【作者】 王修彦；

【导师】 王清照；

【作者基本信息】 华北电力大学 ， 热能工程， 2000， 硕士

【摘要】 凝汽系统是火力发电厂的重要组成部分，对整个火电厂的安全经济运行有着重要影响。本文在大量收集国内外相关资料的基础上，首先明确了一些基本概念，提出了相对清洁系数的概念及其计算公式，对凝汽器的运行有一定的指导作用；论述了火电厂凝汽设备的运行特性，特别是凝汽器的几种过渡工况的瞬间特性；分析了凝汽设备的常见故障，重点分析了凝汽器真空恶化的原因及征兆。 由于引起凝汽器故障的原因很多，其征兆有一定的模糊性，判断起来不容易。本文采用比较先进的模糊神经网络来进行凝汽器的故障诊断。在介绍了模糊逻辑理论和神经网络基础后，引出了二者综合的产物—模糊神经网络（FNN）。模糊神经网络既能表示定性知识，又具有强大的自学习能力和数据处理能力，是一种比较先进的故障诊断方法。 利用模糊神经网络进行凝汽器故障诊断时需要先构造隶属函数。本文根据某电厂的运行规程及运行经验构造了进行凝汽器故障诊断所需的全部17个隶属函数，利用这些隶属函数可以对各个输入进行模糊化处理，使之成为量化输入。对模糊BP网络进行训练，可以得到模糊BP网络的知识库结构，在此基础上，对一个凝汽器实际故障进行了诊断，得出了令人满意的结果。最后，还提出了一个凝汽器在线监测与诊断系统的设想。更多还原

【Abstract】 Condensing system is an important part of a coal-fried power plant, and is of great importance for the whole plant’s safe and economic operating. In this paper, with a great deal of foreign and domestic documents as it basis, conception and calculation formula for coefficient of relative cleanness are put forward, which can guide condenser’s operation to some degree; condensing system’s performance, especially its transient response under some transient conditions, is discussed; condensing system’s common faults are analyzed and the cause and signs of condenser’s vacuum falling are emphasized. The judgement of condenser’s fault is not easy for it has many kinds of reasons and its signs are fuzzy to some extent. In this paper, fuzzy neural networks, which is fairly advanced, is applied in condenser’s fault diagnosis. After introducing fuzzy logic theory and neural networks basis, their complex --fuzzy neural networks is dra~ forth. Fuzzy neural networks, which not only can indicate qualitative knowledge, but also has powerful self-study ability and data process ability, is an advanced fault diagnosis method. Subordinate functions are necessary for condenser’s fault diagnosis by fuzzy neural networks. In this papei; according to operating code and experience of a power plant, seventeen subordinate functions, which are enough for condenser’s fault diagnosis, are constructed. By these functions, each input can be fuzzified to quantitative input. Fuzzy BP networks? knowledge base structure is attained by training. Having these as its basis, a condenser’s practical fault is diagnosed, and the result is satisfied. Finally, a project for on-line monitoring and diagnosis of condenser is put forward.更多还原