基于机会策略的复杂系统视情维修决策模型  

Condition-based Maintenance Model of Deteriorating Complex System Based on Opportunistic Policy

作　　者：程志君[1] 杨征[1] 谭林[1] 

机构地区：[1]国防科学技术大学信息系统与管理学院,长沙410073

出　　处：《机械工程学报》2012年第6期168-174,共7页Journal of Mechanical Engineering

基　　金：国家自然科学基金资助项目(60904002)

摘　　要：针对部件间存在经济相关性的复杂系统,提出一类基于机会策略的视情维修优化模型。模型从部件本身的失效规律分析出发,确定不同失效模式下部件的维修方式及对应的预防性维修阈值。对于整个复杂系统的维修决策问题,借鉴机会维修思想,通过增加机会维修阈值,将时间和状态两类不同度量空间上的维修控制参数组合起来,解决系统层事后维修与视情维修的综合优化问题。同时利用更新过程理论,提出一类基于马氏决策过程的值迭代算法,确定系统的最优检测时间间隔与机会维修状态阈值。以某机械设备为例对机会维修决策模型和值迭代算法的有效性进行了验证。通过对比分析发现,应用模型能够降低系统的运行维护费用,如果部件级失效率之比在适当范围内,其经济相关性越强,节省费用的效率越高。To investigate the optimal policy of complex system with economical dependence,a kind of condition-based maintenance model is presented based on opportunistic policy.The maintenance mode and corresponding control threshold are determined by the analysis of failure rules of system units.Then the opportunistic maintenance threshold is proposed to unify two kinds of control parameters-time and state-on the system level,which are laid on the different measurement spaces.Therefore the optimization problem of system is solved through combining the corrective maintenance and on-condition maintenance.According to the theory of renewal process and Markov decision process,the value iteration algorithm is given to determine the optimal inspection interval and the opportunistic threshold of system,which achieve the minimization of long-run average cost rate of the whole system.Finally,a kind of mechanical equipment is presented as an example to illustrate the implementation and effectiveness of the model and iteration algorithm.Simultaneously,the applicable area of this model is qualified by comparing the trend curve of system average cost rate as the value of parameters change.The results show that the stronger economic dependence is,the more effective in reduction of system maintenance cost the decision model is,if proportion of units＇ failure rate is in proper range.

关 键 词：视情维修 机会策略 劣化系统 马氏决策过程 

分 类 号：N945[自然科学总论—系统科学] O213[理学—概率论与数理统计]