基于动态贝叶斯网络的多阶段任务系统维修优化  

Maintenance policy of phased mission system based on dynamic Bayesian network

作　　者：朱康 赵新文[1] 张黎明 余航 ZHU Kang;ZHAO Xinwen;ZHANG Liming;YU Hang(College of Nuclear Science and Technology,Naval Univ.of Engineering,Wuhan 430033,China)

机构地区：[1]海军工程大学核科学技术学院,武汉430033

出　　处：《海军工程大学学报》2023年第2期96-101,共6页Journal of Naval University of Engineering

摘　　要：针对多阶段任务系统各部件及不同阶段间的相关性,提出了一种多阶段任务系统维修优化的方法。该方法采用4种具有不同有效性测度的维修策略,使得在维修活动尽可能少的情况下获得了较高的任务可靠性;利用动态贝叶斯网络来表征多阶段任务系统各部件及不同阶段间的相关性,在使用隐马尔可夫模型对性能退化部件进行量化的同时,基于观测数据实现了对部件的状态及退化速率的同步更新。Aiming at the components and phases dependencies in mission systems(PMSs),a general method to the PMSs'maintenance optimization was proposed,which used four methods having different efficiency measures to make the mission reliability higher with fewer maintenance actions.By using dynamic Bayesian network model,the dependences between the PMS and its components as well as the phases dependence of each component across consecutive phases was characterized.And the components'states and degradation rates were both updated based on the observation data when using hidden Markov model method to quantify the degraded components.

关 键 词：多阶段任务系统 维修优化 动态贝叶斯网络 隐马尔可夫模型 

分 类 号：TL364[核科学技术—核技术及应用]