基于贝叶斯网络的长江航运安全状况监测模型研究

Yangtze River Navigation Safety Monitoring Model Based on Bayesian Network

为提升航运安全监测水平,依据系统工程学理论构建贝叶斯网络拓扑结构,提出基于贝叶斯网络的长江航运安全监测模型。通过对1 465条长江航运事故数据分析与处理,借助Genie软件对长江航运安全贝叶斯网络进行模拟,标定模型参数。利用改进马尔科夫链蒙特卡洛算法可以满足细致平衡的要求且能与实际的后验概率保持一致的特点,采用马尔科夫覆盖对模型下一阶段的状态进行预测,最后以武汉长江大桥航段某日观测数据为例进行实例验证。结果表明,影响长江航运安全的4个主要因素的22个子节点相互独立,所建长江航运安全状况监测模型操作性强、可视化程度高、监测结果连续、符合实际情况。

In order to improve the safety monitoring level of Yangtze River navigation, according to thesystem engineering theory, a safety monitoring model of Yangtze River navigation was established by con-structing Bayesian network topology structure. By analyzing and processing 1465 Yangtze River naviga-tion accident samples, the Bayesian network of Yangtze River navigation safety model was simulated andparameters were calibrated by using software Genie. Because the improved Markov method could meetthe requirements of detailed balance and the actual posterior consistent characteristics, the condition ofthe next stage of model was predicted by using Markov blanket. Finally, taking Wuhan Yangtze Riverbridge′s observation data as an example, the operability and accuracy of the model was tested. The re-sults show that the 22 nodes of 4 main factors affecting the safety of the Yangtze River navigation are in-dependent with each other; Yangtze River navigation safety monitoring model has strong operability,high degree of visualization, continuous monitoring results, and conforms to the actual situation.