中文摘要：

在复杂的灾害风险系统中,风险并非简单相加,但目前的研究成果基本是单一灾种简单相加得到的综合风险,缺乏可靠性。因此,此研究基于灾害风险系统理论,引入模糊信息粒化方法和模糊转化函数,利用模糊近似推理理论和方法,建立一个多灾种综合风险评估软层次模型。研究表明该模型的优势：1）不仅考虑了灾害风险系统中的确定性,而且还包括了随机不确定性和模糊不确定性;2）利用模糊信息粒化方法不仅减少了数据的不确定性,而且还包括了一些主观信息,使得评估结果更加接近实际,理论与实际紧密结合,更有利于风险管理者和决策者为减少损失规避风险提供依据;3）通过模糊转化函数将不同灾种得到的不同量纲的量转化同一量纲的量,以便于综合分析和模糊近似推理,获得多灾种综合风险。以云南省丽江地区（市）的地震-洪水灾害为例,验证多灾种综合风险评估软层次模型的实用性,并将其结果与世界银行灾害管理中心和哥伦比亚大学灾害和风险研究中心所建议的风险评估模型（HMU-CHRR模型）的结果进行比较分析,讨论了本研究所建模型在多灾种综合风险评估中的特点。

英文摘要：

Integrated risk assessment is one of the important parts of integrated risk management, which can help governments to take maeasures on preventing and mitigating the losses of disasters. What＇s more, multi-haz- ards integrated risk assessment is currently very important and difficult in the field of disaster reduction at home and abroad, with the systematic theory and methodology not yet obtained. Based on the system theory of disas- ter risk, this study has introduced the methods of fuzzy information granulation and fuzzy transformation func- tion, and built a soft hierarchical model of integrated risk assessment for multi-hazards, which is characterized by the following aspects. 1） Not only the certainty of disaster risk, but also the random uncertainty and fuzzy un- certainty are considered. 2） The approach of fuzzy information granulation can reduce the uncertainty of data, and it includes some subjective information so as to make the results of assessment more realistic, theory and practice more tightly combined, and thus, it is more conducive to risk managers and decision makers to provide the basis of loss reduction and risk aversion. 3） Fuzzy transformation function transforms different types of quantity into the same type of quantity in order to comprehensively analyze and obtain the assessment result of multi-hazards integrated risk. Using earthquake-flood disaster risk as a case study, the demonstration analysis of soft hierarchical model of integrated risk assessment for earthquake-flood disaster illustrates that this soft hierar- chical model suggested in this study holds the ability of synthetically assessing multiple disasters.