲁棒离散优化模型,实现了模型鲁棒对等式的转化,从而将初始不确定鲁棒离散优化问题转化为一个计算可处理的混合0-1整数规划问题。随后,深入研究了鲁棒离散优化问题的求解方法,通过对不同方法的比较分析,最终采用整数规划方法来完成电梯群控调度鲁棒离散优化问题的求解工作。最后论文完成了电梯群控调度RDO算法的实现与仿真实验。对于电梯群控调度问题的鲁棒对等式,即一个混合整数规划问题,可利用优化软件进行辅助求解。结合实验室已有的电梯群控虚拟仿真环境,利用优化软件设计基于鲁棒离散优化方法的电梯群控调度算法,定义算法框架中各部分功能函数的接口。最后进行仿真验证,通过与其他调度算法的比较,证明了鲁棒离散优化调度算法在不同交通流下均具有优良的性能和适应能力。实验表明,鲁棒离散优化调度方法可以解决交通流预测误差的影响,很好地改善电梯群控调度性能。更多还原

【Abstract】 For the significance of the research on uncertain optimization, in recent years Robust Optimization (RO), as a new approach for the uncertain optimization, has gained much attention of an increasing number of researchers all over the world. This paper mainly focus on the establishment and transformation of Robust Discrete Optimization (RDO) model, the solution method and the application of RDO to the practical research on elevator group scheduling. The detail contents of this paper are:This paper introduces the main methods to uncertain optimization theory, illustrating the significance of the research on RO. And analysis of the present research situation on RO and the summary of the research achievements are provided.Towards uncertain RO theory, investigates the basic principle of RO and represents the RO theory structure of Bertsimas. This paper focuses on RDO theory and provides the methods for the establishment and transformation of RDO. Besides, the probability bound of constraint violation is also studied.Focusing on the practical application of elevator group scheduling, in addition, the establishment of RDO model and the solution are concentrated on. Firstly, elevator group scheduling is introduced briefly. Then the paper reviews the present methods of group scheduling and analyzes the key problem of the uncertain traffic flow. The model of RDO based on Bertsimas theory is established and the transformation of robust counterpart is realized to make the original uncertain RDO problem transfer to a computational tractable problem of mixed 0-1 integer programming. And the last section launches a deep research on the solution to RDO. Through the comparative analysis, integer programming method is finally adopted to study the solution to the problem of elevator group scheduling RDO.Finally, this paper realizes elevator group scheduling RDO algorithms and takes the simulation experiment. In terms of robust counterpart of the elevator group scheduling, which is a mixed integer programming problem, optimization software is available to help the process of solution. Combined elevator group stimulated environment and utilized C++ and LINGO optimization software, elevator group scheduling RDO algorithms is realized. At last, the stimulation experiment is implemented. Comparing with other algorithms, it is verified that the RO algorithm has more advantages and suitability under different traffic flow. Simulation result shows that the RO algorithm could decrease the influence from traffic predict errors and improve the elevator group scheduling performance greatly.更多还原