应急服务设施的排队最大覆盖选址模型与算法研究
发布时间：2019-05-19 11:58
【摘要】：近年来,随着突发自然灾害事故发生的日益频繁,其对社会所造成的破坏与影响也呈现出加剧和扩大的趋势。在无法准确预知灾害发生时间和地点的情况下,研究如何有效降低灾害发生后所造成的损失和影响,已成为灾害应急管理的重要课题。已有的研究表明：同灾后的救援相比,灾前的物质储备在时效性和经济性上具有明显的优势,对于有效降低灾害的人员和财产损失有重要作用。应急服务设施的选址问题作为战略决策,它对灾害救援效果的影响是持久和深远的。而在众多的选址理论中,覆盖选址模型在应急服务设施选址问题中具有十分广泛的应用。本文梳理了覆盖选址模型自其提出以来至今的主要发展过程,并对其分类进行了阐述。具体来讲,按照模型的目标函数划分,主要分为集合覆盖和最大覆盖。按照建模所使用的方法,分为确定性模型,概率模型和应用排队论的模型。为了加深对基本模型的理解和提高对模型的应用能力,重点对几类重要的覆盖选址模型做了进一步介绍,包括模型的具体表现形式及其重要假设和不足之处。随后,重点研究了在假设条件和处理方法上更加精确的应用排队论的概率覆盖选址模型。按照排队系统的不同,分别研究了M/M/s/s损失系统和4/G/s非损失系统条件下的排队覆盖选址模型。对于前者,建立了有服务水平保证的排队覆盖选址模型：无法立刻得到服务的需求因不能排队等候而损失,但是任何被覆盖的需求点能够得到服务的概率不小于事先给定的服务水平。并从覆盖量的角度建立了描述模型MCLP-R1和便于求解的整数线性规划模型MCLP-IR2,设计了在需求可以部分满足时的求解算法。结果表明,该算法可以在较短的时间内求得令人满意的解。对于后者,无法立刻得到服务的需求将进行排队等候,但是额外限制了排队等候的时间。考虑到不同的需求指派方式,分别建立了基于系统指派和顾客选择两种指派方式下的模型,QMCLP-SA和QMCLP-NS。对于系统指派方式,设计了求解问题的算法,结果表明：该算法可以在较短的时间内求得令人满意的解,因此可以迅速得出问题的下界。同时,对比分析了以上两种不同指派方式所造成的选址差异,结果表明：系统指派方式具有较高的效率,但是从顾客的角度来看,则可能存在不公平问题；而顾客最近选择方式所得到的结果则恰好相反。最后,对研究的不足和进一步改进的方向做了总结和建议。
[Abstract]:In recent years, with the increasing frequency of sudden natural disasters, the damage and impact on society has also shown a trend of aggravation and expansion. Under the condition that the time and place of disaster can not be predicted accurately, it has become an important topic of disaster emergency management to study how to effectively reduce the loss and impact caused by disaster. Previous studies have shown that compared with post-disaster rescue, pre-disaster material reserves have obvious advantages in timeliness and economy, and play an important role in effectively reducing the loss of people and property in disasters. As a strategic decision, the location of emergency service facilities has a lasting and far-reaching impact on disaster rescue. Among many location theories, coverage location model is widely used in emergency service facility location problem. This paper combs the main development process of the coverage location model since it was put forward, and expounds its classification. Specifically, according to the objective function of the model, it is mainly divided into set coverage and maximum coverage. According to the method used in modeling, it is divided into deterministic model, probability model and applied queuing theory model. In order to deepen the understanding of the basic model and improve the application ability of the model, several important coverage location models are further introduced, including the specific forms of the model and its important assumptions and shortcomings. Then, the probabilistic coverage location model with more accurate application of queuing theory in hypothetical conditions and processing methods is studied. According to the difference of queuing system, the queuing coverage location model of M/s/s loss system and 4/G/s non-loss system is studied respectively. For the former, a queuing coverage location model with service level guarantee is established: the demand that can not get the service immediately is lost because it can not wait in line. However, the probability that any covered demand point can get a service is not less than the given service level in advance. From the point of view of coverage, the description model MCLP-R1 and the integer linear programming model MCLP-IR2, which is convenient to solve, are established to design an algorithm when the requirements can be partially satisfied. The results show that the algorithm can obtain satisfactory solutions in a short time. For the latter, the need for immediate service will be lined up, but the waiting time is extra limited. Considering different requirements assignment methods, the models based on system assignment and customer selection are established, QMCLP-SA and QMCLP-NS., respectively. For the system assignment mode, an algorithm is designed to solve the problem. The results show that the algorithm can obtain a satisfactory solution in a short time, so the lower bound of the problem can be obtained quickly. At the same time, the location differences caused by the above two different assignment methods are compared and analyzed. The results show that the system assignment mode has higher efficiency, but from the customer's point of view, there may be unfair problems. The results of customers' recent choices are the opposite. Finally, the shortcomings of the research and the direction of further improvement are summarized and suggested.
【学位授予单位】：南京大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TU984.116;D63
本文编号：2480707
[Abstract]:In recent years, with the increasing frequency of sudden natural disasters, the damage and impact on society has also shown a trend of aggravation and expansion. Under the condition that the time and place of disaster can not be predicted accurately, it has become an important topic of disaster emergency management to study how to effectively reduce the loss and impact caused by disaster. Previous studies have shown that compared with post-disaster rescue, pre-disaster material reserves have obvious advantages in timeliness and economy, and play an important role in effectively reducing the loss of people and property in disasters. As a strategic decision, the location of emergency service facilities has a lasting and far-reaching impact on disaster rescue. Among many location theories, coverage location model is widely used in emergency service facility location problem. This paper combs the main development process of the coverage location model since it was put forward, and expounds its classification. Specifically, according to the objective function of the model, it is mainly divided into set coverage and maximum coverage. According to the method used in modeling, it is divided into deterministic model, probability model and applied queuing theory model. In order to deepen the understanding of the basic model and improve the application ability of the model, several important coverage location models are further introduced, including the specific forms of the model and its important assumptions and shortcomings. Then, the probabilistic coverage location model with more accurate application of queuing theory in hypothetical conditions and processing methods is studied. According to the difference of queuing system, the queuing coverage location model of M/s/s loss system and 4/G/s non-loss system is studied respectively. For the former, a queuing coverage location model with service level guarantee is established: the demand that can not get the service immediately is lost because it can not wait in line. However, the probability that any covered demand point can get a service is not less than the given service level in advance. From the point of view of coverage, the description model MCLP-R1 and the integer linear programming model MCLP-IR2, which is convenient to solve, are established to design an algorithm when the requirements can be partially satisfied. The results show that the algorithm can obtain satisfactory solutions in a short time. For the latter, the need for immediate service will be lined up, but the waiting time is extra limited. Considering different requirements assignment methods, the models based on system assignment and customer selection are established, QMCLP-SA and QMCLP-NS., respectively. For the system assignment mode, an algorithm is designed to solve the problem. The results show that the algorithm can obtain a satisfactory solution in a short time, so the lower bound of the problem can be obtained quickly. At the same time, the location differences caused by the above two different assignment methods are compared and analyzed. The results show that the system assignment mode has higher efficiency, but from the customer's point of view, there may be unfair problems. The results of customers' recent choices are the opposite. Finally, the shortcomings of the research and the direction of further improvement are summarized and suggested.
【学位授予单位】：南京大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TU984.116;D63
【参考文献】
相关期刊论文 前1条
1 杨丰梅;华国伟;邓猛;黎建强;;选址问题研究的若干进展[J];运筹与管理;2005年06期
，本文编号：2480707
本文链接：https://www.wllwen.com/kejilunwen/sgjslw/2480707.html
