面向风电机组齿轮箱的故障诊断系统研究

发布时间：2018-04-18 02:36

本文选题：风电机组 + 齿轮箱　；参考：《华北电力大学》2012年硕士论文

【摘要】：齿轮箱是风电机组传动系统的重要组成部分,也是风电机组中故障率较高的部件之一。由齿轮箱故障引起的机组停运事件时有发生,由此带来的直接和间接损失也越来越大。由于其安装位置狭小,又位于高空作业,一旦发生故障,维修非常困难。因此对风电机组齿轮箱故障进行快速准确诊断并正确处理具有很大的现实意义和经济价值。本文以风电机组齿轮箱为研究对象,对其故障诊断方法进行了研究。在人工智能技术不断发展,并逐步渗透到机械故障诊断领域的背景下,提出了基于故障树分析法的专家系统故障诊断与基于广义回归神经网络(GRNN, General Regression Neural Network)故障诊断相结合的风电机组齿轮箱故障诊断方法。在对风电机组齿轮箱零部件的失效形式进行整理并剖析其失效机理的基础上,绘制了风电机组齿轮箱故障树；将故障树知识以分级存储方式存储到数据库,构建了知识库；以反向推理方式设计了推理流程,建立了推理机。通过分别在时域和频域内对风电机组齿轮箱正常、齿面磨损、轮齿折断三种典型故障状态下的振动信号进行分析,提取了裕度指标、峭度指标、峰值指标、脉冲指标和功率谱熵五个特征参数作为广义回归神经网络的输入并对网络进行训练,建立了基于广义回归神经网络的风电机组齿轮箱故障状态识别模型。通过使用预留的振动信号对基于广义回归神经网络的故障诊断模型进行检测,发现诊断结果与实际状况具有良好的一致性；通过使用系统的搜索查询功能,维修人员可以根据基于故障树分析法的专家系统故障诊断迅速找到故障原因以及专家级解决方案。实验表明,该系统能够以振动信号为依据对风电机组齿轮箱故障进行准确有效的状态识别；能够对已知故障进行快速诊断并给出专家级解决方案。因此,该系统有利于维修人员实现对风电机组齿轮箱故障的精确、快速诊断与维修。
[Abstract]:Gearbox is an important part of wind turbine transmission system and one of the components with high failure rate.The outage of the unit caused by the gearbox failure occurs from time to time, and the direct and indirect losses are becoming larger and larger.Due to its small installation position and high altitude operation, once the failure, maintenance is very difficult.Therefore, it is of great practical significance and economic value to diagnose the gearbox fault of wind turbine unit quickly and accurately and deal with it correctly.In this paper, the fault diagnosis method of wind turbine gearbox is studied.In the context of the continuous development of artificial intelligence technology and the gradual infiltration into the field of mechanical fault diagnosis,A fault diagnosis method for wind turbine gearbox based on expert system fault diagnosis based on fault tree analysis and fault diagnosis based on generalized regression neural network (General Regression Neural Network) is presented.On the basis of sorting out the failure form of gear box parts of wind turbine unit and analyzing its failure mechanism, the fault tree of wind turbine gear box is drawn, the knowledge of fault tree is stored in database by hierarchical storage, and the knowledge base is constructed.The reasoning process is designed and the inference engine is established by reverse reasoning.By analyzing the vibration signals of wind turbine gearbox in time domain and frequency domain respectively under three typical fault states: normal gear box, tooth surface wear and gear tooth fracture, the margin index, kurtosis index and peak value index are extracted.Five characteristic parameters of impulse index and power spectrum entropy are used as input of generalized regression neural network and trained in the network. A fault state identification model of wind turbine gearbox based on generalized regression neural network is established.By using the reserved vibration signal to detect the fault diagnosis model based on the generalized regression neural network, it is found that the diagnosis results are in good agreement with the actual situation.According to the fault tree analysis, the maintainers can quickly find the fault cause and the expert level solution.The experimental results show that the system can accurately and effectively identify the gearbox faults based on the vibration signal, diagnose the known faults quickly and give an expert level solution.Therefore, the system is beneficial to the maintenance personnel to achieve accurate, rapid diagnosis and maintenance of wind turbine gearbox faults.
【学位授予单位】：华北电力大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TH165.3

【参考文献】