中文摘要：

为解决现有输变电设备检修决策很大程度依赖于决策人员经验的问题,应用Markov决策过程为检修决策提供定量、具体的决策依据。首先通过Markov过程求解设备的状态转移概率;然后运用策略迭代法对模型最优决策进行求解;最后将数据代入检修决策模型中,改变故障损失得到决策结果的变化。比较决策变化对应的设备故障损失和设备稳态概率可得知：所建立的基于状态的检修（CBM）模型适用于安装了在线监测装置的设备,而基于检测的检修（IBM）模型适用于无在线监测装置的设备;最优决策从维修成本最小的决策过渡到最后每次决策时都选择大修,整个变化过程证明了应用Markov决策可以折中维修成本和故障损失得到经济最优决策。研究结果可为检修决策人员提供定量的检修计划。

英文摘要：

The existing maintenance decision process depends mostly on the experiences of maintenance staff. In order to improve this condition, the Markov decision process was applied to provide quantitative decision basis. First, the state transition probability was obtained using the Markov process, then, the optimum maintenance policy was put forward us- ing the policy iteration method, finally, the condition data of power delivery equipment were applied to the model, and the failure cost was changed to obtain the optimum policy variation. By comparing the failure cost with different decision and the steady state possibility, it can be concluded that condition based maintenance （CBM） model is suitable for equipment with online monitoring device and inspection based maintenance （IBM） model for equipment in absence of online moni- toring device. The optimum decision changes from the one with minimum maintenance cost to major maintenance every time, which proves that the proposed model can compromise the maintenance and failure cost to obtain the optimum maintenance decision. The research results can provide the maintenance staff with quantitative maintenance plan.