风机齿轮箱多故障诊断问题研究

发布时间：2019-04-11 17:27

【摘要】：随着风电产业的发展,风力发电机组的稳定安全运行和故障诊断越来越受到科学研究者的注意。齿轮箱是风机传动链的一个重要组成部件,它在运行中会受多种因素影响;齿轮箱一旦发生故障,就可能引发风机传动链的崩溃。因此,齿轮箱的故障诊断研究对于维持风机的正常运行具有重要意义。本文的主要研究内容是风机齿轮箱多故障诊断,为了解决这个问题,文章提出了两种不同的解决方案:1、本文提出了一种新的欠定盲源分离算法来解决齿轮箱多故障诊断问题。该算法将盲源分离问题分解为两个子问题,即源信号数目估计和源信号恢复。源信号数目由经验模态分解(empirical mode decomposition,EMD)、奇异值分解(singular value decomposition,SVD)和 K 均值(K-means)聚类联合算法估计。然后,输入信号通过短时傅立叶变换转换到时-频域。最后,通过模糊C聚类估计混叠矩阵,恢复源信号采用的是最小化l1范数。实验结果清晰地验证了算法在处理齿轮箱非线性多故障问题时的有效性。2、本文的另一种方法为基于支持向量机(support vector machine,SVM)概率估计的多故障诊断方法。该方法对安装在齿轮箱上不同位置的传感器分别建立支持向量机模型。每个模型都会输出样本归属于各个类的概率,最终诊断结果是这些概率的综合。为了提高模型的诊断率,方法引入了总体经验模态分解(ensemble empirical mode decomposition,EEMD)来进行特征提取。该算法的有效性经仿真数据和真实数据验证。
[Abstract]:With the development of wind power industry, more and more scientific researchers pay attention to the stable and safe operation and fault diagnosis of wind turbine. Gear box is an important component of fan transmission chain, it will be affected by many factors in operation, once the gear box failure, it may lead to the failure of fan transmission chain. Therefore, the research on fault diagnosis of gearbox is of great significance for maintaining the normal operation of fan. The main research content of this paper is multi-fault diagnosis of fan gearbox. In order to solve this problem, this paper puts forward two different solutions: 1, In this paper, a new blind source separation algorithm is proposed to solve the problem of multi-fault diagnosis of gearbox. The algorithm decomposes the blind source separation problem into two sub-problems, that is, the estimation of the number of source signals and the restoration of the source signals. The number of source signals is estimated by the combined empirical mode decomposition (empirical mode decomposition,EMD), singular value decomposition (singular value decomposition,SVD) and K-means (K-means) clustering algorithms. Then, the input signal is converted to time-frequency domain by short-time Fourier transform. Finally, the aliasing matrix is estimated by fuzzy C clustering, and the minimum L1 norm is used to recover the source signal. The experimental results clearly verify the effectiveness of the algorithm in dealing with the nonlinear multi-fault problem of gearbox. 2. Another method in this paper is the multi-fault diagnosis method based on support vector machine (support vector machine,SVM) probability estimation. The support vector machine (SVM) models for sensors installed in different locations of gearbox are established by this method. Each model outputs the probability that the sample belongs to each class, and the final diagnosis result is a synthesis of these probabilities. In order to improve the diagnostic rate of the model, the ensemble empirical mode decomposition (ensemble empirical mode decomposition,EEMD) is introduced to extract the features. The validity of the algorithm is verified by simulation data and real data.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM315

【参考文献】