基于贝叶斯网络的空间受限项目风险管理研究

Research on Risk Management of Space-limited Project Based on Bayesian Network

摘要:

空间受限项目的事故风险源多，且易诱发连锁反应。如何快速识别空间受限项目的风险源，并有效评估各个 诱发节点的风险大小，对空间受限项目的风险管理具有重要意义。识别空间受限项目风险源并构造贝叶斯网络，通 过贝叶斯网络推理的正向预测、反向诊断，获得各个风险源节点的后验概率，并计算风险源概率与事故发生的互信 息，同时引入风险可控值概念及指标量化，构建基于贝叶斯网络的二维四区风险管理模型，提出空间受限项目风险 源管理策略，最后通过一个实例对研究结果进行说明。

Abstract:

There are various risk resources of accident in space-limited project, and easily cause relative chain reactions. It is significant for risk management of space-limited project to rapidly identify risk sources and effectively get assessment of the risk of each induced accident node. Authors identify risk sources of space-limited project and construct a Bayesian Network risk model to get the posterior probability of each risk source node which uses Bayesian Network’s forward prediction and backward diagnosis. Then authors calculate the mutual information of probability of risk source and accident, introduce a concept of controllable and quantified value of risk sources to construct Twodimensional & Four-zones (TDFZ) risk management model based on mutual information, and propose corresponding management strategies. Finally, authors explain the research result through an example.