摘要:燃气轮机维护费用高,可靠性要求严苛,现行的基于时间的维护策略存在若干缺陷,可以借助以可靠性为中心的维修( Reliability-centered Maintenance, RCM)技术,降低维护成本,保障设备可靠性. RCM基于逻辑决断图选择维护任务,针对性不强,且主观性太强. 提出了一种基于RCM知识库的维护任务逻辑决断方法. 该方法按照功能属性、故障属性、维护属性分析已有故障,形成知识库;对新发现的故障按照功能属性,基于知识库,应用C4 .5算法生成逻辑决断图,根据其故障属性判断合适的维护任务. 该方法被应用于某燃气轮机,对29台该燃气轮机5年的故障数据进行RCM分析,生成RCM知识库,并为预防可燃气体探测器误报警与防冰系统高背离两种故障,提供了解决方法.%Gas turbines with high maintenance costs and stringent reliability requirements are not suitable for current time based mainte-nance strategy.With the help of Reliability-centered Maintenance( RCM) maintenance costs of gas turbines can be reduced and relia-bility of gas turbine can be guaranteed.Maintenance task is selected based on the logical decision graph in RCM, which is inaccurate and subjective.A maintenance task logic decision method based on RCM knowledge base is proposed in this paper.Function character-istics, fault features, and maintenance properties of existing fault are analyzed to build knowledge base.Then the logical decision graph for a new fault is established based on its function characteristic via C4.5 algorithm.Then the maintenance property of this fault is de-termined according to its fault features.The method is applied to a type of gas turbine, whose RCM knowledge base was generated from 5 years of fault data on 29 gas turbines.Solutions for false alarm of combustible gas detectors and anti-ice system high deviation are provided based on this new method.