基于复杂网络聚类的应急救援中心动态选址优化研究

the location allocation of the supply center is not only the great demand for disaster relief, but also the research of complex net, net clustering, optimal multi-object distributed in the space. The disaster mitigation group can make their effort to disaster alleviation. however, few literatures reported that the disaster alleviation has reached the minimum casualties and loss. One of the import reasons is the disaster relief command cannot know how to get the optimal supply center for disaster alleviation. This project proposes the complex net clustering model based on the shortest path to seek for the optimal dynamic location allocation for supply center. Single center net clustering calculates the centroid of all net nodes with the iterative, firstly. Then, the optimal results would be reached with the descend method, which followed by the dichotomy of the neighbor of the entire net. Multi-center net clustering would split and combine the classes automatically, according to the shortest path from the center of the class to the vertex located in the border. Net clustering can approach the optimal results to meet the demand of the disaster relief. It is an effective approach to study the optimal dynamic location allocation for supply center for different disasters. The disaster mitigation is complex system, and the complex net clustering provides an approach for the optimal rescue. Moreover, complex net clustering has proposed the support for the theory of the multi-center location allocation, complex net, optimal problems.

应急救援中心动态选址是重大社会需求，也是最优救援、选址、复杂网络、网络聚类、多目标空间优化的理论需求。救援组织可能尽了最大能力救灾减灾，但鲜有文献报道某次灾害救援的应急救援中心动态选址达到了最优。本项目提出基于最短路径的复杂网络聚类模型，研究最优的灾民应急安置点、野战医院、伤病员临时集聚转运点、直升机救援着陆地点的动态选址。单中心的复杂网络聚类模型采用迭代法求取近似最优解，利用二分邻域下降法获取最优解。多中心的复杂网络聚类模型基于单中心的复杂网络聚类模型，采用ISODATA算法获取近似最优解，通过比较每类边界上的网络节点与各聚类中心的最短路径，调整网络节点所属的类别，自动进行类别的"分裂"与"合并"。复杂网络聚类法具有很快的收敛速度，能快速获取最优解。本项目应用复杂网络聚类法探索应急救援中心选址，揭示灾害救援的复杂性，为最优救援提供科学支撑，为网络聚类、复杂理论及最优化等提供理论支持。

应急救援中心动态选址是重大社会需求，也是最优救援、选址、复杂网络、网络聚类、多目标空间优化的理论需求。救援组织可能尽了最大能力救灾减灾，但鲜有文献报道某次灾害救援的应急救援中心动态选址达到了最优。本项目提出基于最短路径的复杂网络聚类模型，研究最优的灾民应急安置点、野战医院、伤病员临时集聚转运点、直升机救援着陆地点的动态选址。单中心的复杂网络聚类模型采用迭代法求取近似最优解，利用二分邻域下降法获取最优解。多中心的复杂网络聚类模型基于单中心的复杂网络聚类模型，采用ISODATA 算法获取近似最优解，通过比较每类边界上的网络节点与各聚类中心的最短路径，调整网络节点所属的类别，自动进行类别的"分裂"与"合并"。复杂网络聚类法具有很快的收敛速度，能快速获取最优解。重要研究成果《基于复杂网络聚类的最优选址模型》应用复杂网络的原理与方法，可以解决灾害救援复杂系统中的应急救援中心选址。重要研究成果《灾害范式及其历史演进》表明人们对灾害的认识和理解是逐步深化和发展，灾害范式与减灾实践的发展互动生成，范式的产生与转变都有深刻的社会动因。本项目应用复杂网络聚类法探索应急救援中心选址，揭示灾害救援的复杂性，为最优救援提供科学支撑，为网络聚类、复杂理论及最优化等提供理论支持。