The Bayesian network method is introduced into the process of fire risk quantitative assessment. The event tree model is established, and the Bayesian network model is transformed from the event tree model based on the typical fire scenarios in high-rise space. A Bayesian fire risk assessment algorithm for high-rise buildings based on mutual information reliability is proposed. Bayesian network is modified considering the influence of uncertainties. Finally, the modified Bayesian network model is used to calculate the probability of fire developing to different stages, and the estimated value of property loss is used to express the severity of the accident and calculate the fire risk value. The results show that the existence of uncertainties has a significant impact on the results of risk assessment; the quantitative assessment method based on Bayesian network is better than the ETA method based on event tree analysis in dealing with uncertainties and is more suitable for high-rise space fire risk assessment.

贝叶斯网络方法被引入到火灾风险定量评估的过程中。建立事件树模型，并根据高层空间中典型的火灾场景，从事件树模型转换贝叶斯网络模型。提出了一种基于互信息可靠性的高层建筑贝叶斯火灾风险评估算法。考虑不确定性的影响，对贝叶斯网络进行了修改。最后，使用改进的贝叶斯网络模型来计算火灾发展到不同阶段的可能性，并使用财产损失的估计值来表示事故的严重程度并计算火灾风险值。结果表明，不确定性的存在对风险评估的结果有重大影响。