【作者】 王宏；

【导师】 李朝晖； 毕亚雄；

【作者基本信息】 华中科技大学 ， 水利水电工程， 2009， 博士

【摘要】 继控制自动化、管理信息化与办公自动化后，设备维护信息化与自动化已成为水电站建设和改造的焦点。自20世纪90年代以来，一些大型电厂陆续开展“状态检修”系统建设，在机组状态监测、分析及诊断方面取得了较大的进展，但由于认识、技术和管理多方面的原因，没有达到预定的目标。为了明确水电站维护领域研究、开发和建设的最终目标，加速维护现代化，提出了水电站维护自动化的概念。首先对维护自动化进行了定义，即由维护系统自动地完成各种维护行为。这些行为包括自动状态监测、自动维护决策和自动维护实施。给出了系统运行模型，模型清楚地描述了系统运动状态和设备健康状况之间的关系。控制自动化就是根据我们的愿望自动地控制系统的状态，使系统优质地完成指定的任务；而维护自动化的目的是自动地维护设备，使设备在健康状态下运行。此外还对维护系统、控制系统和管理系统的集成进行了描述。根据系统构成、行为描述及评价关系模型，制订了一套水电设施自动监测、分析与诊断的方法：(1)运动状态监测与分析。在控制系统的基础上，补充监测设施的运行状态，并采用越限判断、梯度计算、关联分析等方法，捕捉设施健康状态异常信息。(2)设施性能评价。根据完成规定任务过程中的运行状态变化曲线，计算设施完成任务的性能指标，并通过与标准和规程中的性能要求对比，以及与历史上同样条件下完成同样任务质量的对比，检测设施功能的失效和健康状态的降低。(3)故障树分析。以在运行状态分析和设施性能评价中检测出来的异常现象为线索，根据故障树模型，查找并列出嫌疑设备清单。(4)故障或降级诊断。采用模型参考故障诊断、签名分析等方法，对故障和健康状况降低进行诊断定位。在掌握设备健康状况的基础上，综合水情、电力市场、物资供应采购与仓储、检修人力与物力、检修场地等信息，考虑水利与航运调度等约束条件，以可靠性为中心，以经济效益最优为目标，做出检修决策，包括：维修设备、维修等级、维修时间。目前，虽然维护实施还需要由人工来完成，但实施方案和工艺的设计可以由计算机专家系统根据检修规程自动地完成，在数字机组上实施和优化，并采用三维可视化仿真等虚拟现实手段培训、指导检修人员。同时，长江电力已开展水轮机叶片打磨与补焊机器人的研究，可以预计将来在虚拟机组上设计好的检修实施方案可以直接输出给机器人，由机器人实施。中国长江电力股份有限公司与华中科技大学合作，自2001年开始开展最优维护信息系统研究，目前已在葛洲坝电站建立了10台机组和4台变压器的健康状况监测系统和远程监测与诊断系统，并采用三维可视化仿真技术建立了数字机组。该系统在葛洲坝电站的设备分析、故障预警与检修中发挥了重要作用。为实现维护自动化迈出了坚实的一步。更多还原

【Abstract】 Condition Based Maintenance (CBM) becomes the focus area after controlautomation and office automation in hydro power plants.Remarkable progress has beenachieved in condition monitoring.However,CBM is only one of the advancedmaintenance strategies.In order to clarify the goal of research and accelerate the modernization inmaintenance domain,maintenance automation is proposed.Maintenance automation is firstly defined as that maintenance system is developed toautomatically carry out maintenance activities for human being.These activities includeautomatic condition monitoring,automatic maintenance decisions and automaticmaintenance actions.A system operation model is given.It clearly illustrates the relationships betweensystem operating states and facility health conditions.The goal of maintenance automationis to automatically maintain facilities in good conditions as well as that of controlautomation is to automatically regulate system states to fellow the expected orbit and tocomplete the appointed tasks.The integration of maintenance system with control systemand management system is described.A condition monitoring secheme is proposed on base of the diagram of systemconstitution and its behavior descriptions:-Operating states measurement and analysis.The operating states are integratedlymeasured.They are compared with their limits.An abnormal event is recorded when thestate(s) exceed the limits.-Performance evaluation.The operation space of a turbine-generator set is dividedto operating sections such as stand-still,start-up,generation with steady load,etc.Theturbine generator set and its equipments accomplish specific task(s) in each operatingsection.The operating section is automatically identified.The performances of tasksaccomplishment are evaluated at the end of each operating section.The performanceindices are compared with the specifications in standards to detect unacceptable event(s).The variation trend of each performance index under similar operating section is made assoon as a new index is figured out or periodicly to detect performance degradationevent(s). -Fault tree analysis.All the possible causes of any abnormal event are searchedand listed on base of fault tree.-Fault or degradation diagnosis.The model based fault detection and signatureanalysis approaches are employed to analyse the suspicious equipments in order to locatethe fault or degradation.Although the maintenance has to be carried out by human being at this moment,theplan and the technics of maintenance can be concepted and optimized on the virtualmachine realized by use of 3D simulation techniques.The conception results can bedemonstrated on the virtual machine to direct the maintenance actions.Furthermore,China Yantze Power Cooperation Limited (CYPC) has started the research on the robotsfor turbine repair.So the conception results will be directly output to robots andaccomplished by robots in the future to realize complete maintenance automation.A joint study on optimal maintenance information system has been started betweenCYPC and Huazhong University of Science and Technology since 2001.The on-linecondition monitoring systems of 10 turbine generator sets and 4 transformers,the remotemonitoring and diagnosis system,and a digital emulator of turbine generator sets havebeen in service in Gezhouba Hydro Power Plant.These systems play more and moreimportant role in the analysis,diagnosis and maintenance in Gezhouba.Their successfulapplication makes significant contributions to maintenance automation。更多还原