中文摘要：

提出了一种基于动态贝叶斯网络的研发项目风险概率评估模型,将时间序列理论与贝叶斯理论相结合,通过概率和有向无环图表达了不同时间片段之间风险因素状态变化的关系。同时发现在样本数据缺失的情况下,采用Leaky Noisy-or gate模型来计算节点的条件概率,这样可以得出更为客观的评估结果。通过与静态贝叶斯网络评估结果对比,动态贝叶斯评估模型提高了研发项目风险概率评估的准度,为风险控制提供了更为科学的依据。

英文摘要：

A research and development （R＆D） project risk probability assessment model based on dynamic Bayesian net- work is proposed, which combines time series theory and Bayesian theory. Through probability and directed acyclic graph, it expresses the changing relationship of risk factors state between different time segments. Moreover, in the case of missing data samples, it uses the Leaky Noisy-or gate model to calculate the conditional probability. Then it can obtain a more ob- jective assessment result. Compared with the assessment of static Bayesian networks, dynamic Bayesian network model en- hances the R＆D project risk probability assessment accuracy, which can provides a more scientific basis for risk control.