基于灰色关联和聚类算法的避难所选址研究

Study on Locating Urban Emergency Shelter Based on Gray Correlation and Mean Deviation Clustering

以合理利用空间资源、提供城市应急管理决策建议为目标,针对现有避难所选址模型中避难所影响因素考虑不全面和避难所数量确定过于主观的问题,选择7个避难所评价指标,使用熵权法计算各指标的权重,采用灰色关联法计算每个避难所加权灰色关联度,基于该指标体系评估避难所的条件,建立多目标选址模型,使用均值偏移聚类算法求解最优避难所。最后,以福州市仓山区三环以内为实验区验证该方法解决避难所选址问题的科学性和合理性。结果表明,该模型提高了避难所选址质量,优化了避难所的空间分布。

To reasonably use the spatial resource and offer suggestions on urban emergency management decision for governments,aiming at problems caused by incompletely considering advantages and disadvantages ofshelters and over-subjectively determining the number of shelters in existing shelter location models,this paper constructed a multi-objective site selection model by successively selecting 7 shelterevaluation indexes,computing the weight of each index with the entropy weight method and the weighted gray correlation degree of each shelter with gray correlation method,based on which the quality of shelters were evaluated.The optimal shelters are obtained by means of solving the model with the mean shift clustering algorithm.Finally,the effectiveness and rationality of the proposed method in processing locating shelterswere verified by the case study which took the part of Cangshan District within the third ring road in Fuzhou City as the experimental area.The results show that the model improves the quality of shelter location and optimizes the spatial distribution of shelters dramatically.