基于聚类异动搜索的风电机组齿轮箱早期故障识别研究
发布时间：2018-10-16 10:53
【摘要】：由于地处偏僻,工作条件恶劣,风电机组故障频发,而齿轮箱故障率也相对较高,而风电机组的SCADA系统并不能很好的对振动监测参数进行分析,在这种情况下,风电机组的状态监测与故障预警系统对减少维修成本、提高维修效率、节省维修时间有着重要的意义。本文主要研究风电机组传动系统齿轮箱,从齿轮箱振动信号分析出发,研究齿轮箱故障机理,故障征兆,故障原因等,并对振动信号处理的时频分析方法进行深入的研究。 (1)论文首先分析了风电机组齿轮箱的主要结构、振动机理,针对齿轮箱典型的故障模式,分析了这些故障的征兆,故障原因,故障的影响,并结合齿轮、轴、轴承等振动故障特点,给出了风电机组齿轮箱振动监测传感器的选择及安装形式和监测测点的布置方案。 (2)分析了风电机组齿轮箱振动信号的变工况特点,针对变工况对故障诊断带来的困难,基于无量纲幅域参数重复性描述因子、相似性描述因子、跳跃性描述因子能有效的解决齿轮箱变转速造成的齿轮箱振动能量的变化给故障诊断和预警带来的困难,对齿轮箱振动信号进行高维特征空间转化,利用数据挖掘中k均值聚类方法对齿轮箱存在的异常信号和正常信号进行分类,发现风电机组齿轮箱振动信号存在早期故障。 (3)针对齿轮箱振动信号非线性、非平稳的特点,采用了时频分析中的Hilbert-Huang变换方法,通过提取时频熵、内禀模态能量熵特征值来判断齿轮和轴承运行状态,提出基于IMF包络谱来实现风电机组齿轮箱复合故障的诊断,研究对齿轮箱振动信号进行EMD频率族分离,并对IMF分量进行Hilbert包络解调,通过包络谱的分析来故障诊断,实验仿真这种方法的可行性。 (4)初步研究了风电机组状态监测与故障预警系统的系统设计,从系统构成上,给出了传感器选择的要求和意见、信号采集系统参数要求和状态监测系统和故障预警所要包括的故障分析功能,为风电机组状态监测与故障预警系统的系统开发提供了有效的帮助。 文章最后对本论文进行了总结,并对相关技术进行了展望。
[Abstract]:Because of its remote location, poor working conditions, frequent failures of wind turbine units and relatively high failure rate of gearboxes, the SCADA system of wind turbines is not able to analyze the vibration monitoring parameters well, in this case, The condition monitoring and fault warning system of wind turbine is of great significance to reduce maintenance cost, improve maintenance efficiency and save maintenance time. In this paper, the gearbox of the transmission system of wind turbine is studied. From the analysis of the vibration signal of the gearbox, the fault mechanism, fault symptoms and causes of the gearbox are studied. Furthermore, the time-frequency analysis method of vibration signal processing is deeply studied. (1) the main structure and vibration mechanism of the gearbox of wind turbine are analyzed firstly, aiming at the typical fault mode of the gearbox. The symptoms, causes and effects of these faults are analyzed, and the characteristics of vibration faults, such as gears and bearings, are analyzed. The selection and installation of vibration monitoring sensors and the layout of monitoring points are given. (2) the characteristics of vibration signal of wind turbine gearbox are analyzed. In view of the difficulty of fault diagnosis caused by variable working conditions, based on the repeatability description factor and similarity description factor of dimensionless amplitude domain parameters, The jump description factor can effectively solve the difficulty of fault diagnosis and early warning caused by the change of vibration energy of gearbox caused by variable speed of gearbox, and transform the vibration signal of gearbox into high dimensional characteristic space. By using the k-means clustering method in data mining, the abnormal signals and normal signals of gearbox are classified, and it is found that there are early faults in the vibration signals of gearbox of wind turbine. (3) aiming at the nonlinearity of gearbox vibration signal, In order to judge the running state of gear and bearing, the Hilbert-Huang transform method in time-frequency analysis is used to extract the eigenvalues of time-frequency entropy and intrinsic modal energy entropy. Based on the IMF envelope spectrum, this paper presents a method to diagnose the complex fault of the gearbox of wind power unit. The EMD frequency family separation of the vibration signal of the gearbox is studied, and the IMF component is demodulated by the Hilbert envelope demodulation, and the fault diagnosis is made through the analysis of the envelope spectrum. (4) the system design of wind turbine condition monitoring and fault warning system is studied preliminarily, and the requirements and opinions of sensor selection are given from the system composition. The parameter requirements of signal acquisition system and the fault analysis functions of the condition monitoring system and fault warning system provide effective help for the development of the wind turbine condition monitoring and fault warning system. At the end of the paper, the thesis is summarized, and the related technology is prospected.
【学位授予单位】：华北电力大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TM315;TH165.3
本文编号：2274137
[Abstract]:Because of its remote location, poor working conditions, frequent failures of wind turbine units and relatively high failure rate of gearboxes, the SCADA system of wind turbines is not able to analyze the vibration monitoring parameters well, in this case, The condition monitoring and fault warning system of wind turbine is of great significance to reduce maintenance cost, improve maintenance efficiency and save maintenance time. In this paper, the gearbox of the transmission system of wind turbine is studied. From the analysis of the vibration signal of the gearbox, the fault mechanism, fault symptoms and causes of the gearbox are studied. Furthermore, the time-frequency analysis method of vibration signal processing is deeply studied. (1) the main structure and vibration mechanism of the gearbox of wind turbine are analyzed firstly, aiming at the typical fault mode of the gearbox. The symptoms, causes and effects of these faults are analyzed, and the characteristics of vibration faults, such as gears and bearings, are analyzed. The selection and installation of vibration monitoring sensors and the layout of monitoring points are given. (2) the characteristics of vibration signal of wind turbine gearbox are analyzed. In view of the difficulty of fault diagnosis caused by variable working conditions, based on the repeatability description factor and similarity description factor of dimensionless amplitude domain parameters, The jump description factor can effectively solve the difficulty of fault diagnosis and early warning caused by the change of vibration energy of gearbox caused by variable speed of gearbox, and transform the vibration signal of gearbox into high dimensional characteristic space. By using the k-means clustering method in data mining, the abnormal signals and normal signals of gearbox are classified, and it is found that there are early faults in the vibration signals of gearbox of wind turbine. (3) aiming at the nonlinearity of gearbox vibration signal, In order to judge the running state of gear and bearing, the Hilbert-Huang transform method in time-frequency analysis is used to extract the eigenvalues of time-frequency entropy and intrinsic modal energy entropy. Based on the IMF envelope spectrum, this paper presents a method to diagnose the complex fault of the gearbox of wind power unit. The EMD frequency family separation of the vibration signal of the gearbox is studied, and the IMF component is demodulated by the Hilbert envelope demodulation, and the fault diagnosis is made through the analysis of the envelope spectrum. (4) the system design of wind turbine condition monitoring and fault warning system is studied preliminarily, and the requirements and opinions of sensor selection are given from the system composition. The parameter requirements of signal acquisition system and the fault analysis functions of the condition monitoring system and fault warning system provide effective help for the development of the wind turbine condition monitoring and fault warning system. At the end of the paper, the thesis is summarized, and the related technology is prospected.
【学位授予单位】：华北电力大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TM315;TH165.3
【参考文献】
相关期刊论文 前10条
1 高胜利;赵小荣;申强;;风力发电机组齿轮箱故障诊断[J];电力电子技术;2011年10期
2 陈志新;徐金梧;杨德斌;章立军;;基于经验模态分解的混沌噪声背景下弱信号检测与信号提取[J];机械科学与技术;2006年02期
3 靳同红;莫正波;郑德亮;王胜春;;时频分析技术及应用研究[J];机械科学与技术;2009年01期
4 侯敬宏,黄树红,申_",张燕平;基于小波分析的旋转机械振动信号定量特征研究[J];机械工程学报;2004年01期
5 于德介,程军圣,杨宇;Hilbert-Huang变换在齿轮故障诊断中的应用[J];机械工程学报;2005年06期
6 朱瑜;王海洋;;基于小波去噪和EMD的齿轮箱故障诊断研究[J];煤矿机械;2012年04期
7 王凤利,马孝江;基于局域波时频谱的旋转机械早期故障诊断技术[J];农业机械学报;2005年01期
8 彭华东;陈晓清;任明;杨代勇;董明;;风电机组故障智能诊断技术及系统研究[J];电网与清洁能源;2011年02期
9 陈忠,郑时雄;基于经验模式分解(EMD)的齿轮箱齿轮故障诊断技术研究[J];振动工程学报;2003年02期
10 来五星,轩建平,史铁林,杨叔子;Wigner-Ville时频分布研究及其在齿轮故障诊断中的应用[J];振动工程学报;2003年02期
相关硕士学位论文 前1条
1 李虎;大型风电机组振动状态监测系统开发[D];华北电力大学（北京）;2009年
，本文编号：2274137
本文链接：https://www.wllwen.com/kejilunwen/jixiegongcheng/2274137.html
