针对洪水灾害风险评价中单一赋权法存在一定局限性和风险等级难以客观划分的问题, 提出改进组合赋权-模糊聚类算法, 开展洪水灾害风险评价研究。考虑洪灾危险性、敏感性和易损性等 3 个方面筛选构建评价指标体系, 并分别采用直觉模糊层次分析法和 VC-CRITIC 法赋予主、客观权重, 通过改进博弈论组合赋权法计算最优组合权重, 加权计算不同评价单元洪灾风险度, 利用高斯混合模型模糊聚类算法划分区域洪水灾害风险等级。以茨南淝左片防洪保护区遭遇淮河干流百年一遇洪水为例进行洪灾风险评价算法应用研究, 结果表明: 极高风险区和 高风险区共占保护区总面积的 24. 87 % , 基本为淹没水深较大、地形位指数较低和社会经济价值较高的区域。评价结果较为合理可靠, 所提改进组合赋权2模糊聚类算法可为防洪保护区洪水灾害风险评价和防灾减灾决策提供技术支持。

Given the limitation of the single weighting method in flood risk assessment and the difficulty of the objective division of risk level, an improved combination weighting-fuzzy clustering algorithm is proposed to carry out the research of flood risk assessment. Co nsidering the three aspects of flood risk, sensitivity and vulner ability, the evaluation index system is constructed. The subjective and objective weights are given by the intuitionistic fuzzy analytic hierarchy process and VC-CRITIC method. The optimal combination is calculated by the game-theory-based combination weighting method. Combination weights are used to calculate the flood risk of different evaluation units, and the Gaussian-mixture-model-based fuzzy clustering algorithm is used to classify the regional flood risk levels. T aking the 100-year flood of the mainstream of the Huai River as an example, the application of the flood risk assessment algorithm was carried out in the Cinanfeizuopian flood control protected zone. The results show that the extremely high-risk area and high-risk ar ea which account for 24. 87% of the total area of the protected area submerge. In areas with large water depth, low to pographic index and high socio-economic value, the evaluation results are more reasonable and reliable. The improved combination weighting-fuzzy clustering algorithm may provide flood disaster risk assessment and disaster prevention and mitigation decision-making in flood control protected zones with technical support.

国家重点研发计划( 2017YFC0405601) ; 高等学校学科创新引智计划( B14012)