大型风力发电机组齿轮传动系统故障特征提取与识别方法研究
发布时间：2018-11-20 21:14
【摘要】：针对大型风力发电机组齿轮传动系统容易出现故障的特点,对其进行故障诊断中故障特征提取方法和故障识别方法进行了研究。简单介绍了风力发电机组成和工作机理,重点介绍了齿轮传动系统的组成和常见故障,列出了风力发电机组齿轮传动系统故障特征频率的计算公式,介绍了引起齿轮振动的原因,简述了齿轮箱振动信号的特点。利用实验采集的原始振动加速度信号对时域统计指标进行了计算,根据时域统计指标的方差可表示不同状态的离散程度,指出了时域统计指标中可以作为故障特征元素的指标。利用幅值谱和细化谱分析方法对各故障状态下的频域特征进行了分析,说明了各故障状态下信号调制的边频带特点。通过对风力发电机齿轮传动系统故障状态振动信号的时域特征和频域特征分析,帮助我们了解故障特点和故障产生原因,为下一步故障特征提取提供指导和依据。针对经验模态分解(EMD)方法用于齿轮故障诊断的优越性和不足,对集合经验模态分解(EEMD)可以减小模态混叠效应的观点进行了仿真验证,提出了该方法中两个主要参数的确定方法。运用相关系数法对集合经验模态分解得到的内禀模态函数(IMF)分量进行了筛选,计算了筛选后的有意义的IMF分量的能量和占总能量的能量比,构造故障特征向量。根据现有的灰色关联度算法和缺陷提出了改进的灰色相似关联度算法,将改进的灰色相似关联度算法用于风力发电机组齿轮传动系统的故障分类识别,实验验证了其有效性,并与多分类支持向量机方法做了比较,结果证明灰色相似关联度算法的准确性更好,实时性更高。
[Abstract]:In view of the characteristic that the gear transmission system of large wind turbine is prone to failure, the method of fault feature extraction and fault identification in fault diagnosis is studied. This paper briefly introduces the composition and working mechanism of wind turbine, emphasizes on the composition and common faults of gear transmission system, and lists the formula for calculating the characteristic frequency of gear transmission system of wind turbine. The causes of gear vibration are introduced, and the characteristics of gear box vibration signal are briefly described. The time-domain statistical index is calculated by using the original vibration acceleration signal collected by the experiment. According to the variance of the time-domain statistical index, the discrete degree of different states can be expressed, and the time-domain statistical index can be used as the index of fault characteristic element. The frequency domain characteristics of each fault state are analyzed by means of amplitude spectrum and thinning spectrum analysis method, and the edge band characteristics of signal modulation in each fault state are explained. By analyzing the time-domain and frequency-domain characteristics of the vibration signals in the fault state of the gear transmission system of the wind turbine, this paper helps us to understand the fault characteristics and the causes of the faults, and provides guidance and basis for the next step of the fault feature extraction. In view of the advantages and disadvantages of the empirical mode decomposition (EMD) method for gear fault diagnosis, the viewpoint that the set empirical mode decomposition (EEMD) can reduce the modal aliasing effect is verified by simulation. A method for determining two main parameters of this method is presented. The intrinsic mode function (IMF) components obtained from the empirical mode decomposition of the set are screened by the correlation coefficient method. The energy and the energy ratio of the significant IMF component to the total energy are calculated, and the fault eigenvector is constructed. According to the existing grey correlation degree algorithm and the defects, the improved grey similar correlation degree algorithm is proposed. The improved grey similar correlation degree algorithm is applied to the fault classification and identification of the wind turbine gear transmission system, and the effectiveness of the algorithm is verified by experiments. Compared with the multi-classification support vector machine method, the results show that the grey similarity correlation algorithm has better accuracy and higher real-time performance.
【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;TH132.41
[Abstract]:In view of the characteristic that the gear transmission system of large wind turbine is prone to failure, the method of fault feature extraction and fault identification in fault diagnosis is studied. This paper briefly introduces the composition and working mechanism of wind turbine, emphasizes on the composition and common faults of gear transmission system, and lists the formula for calculating the characteristic frequency of gear transmission system of wind turbine. The causes of gear vibration are introduced, and the characteristics of gear box vibration signal are briefly described. The time-domain statistical index is calculated by using the original vibration acceleration signal collected by the experiment. According to the variance of the time-domain statistical index, the discrete degree of different states can be expressed, and the time-domain statistical index can be used as the index of fault characteristic element. The frequency domain characteristics of each fault state are analyzed by means of amplitude spectrum and thinning spectrum analysis method, and the edge band characteristics of signal modulation in each fault state are explained. By analyzing the time-domain and frequency-domain characteristics of the vibration signals in the fault state of the gear transmission system of the wind turbine, this paper helps us to understand the fault characteristics and the causes of the faults, and provides guidance and basis for the next step of the fault feature extraction. In view of the advantages and disadvantages of the empirical mode decomposition (EMD) method for gear fault diagnosis, the viewpoint that the set empirical mode decomposition (EEMD) can reduce the modal aliasing effect is verified by simulation. A method for determining two main parameters of this method is presented. The intrinsic mode function (IMF) components obtained from the empirical mode decomposition of the set are screened by the correlation coefficient method. The energy and the energy ratio of the significant IMF component to the total energy are calculated, and the fault eigenvector is constructed. According to the existing grey correlation degree algorithm and the defects, the improved grey similar correlation degree algorithm is proposed. The improved grey similar correlation degree algorithm is applied to the fault classification and identification of the wind turbine gear transmission system, and the effectiveness of the algorithm is verified by experiments. Compared with the multi-classification support vector machine method, the results show that the grey similarity correlation algorithm has better accuracy and higher real-time performance.
【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;TH132.41
【参考文献】
相关期刊论文 前10条
1 郝旺身;王洪明;董辛e，
本文编号：2346046
本文链接：https://www.wllwen.com/jixiegongchenglunwen/2346046.html
