基于投影寻踪动态聚类模型的城市内涝风险评估

Urban Waterlogging Risk Assessment Based on Projection Pursuit Dynamic Clustering Model

摘要: 城市内涝风险评估对于城市防灾减灾以及城市排水设施改造是至关重要的。采用多个指标对城市内涝风险进行评估时，指标权重对风险评估结果具有重大影响。因此，针对风险评估中指标权重模糊不确定性问题，本文基于投影寻踪动态聚类方法对城市内涝风险进行评估。投影寻踪动态聚类是一种对样本数据进行综合评价的客观方法，并能通过设置约束条件来考虑决策者的偏好。选取某排水区建立SWMM模型进行实例研究，模拟计算得到评价指标值，投影寻踪计算得到不同区域的风险等级。通过对结果进行分析可知，投影寻踪动态聚类可以用于城市内涝风险评估，而且对城市防灾减灾具有重要指导意义。

Abstract: Urban flood risk assessment is critical to the prediction and reduction of flood disaster, and also the mitigation of drainage facilities. However, for flood risk assessment using multiple indicators, the index weight has a great impact on the results of flood risk assessment. In this paper, projection pursuit dynamic cluster method is used in order to reduce the uncertainties of the index weight. Projection pursuit dynamic cluster is a relatively objective method, which can evaluate the sample data comprehensively, and can also consider the preference of decision makers by setting constraint conditions. The SWMM model was established for a drainage system. Then different values for the evaluation index were got through simulation. Then the results of urban waterlogging risk assessment are obtained by the dynamic iteration between clustering center and relative membership degree. Through the analysis of the results, it shows that projection pursuit dynamic cluster method can be used for urban waterlogging risk assessment. Furthermore, it brings important guiding significance for urban disaster prevention and mitigation.