【作者】 叶鹏；

【导师】 李波；

【作者基本信息】 电子科技大学 ， 工程硕士（专业学位）， 2020， 硕士

【摘要】 透平机械是冶金、石油、化工等领域的核心装备,一旦发生计划外停机,会造成企业极大的经济损失。对于大型透平机械机组,现有的维护手段一般是根据概率对设备所处的性能区间进行估计,再根据估计的结果对设备的单独部件进行维护,但是这种方法并没有很周密的考虑到设备部件间存在各种相关关系,可能存在一定的偏差,不能满足智能化维护的需求,故提出一种高效、经济的维护方法是很有必要的。本论文以2019年度四川省重大科技专项——“先进制造智能服务”为依托,以大型透平机械为研究对象,针对大型机械现有的维护手段落后及低效的问题,基于预防性维护理论和可靠性理论,结合部件的成组维护与数字孪生技术,对大型设备的多个部件进行成组预防性维护开展研究,主要包括以下几个部分内容:首先针对透平机械历史数据、故障数据少的问题,结合现在工业界流行的数字孪生技术,对设备的部分组件建立数字孪生模型,并根据采集数据对孪生模型的运行数据进行训练使其接近真实情况,得到设备运行的数字孪生模拟数据用于下文分析,同时根据运行数据对主要部件的退化过程进行分析,得到退化程度量化模型。其次基于数字孪生体的运行数据结合退化过程对部件间的相关性进行研究,得到了一种部件间退化相关性的分析方法,基于此进一步分析了部件的独立可靠度和基于相关性的设备可靠度,再结合部件在维护过程中的经济相关性以及结构相关性,建立部件在维护过程中的成本模型,为后文的预防性维护方法研究提供基础。最后根据上文所建立的退化过程、可靠度、以及维护成本为目标,建立多目标预防性维护问题模型,结合协同算法,改进粒子群算法的搜索更新公式对模型进行求解,得到了单个部件的维护窗口等,并进行对比验证说明有效性,同时基于上述求解得到的窗口,采用一种改进聚类半径的聚类算法,结合实例,对采用成组方法进行预防性维护和传统基于概率的单部件预防性维护通过数字孪生体的模拟运行进行了对比,说明本文所提出的方法在实际情况下的适用性,能一定程度上的节省维护费用,提高维护后可用时间,对于透平机械的维护有一定的指导意义和实用价值,对于其他类似大型机械设备有一定的启发意义。更多还原

【Abstract】 Turbine machinery is the core equipment in the fields of metallurgy,petroleum,and chemical industry.Once an unplanned shutdown occurs,it will cause great economic losses to the enterprise.For large turbomachinery units,the existing maintenance methods generally estimate the performance interval of the equipment based on probability,and then maintain the individual components of the equipment based on the estimated results,but this method has not been carefully considered that are various correlations between equipment components,and there may be certain deviations that cannot meet the needs of intelligent maintenance.Therefore,it is necessary to propose an efficient and economical maintenance method.This thesis is based on the major scientific and technological project of Sichuan Province in 2019-"Advanced Manufacturing Intelligent Service",and taken large turbine machinery as the research object.It is based on the preventive maintenance theory for the problems of backward and inefficient maintenance methods of large machinery.And reliability theory,combined with group maintenance of components and digital twin technology,to carry out research on group preventive maintenance of multiple components of large equipment,mainly including the following parts:Firstly,for the problem of low historical data and failure data of turbine machinery,combined with the popular digital twin technology in the industry,a digital twin model is established for some components of the equipment,and the operating data of the twin model is trained according to the collected data to make it close to the real situation,the digital twin simulation data of the equipment operation is used for the following analysis,and the degradation process of the main components is analyzed according to the operation data to obtain a quantitative model of the degradation degree;Secondly,based on the operation data of the digital twin combined with the degradation process,the correlation between the components was studied,and an analysis method of the degradation correlation between the components was obtained.Based on that,the independent reliability of the components and the reliability of the equipment based on the correlation were further analyzed.And then,combined with the economic relevance and structural relevance of the components in the maintenance process,the cost model of the components in the maintenance process is established to provide a basis for the later study of preventive maintenance methods;Finally,based on the degradation process,reliability,and maintenance cost established above,a multi-objective preventive maintenance problem model is established.Combined with the collaborative algorithm,the search and update formula of the particle swarm optimization algorithm is improved to solve the model,and the individual components are obtained.Maintain windows,etc.,and compare and verify the effectiveness.At the same time,based on the window obtained by the above solution,a clustering algorithm with improved cluster radius is used.Combined with examples,preventive maintenance using the group method and traditional probability-based single.The preventive maintenance of components is compared through the simulation operation of digital twins,which shows the applicability of the method proposed in this paper in the actual situation,which can save maintenance costs to a certain extent,improve the available time after maintenance,and maintain the turbine machinery.It has certain guiding significance and practical value,and has certain enlightening significance for other similar large-scale mechanical equipment.更多还原