基于云平台的风电机组轴承的故障诊断研究

发布时间：2018-06-14 13:48

本文选题：风电机组轴承 + 故障诊断　；参考：《新疆大学》2017年硕士论文

【摘要】：随着全球风力发电行业快速发展,风电机组运维和故障诊断市场需求逐步增加,而风电机组轴承作为风电机组关键部件之一,其正常、稳定运行直接影响着风电机组能量转化率和机组其他部件健康状态。振动监测是目前轴承状态监测故障诊断的常用方法,风电机组数量多、振动测点多、采样率高造成数据量非常大,达到PB甚至TB,给数据传输、分析和诊断提出了挑战。随着互联网快速发展,各种大数据、云计算分析和处理新方法、新技术出现,大数据分析主要基于小数据的探索。因此,本文提出了集成经验模态分解与峭度系数和相关系数的关联度提取方法,通过时域参数、AR模型参数、能量熵参数提取了轴承故障和正常轴承之间的特征值矩阵,将特征值输入径向基核函数的支持向量机,训练故障严重程度的诊断模型,通过实验室轴承数据和风电机组实际运行轴承数据,验证了模型故障识别的准确率。通过对部分数据探索和研究,提出了风电机组轴承故障诊断云端化,运用亚马逊提供的AWS云计算平台,搭建基于AWS的风电机组轴承故障诊断研究平台,将采用的分析方法向云端进行算法的并行化和迁移,主要通过Python开发语言实现了多风电场多台机组振动信号实时信号采集、传输和处理,同时,针对每台风电机组轴承振动信号历史数据定期进行批处理,将其故障诊断和识别模型迭代和更新,实现了风机主轴承故障诊断专业化和定制化,验证和实现了风电机组故障诊断与云计算技术结合,对风电机组运维和故障诊断等领域具有较强的指导意义和参考价值。
[Abstract]:With the rapid development of the global wind power industry, the market demand for wind turbine operation and fault diagnosis is gradually increasing. As one of the key components of wind turbine, the bearing of wind turbine is normal. Stable operation directly affects the energy conversion rate of wind turbine and the health state of other components of wind turbine. Vibration monitoring is a commonly used method for fault diagnosis of bearing condition monitoring at present. The large amount of data caused by the large number of wind turbine units, the large number of vibration measuring points and the high sampling rate lead to the achievement of PB or even TB, which poses a challenge to data transmission, analysis and diagnosis. With the rapid development of the Internet, all kinds of big data, cloud computing analysis and processing new methods, new technologies appear, big data analysis is mainly based on the exploration of small data. Therefore, this paper presents a method of extracting correlation degree by integrating empirical mode decomposition with kurtosis coefficient and correlation coefficient. The eigenvalue matrix between bearing fault and normal bearing is extracted by time domain parameter AR model parameter and energy entropy parameter. The eigenvalue is input into the support vector machine of radial basis function and the diagnosis model of fault severity is trained. The accuracy of fault identification of the model is verified by the laboratory bearing data and the actual running bearing data of wind turbine. Through the exploration and research of some data, the paper puts forward the cloud diagnosis of wind turbine bearing fault, and builds the research platform of wind turbine bearing fault diagnosis based on AWS by using the AWS cloud computing platform provided by Amazon. The algorithm is parallelized and migrated to the cloud, and the real-time signal acquisition, transmission and processing of vibration signal of multi-wind farm and multi-unit are realized by Python development language, at the same time, According to the historical data of bearing vibration signal of each wind turbine unit, batch processing is carried out periodically, and the fault diagnosis and identification model is iterated and updated to realize the specialization and customization of fault diagnosis of main bearing of fan. The combination of wind turbine fault diagnosis and cloud computing technology is verified and realized, which has strong guiding significance and reference value for wind turbine operation and fault diagnosis.
【学位授予单位】：新疆大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM315

【参考文献】