基于信息扩散理论的山洪灾害风险因子隶属度分析

Using Information Diffusion to Analyze the Membership Degree in Risk Assessment of Flash Flood

摘要: 山洪灾害多发生于山丘区小流域，其环境条件因素受多方面影响且具有不确定性。针对小流域山洪灾害风险因子概率分布未知，样本数量较少的特点，本研究提出了一种基于信息扩散理论的风险因子隶属度分析方法。首先对各山洪灾害风险因子进行量化，利用正态扩散函数，构造研究网格内风险因子的原始信息矩阵，并确定控制点；在此基础上，根据一维线性信息分配函数，计算各风险因子的隶属度，从而构造相应的模糊关系矩阵。最后，应用信息扩散方法对山洪灾害实验网格内的风险因子状态进行实例应用，构建了实验网格内各山洪风险因子的隶属度，反映了风险因子状态与山洪风险等级的隶属关系。不同于传统的隶属度经验公式，基于信息扩散理论的山洪灾害风险因子隶属度分析，通过有限的知识快速地进行风险定量化评估，极大提高了隶属度计算的科学性和鲁棒性，为山洪灾害的风险评估提供科学依据。

Abstract: Flash flood often occurs at small watershed, its forming condition is uncertain and multi-factorial. Con-sidering the uncertainty of probability distribution and the incompleteness of disaster data at small wa-tershed, a method of membership degree analysis based on information diffusion is proposed in this study. Firstly, the risk indicators of flash flood are quantified, using normal information diffusion function, the original information matrix of risk indicators in study grid is established, and the control point is calculated. Based on one-dimensional linear information distribution function, the membership degree is calculated and the fuzzy relation matrix is established. Finally, case study of flash flood in research grid is present, the membership degree of the risk indicators is calculated, which describe the membership relationship of risk indicators and risk grades. Different from the membership function using empirical formulas, membership function using information diffusion presents a quantitative risk assessment based on limited knowledge. Results show an increased scientific and robustness on membership degree, and contribute to risk assessment of flash flood disaster.