基于专家系统的诊断技术在电力系统状态监测中的应用研究-电力系统及其自动化专业论文.docxVIP

=={==≈；=；={==自==；=；；=∞；#自==≈== =={==≈；=；={==自==；=；；=∞；#自==≈==华中科技大学硕士学位论文 摘要 f随着电力系统中竞争的日益激烈，减少电力生产成本也显得越来越重要，而 实行状态检修就是一种十分有力的手段。所谓状态检修又叫预知维修，它是从预 防性检修发展而来的更高层次的检修体制，它是一种以设备的状态为基础、以预 测设备状态发展为依据的检修方式。它可以大大的节约系统维修的费用，能产生 巨大的经济效益。这种维修体制，由于科学的提高设备的可用率和明确了维修目 标，就为设备安全、稳定、长周期、全性能、优质运行提供了可靠的技术和管理 保障。 而实现状态检修，需要综合一系列的专门知识，才能达到它的预期目的。一 般说来，进行状态检修，除了搞好现场人员的相应知识的培训工作以外，状态监 测与诊断是关键因素之一。f雄论文旨在研究设计开发适应于状态检修为目的的专 家系统。在综合多种不精确推理方法的基础之上，根据电力系统实际，提出了一 种新的隶属度来表示事实成立的不精确性，以模糊理论进行不确定传播计算的新 方法。经过变压器的三比值法对该系统的测试，表明该系统运行结果准确。同时 该系统应用面向对象的方法进行设计，综合列用Oracle和Access数据库，因而灵 活性大，易于维护和升级。 关链词：状态检修÷状态监测}电力系统i专家系统，《疆蚤囊多，4 华中科技大学硕士学位论文==#；=；==z==#=≈；=％；；=目；≈≈=目目一； 华中科技大学硕士学位论文 ==#；=；==z==#=≈；=％；；=目；≈≈=目目一； Abstract With the drastic competition in power industry,it is very important to decrease the cost of power production and Condition-based Maintenance is a useful way to fulfill this task．Condj tion-Based Maintenance which is also named predictive maintenance is a super layer overhaul mode．It is based On the state of equipments and predicting the state of equipments．This maintenance mode can save a lot of charge and bring much economic benefit．With increasing the USe time of equipments and clarifying the goal of Maintenance，Condition-Based Maintenance will offer reliable technique and manage safeguard for equipments secure，steady,long period,full performance and exceUent running． It will take a series of professional knowledge for achieving Condition-Based Maintenance．In general，Condition monitoring and diagnosis are妇po#4mt factors for Condition-Based Maintenance，except the work for the training of spot nmning stu任s． This thesis is concerned with studying and designing an expert system for condition-based maintenance．This paper inmgrates several uncertain inference ways in the expert systems锄d proposes a new approach that we c锄use reliability to denote the uncertain of the fact and to calculate the spread of unce嘣nty with fuzzy logic．The proposed approach Was verified via the test ofthree ratio me