有害物品运输网络逆向设计与优化研究

Many of the materials transported by trucks, trains, vessels, and planes are flammable, explosive, poisonous, corrosive, or radioactive. Because being harmful of those materials, there will be potential risk in hazardous materials(HAZMAT) transportation. Compared with the path selection in given network, this project, from the inverse angle of view, presents an analytical approach for the problem of designing a road network for HAZMAT shipments. That means government's authority can change certain road segments to hazmat transportation by using toll policy or close the segments in the network. Based on the inverse network design and optimization, HAZMAT carriers can naturally pursuit the road with low public and environmental risk. .There are three main problems in this research. First of all, because there are many different risk definitions and requirements in HAZMAT transportation, we need to investigate the effection of the different risk measure and requirements in road selecting and network inverse design. Secondly, we will discuss the HAZMAT transportation network inverse design and optimization in deterministic, stochastic and time-varying network. The effection of the deterministic, stochastic and time-varying network in HAZMAT transportation needs to deeply discuss. Then, the inverse bi-level network design and optimization models for HAZAMT transportation in the deterministic, stochastic and time-varying network will be constructed. With high computational complexity, the heuristic algorithms will be developed. Thirdly, because the cost, risk and travel time coefficients and the data in the HAZMAT network inverse design and optimization are subject to uncertainty, we propose a robust problem in HAZMAT network inverse design. .In this project, there are two main contributions. One is in inverse network optimization and the other is in HAZMAT transportation network design. In inverse network optimization, we discussed the bi-level network design problem, multi-origin and multi-destination network problem, stochastic and time-varying network problem. In HAZMAT transportation network inverse design, based on the risk definitions and HAZMAT transportation requirements, government's authority can regulate the carrier select the minimum possible transport risk road in HAZMAT transportation. Thus, this research can significantly reduce the exposure risk and avoid catastrophe in HAZMAT transportation.

本项研究基于逆向优化的角度，以设计和优化有害物品运输网络为研究对象，旨在丰富有害物品运输的理论与应用。针对有害物品的危害特性，结合逆向优化方法和理论，从宏观的角度出发，通过逆向设计与优化有害物品运输网络，使有害物品运输车辆自动选择风险较低的运输路径，避开人口密集地区，有效降低运输风险。首先，根据有害物品运输风险的多种表达形式，研究其对运输路径选择和逆向网络优化的影响；进而研究基于静态、随机和时变有害物品运输网络的逆向设计和优化，定义和测度随机和时变因素对有害物品运输的影响，构建具有互动效应的双层有害物品运输网络逆向模型，设计有效的求解算法；并进一步研究有害物品运输网络逆向优化的鲁棒性，获得网络逆向优化策略的有效性分析。通过本项研究，将在理论方面完善网络逆向优化的研究，从实际应用方面将加强我国有害物品运输的网络设计，提升管理有害物品运输的能力，降低有害物品运输的风险，保护人民生命财产安全。

课题所要研究的问题得到了比较深入和系统的回答，项目组共有8篇标注本项目的研究成果，《管理科学学报》3篇（含录用）、《管理工程学报》4篇（含录用）、《中国管理科学》1篇（含录用）。还完成了工作论文（审稿中）6 篇，正在完成工作论文4篇。课题组围绕“有害物品供应链管理”的大领域，从“基础网络逆向设计”到“随机网络阻断问题”再到“静态网络下有害物品运输网络逆向设计和优化”，并进一步到“供应链网络中渠道管理的理论与应用”与“供应链网络中技术引进与技术改进管理”，比较完整的研究了“有害物品供应链管理”进行了拓展性研究，取得了一系列的创新成果。具体包括：（1）有向网络中双最短路问题的逆优化问题。为进一步研究有害物品逆向网络设计的研究，提供了基础理论与算法。（2）随机网络下的阻断问题研究。为进一步研究有害物品逆向网络设计的研究，从阻断的角度，提供了基础理论与算法。（3）静态网络下有害物品运输网络逆向设计和优化。完成了静态网络下，有害物品运输网络逆向设计和优化的基本模型，并设计了相应的求解算法。同时，针对非线性风险，完成了静态网络下，有害物品运输网络逆向设计和优化的基本模型，并设计了相应的求解算法。（4）完善了供应链网络中渠道管理的理论与应用研究。完成了考虑互联网发展对零售商商业模式选择的影响，分别以单产品和多产品为研究对象，从经营模式和纵向整合这两个角度研究了零售商商业模式选择。（5）完善了供应链网络中技术引进与技术改进管理研究。完了具有技术转让与改进的最优决策问题，用经典的消费者效用函数构建了采购方进行捆绑采购的效用，并结合随机效用理论建立了离散选择模型研究了技术捆绑采购过程中最优采购量、最优分配比例问题，给出了最优组合决策和条件。