基于语言变量的贝叶斯网络风险管理框架.PDF

Ke Yong-jian and Wang Shou-qing, Risk Management Framework Using a Linguistic Variable-Based Belief Network for Construction 1 Projects, Chinese Journal of Management Science, Vol.14, Special Issue, pp.211-215, Oct 2006 基于语言变量的贝叶斯网络风险管理框架 柯永建，王守清 （清华大学建设管理系，北京 100084) 摘要：工程项目相关历史数据的不充分导致风险因素的发生概率和影响程度很难统计得到，而专家 也很难准确判断出风险度的数值，因此风险度的确定成为风险定量评价和管理最大的障碍。本文在王守 清教授“外星人眼模型”研究的基础上，提出一个基于语言变量的贝叶斯风险管理框架。首先将项目风 险分为关键风险指标(KRIs, Key Risk Indicators)和关键风险诱因(KRDs, Key Risk Drivers)，以关键风险指 标作为目标节点，以关键风险诱因作为父节点，构造出风险因素贝叶斯网络；其次将计算得到的关键风 险指标概率分布代入财务评价模型，通过蒙特卡罗模拟得到财务评价指标的概率分布情况；最后根据所 得到的评价指标进行风险评估，并提供带有优先顺序的风险管理措施建议。 关键词：贝叶斯网络，模糊数，语言变量，风险管理 Risk Management Framework Using a Linguistic Variable-Based Belief Network for Construction Projects Yongjian Ke, ShouQing Wang (Dept of Construction Management, Tsinghua University, Beijing 100084, China) Abstract— Quantitative risk management is becoming more Risk management generally involves risk identification, risk and more important, which however requires a large number of analysis, and risk response. In addition to the different probabilities in numerical form resulting in a major obstacle of definitions of risk, there are various ways for categorizing risk doing so. Following on the second author’s proposed Alien for different purposes too. In the second author’s previously Eyes’ Risk Model, this paper proposes a risk management proposed Alien Eyes’ Risk Model, the identified risks are framework using a linguistic variable-based belief network, sorted into three l