【作者】 李健；

【导师】 张翠华；

【作者基本信息】 东北大学 ， 管理科学与工程， 2011， 硕士

【摘要】 半期望设施是指在为居民提供服务的同时,也会影响居民的生活质量或者降低居民财产价值的设施。半期望设施作为回收物流的重要组成部分,其设施选址问题近年来成为热点问题。纵观目前关于半期望设施选址的研究,大多数研究成果集中在确定条件下的选址分析,忽略了不确定条件对选址结果的影响；关于回收物流的研究集中在确定条件下或者随机环境下的回收物流网络设计,而在实际生产中,不确定参数的概率分布往往很难得到,或者系统无法承担低概率事件发生带来的影响。所以研究具有鲁棒性的半期望设施选址模型以及回收物流优化模型具有很强的现实意义。本文考虑在不确定环境下,将鲁棒优化方法应用到半期望设施选址以及回收物流优化研究中,具体的研究内容包括：(1)在有容量限制和无容量限制的条件下,综合考虑选址成本最小化和社会负效用最小化两个目标,建立基于粒子群算法的半期望设施选址基本模型。运用多目标离散二进制粒子群算法和VC++6.0软件对有容量限制模型进行算例仿真和分析。(2)考虑回收量不确定的两种表现形式,即废弃率不确定和人均回收量不确定的条件下,分别运用区间分析法和情景分析法对不确定性进行描述,融入Bertsimas鲁棒优化方法,建立基于鲁棒优化的半期望设施选址模型。最后,进行模型的算例仿真和分析,验证模型的解鲁棒性和模型鲁棒性。(3)考虑在需求量、单位运输成本以及回收量不确定条件下,运用Ben-Tal鲁棒优化方法,建立回收物流优化的鲁棒优化模型。最后对模型进行算例仿真和分析。重点分析了在不同的问题规模和不同的不确定水平下,确定型模型和鲁棒优化模型在回收物流优化中的表现。结果表明：不确定参数在一定的范围内波动,本文建立的半期望设施选址鲁棒优化模型具有解鲁棒性和模型鲁棒性；在回收物流优化中,鲁棒优化模型相对于确定性模型,具有更高的优越性和稳定性。更多还原

【Abstract】 Semi-obnoxious facility provides a benefit or service to society, while adversely affecting the quality of life or social values in a number of possible ways. As a key link in the returned logistics, semi-obnoxious facility location problem has come to be a hot research area. From the comprehensive survey of research on the semi-obnoxious facility location problem, most of the research results are based on the deterministic background, the influence of uncertainty on the facility location is neglected. The study on returned logistics is focused on the network design, which is under the deterministic or stochastic environment. However, the probability distribution of the uncertain data can’t be easily got in actual production condition, or system can’t hold the influence from the occurrence of small probability event. Thus the rubost approach to semi-facility location problem and returned logistics optimization has a great of academic and realistic meaning.This paper apply the robust optimization method to semi-obnoxious facility location problem and returned logistics optimization, detailed research contents, research methods and corresponding conclusions including:(1) With capacity constraint and without capacity constraint, this paper takes an overall consideration on a minisum function to represent the location costs and another minisum function to represent the obnoxious effects of the facility, and two basic models for semi-obnoxious facility location problem based on bi-objective particle swarm are established. Discrete binary particle swarm optimization algorithm and VC++6.0software are applied to solve the model under capacity constraint.(2) Two forms for recovery amount uncertainty are considered in this paper: disposal rate uncertainty and average recovery amount uncertainty. Under the above uncertain conditions, interval analysis and scenario analysis are applied to describe recovery amount uncertainty, semi-obnoxious facility location robust optimization models are established based on Bertsimas robust optimization method. Finally, a numerical study is designed to analyze the model.(3) Considering demand uncertainty、transportation cost uncertainty and recovery amount uncertainty, a robust model for returned logistics optimization is established based on Ben-Tal robust method. Finally, a numerical study is designed to analyze the model. Focuses on the analysis of the performance for the robust model and deterministic model in the returned logistics optimization under different problem size、 different uncertainty level. Results of a computational example verify the robust model is more superior and stable than the deterministic model.The research work has shown the following:Results of a numerical example verify the model robustness and solution robustness for the semi-obnoxious facility location robust optimization model, and in returned logistics optimization, the robust model is more superior and stable than the deterministic model.更多还原