多目标多情景下的地震灾害避难所选址优化研究

Disaster shelter planning is one of the most effective ways to reduce casualties in a disaster. A rapid progress has been made on disaster shelters research, however there have been few works done on the multi-objective optimization models for shelter location and allocation by considering the disaster condition and evacuee. This project is going to take the location planning of the earthquake disaster shelter as the case study, after systematically summarizing the principles and standards of shelter planning, and considering the collapse of buildings and roads, and the distribution of evacuees, a multi-objective model for shelter location planning with the constraints of safety, accessibility and capacity of shelters is developed. By solving the model, several optimization plans have been obtained under the scenarios of different seismic intensities. The results can provide a scientific basis for optimizing the shelter location and provide decision support for emergency evacuation after an earthquake.

规划建设城市灾害避难所是减轻灾后人员伤亡最积极而有效的途径之一。近年来，灾害避难所的研究工作取得了一定的进展，但考虑灾情大小和避难人口需求的多目标灾害避难所选址优化模型仍极为缺乏。本项目拟以地震灾害避难所选址为例，在系统归纳、总结灾害避难所规划原则与标准的基础上，结合数学规划方法，充分考虑由于地震的不确定性造成的建筑物和道路破坏，以及避难人口分布状况，以避难所和避难道路安全性、避难所可达性和避难所可收容性等为约束条件构建多目标的灾害避难所选址模型。并利用人工智能算法，进行模型的解算，得到不同设定地震下的灾害避难所与避难路径规划方案，以期为灾害避难所优化选址提供科学依据，也为震后应急避难提供决策支持。

合理的地震灾害应急避难场所配置，能有效提高震后灾区居民的避难率，大大减轻地震灾害造成的人员伤亡率。项目以地震灾害避难所选址为例：通过安全性、可达性和可容性分析，以及建筑物和道路破坏情况，确定了潜在灾害候选避难场所的区位，备选疏散路径和避难需求人口。以最小化从居民点到指定避难所的总加权避难时间和最小化选定避难所的总建设成本为目标，建立了地震灾害应急避难所的选址模型。针对上述模型，在基本的PSO算法中，加入了离散机制、约束机制（疏散距离、避难所空间等约束条件）和局部搜索机制（避免早收敛），以及多目标策略（疏散时间和避难所建设成本）等。应用改进的PSO算法，通过求解Pareto优解，对多目标模型进行了解算。并针对朝阳区5个不同的地震情景，对地震应急避难场所的选址优化进行了实证研究，得到不同设定地震情景下的灾害避难所与避难路径规划方案。结果可为灾害避难所优化选址提供科学依据，也为震后应急避难提供决策支持。