Disaster assessment of oil and gas supply chain based on data-driven dynamic Bayesian networks
提出一种结合词嵌入模型和数据驱动动态贝叶斯网络的方法，评估技术、经济等七因素对油气供应链中断的影响，发现中断概率为71%，环境因素影响最大。
The oil and gas supply chain (OGSC) constitutes a fundamental pillar of the global economy, providing vital energy resources, bolstering industrial activities, enabling international trade, and safeguarding national security. However, existing risk assessment approaches exhibit significant limitations: traditional methods predominantly rely on expert judgment or static probabilistic frameworks, often neglecting the temporal evolution of disruption factors and lacking robust data-driven methodologies. These critical gaps necessitate an interpretable, data-driven, and temporally-aware framework capable of accurately identifying, quantifying, and predicting OGSC disruption risks. This study introduces a novel, data-driven methodology that incorporates a word embedding model and develops a Data-driven Dynamic Bayesian Network (DDBN) to assess the impact of seven significant factors, including technology and economy, on disruptions within the OGSC. Our exhaustive analysis reveals a consistent 71% probability of disruptions in the OGSC. Among the seven factors, environmental influences are the most significant, while technical factors have the least impact. Sensitivity analysis and belief propagation are employed to derive valuable managerial insights into the efficacy of the proposed model. The code and associated data supporting this study are accessible at https://github.com/raichll/Research-on-oil-and-gas-supply-chain-disruption.git.