【作者】 欧健；

【导师】 孙才新；

【作者基本信息】 重庆大学 ， 电气工程， 2004， 博士

【摘要】 随着国民经济的发展,我国大型汽轮发电机组的单机容量日益增大,这对机组可用率、运行效率、安全性、可靠性与经济性提出了更高的要求。因此,研究汽轮发电机组振动故障诊断技术对其安全稳定运行有着十分重要的理论意义和巨大的经济效益。本文在总结和借鉴前人有关汽轮发电机组状态监测和故障诊断研究的基础上,重点研究了基于模糊集、人工神经网络、粗糙集、遗传算法以及它们相互结合机制的诊断理论和方法,用已有的汽轮发电机组振动故障实例进行了验证,得出了一些具有实用价值的结论,并提出了可靠、实用的大型汽轮发电机组振动故障诊断的新方法。本文主要内容有:1阐述了本课题研究的目的和意义,对汽轮发电机组振动故障的特点、机组在线监测和故障诊断技术的研究现状和方法以及模糊集、神经网络、粗糙集和遗传算法的发展及其在机组振动故障诊断中的应用进行了全面的综述。2分析和研究了汽轮发电机振动故障的电气、机械和热力机理,提出采用模糊集理论对征兆数据进行规范化处理的方法,为进一步的智能诊断技术研究奠定基础。3深入分析BP神经网络在理论和方法上存在的缺陷,提出采用遗传算法优化初始网络权重,将遗传算法与BP神经网络有机地结合起来,迅速得到BP神经网络最佳的初始权值矩阵,并成功地运用于汽轮发电机组振动故障诊断系统。4通过对粗糙集理论的深入分析,提出改善传统粗糙集约简抗干扰属性影响差的方法和基于遗传算法的粗糙集约简算法。仿真结果表明,该方法提高了粗糙集约简提取诊断规则的准确性和效率,能较好地满足诊断规则提取的技术要求,并成功地运用于汽轮发电机组振动故障诊断系统。5针对大型汽轮发电机组振动故障的复杂性、非线性、影响因素多等特点,首次提出基于粗糙集理论的诊断规则分层发现方法,在诊断实例中取得了较好的效果。6提出了基于粗糙集约简理论的汽轮发电机组振动故障诊断模糊神经网络,有效地压缩了神经网络的输入空间,经实例证明该方法能很好地对各层次的振动故障进行诊断。7综合本文的研究成果,将模糊集、神经网络、遗传算法和粗糙集结合起来,首次提出了结合多种智能方法的汽轮发电机组振动故障集成诊断网络,通过故障实例诊断分析,效果良好,具有较强的实用价值。更多还原

【Abstract】 With the development of national economy, the ever-growing capacity of large steam generator-set demands higher usability, security, reliability and economical efficiency. Thus the research to the technology of the vibration fault diagnosis in the stream generator-set is of vial significance to its safe function, and can bring us substantial benefits. Based on the former study of the condition monitor and fault diagnosis on stream generator-set, this dissertation focuses on fuzzy set theory, artificial neural network technology, rough set theory and the genetic algorithms and theories on their integration mechanism. Attested by the cases of vibration fault of the present steam generator-sets, this dissertation draws some feasible conclusions, and brings forth some reliable and practical new methods, which have further enriched and greatly improved the theories on the vibration fault monitor and diagnosis of the large stream generator-set. The main contents of this paper are described as follows:1.The goals and meaning of this research are presented. It is a general survey to the feature of vibration faults of turbine generator-set, the present research and the research methods of on-line monitor and faults diagnosis, the development of Fuzzy Set Theory, Neural Network, Rough Set Theory and Genetic algorithms and their application in the generator-set vibration diagnosis.2.Based on the analysis of the electric, mechanic and thermo theories of steam turbine generator-sets’ vibration faults, the application of the theory to the standardization of symptom data is put forward, which has provided a foundation for the research of intellectual diagnosis.3.On the analysis of the weaknesses of the BPNN theory and methods, genetic algorithms is adopted to optimize the initial weight, and through their organic integration, the optimal weight matrix of BPNN can be obtained, which can be successfully applied to diagnose the vibration fault of steam generator-set.4.On the analysis of rough set theory, the method to improve the traditional rough reduce weakness in avoiding the influence of interrupting attribute and the rough reduce algorithms based on genetic algorithm are brought forward. Suggested by simulation, this method has improved the exactness of technology requirement of diagnosis rules extraction and has been successfully used in vibration fault diagnosis in turbine generator-set. <WP=8>5.In view of that vibration fault of large steam turbine generator-set has such features as complexity, non-linearity, multi influence factors and so on, the gradual discovery of diagnosis rules has been put forward and has been successfully used in actual fault diagnosis.6.The fuzzy neural network of steam turbine generator-set’s vibration fault diagnosis based on rough set theory is brought forward, so as to compress the input dimension effectively, which has been proved to be successful by actual example. 7. Associated with achievements of this dissertation, the fuzzy set theory, neural network, genetic algorithm and rough set theory are integrated. The model of integrated diagnosis network of steam turbine generator-Set’s Vibration Faults is constructed, and analysis on practical fault case is carried out, the results show that the model has a fairly high practical value.更多还原