中文摘要：

大多数机械设备有三种运行状态：正常状态、异常状态和故障状态。设备被检测诊断为异常后进行状态维修，而设备故障后进行故障维修。假设状态维修能修复如新，故障维修修复非新。引进单调随机过程描述故障维修修复非新的情况。在状态维修能修复如新故障维修修复非新等假设下，利用概率分析和矢量Markov过程方法研究了状态维修机械设备的可靠性和检测更换策略，导出了设备的可靠性指标与稳态收益率的明显表达式，以及最优检测更换策略的可行性判别准则，最优检测更换策略可用解析方法或数值方法求得。实例研究表明，研究结果可为提高机械设备可靠性、安全性和经济效益提供科学的参考依据。

英文摘要：

Most of mechanical equipments have 3 states： normal, abnormal and failure. When the equipments are checked and diagnosed in abnormal, they will accept predictive maintenance. And if the equipments fail, they will accept failure repair. It is assumed that the predictive maintenance makes equipments ＂as good as new＂, while the failure repair does not makes equipment ＂as good as new＂. Monotone process is introduced to describe the case that failure repair does not makes equipment ＂as good as new＂. Under above assumption and others, reliability and check replacement policy of mechanical equipments are studied by using probability analysis and vector markov process method. The explicit expression of the reliability index and average profit rate （i.e., the long-run average profit per unit time） are derived and the rule of feasibility for the optimal check replacement policy under the largest average profit rate is obtained. The corresponding optimal check replacement policy can be determined analyticarly or numerically The research on the actual example shows that the result can provide some scientific basis for improving the reliability, safety and economic profit of mechanical equipments.