齿轮故障特征参数提取及最佳特征参数选择研究

发布时间：2018-06-07 04:00

本文选题：齿轮 + 特征提取　；参考：《沈阳航空航天大学》2017年硕士论文

【摘要】：在旋转机械设备中,齿轮是一个十分重要的、也是故障率较高的零部件之一。齿轮一旦发生故障就可能会对机械设备造成特别严重的影响,会给经济和安全带来相当严重的损失。所以,及时发现并排除齿轮故障对机械设备的维护来说十分关键。而在机械设备诊断中,最为重要的一个课题是特征提取,其次是最佳特征参数选择问题,提取特征的优劣直接影响着故障监视和诊断的准确率。本文以齿轮为研究对象,主要完成了以下两个方面的研究:1.基于VMD平均能量的齿轮故障特征提取方法。与EMD方法相比,VMD方法有效解决了模态混叠问题且运算效率高。因此,本文提出了一种基于VMD的特征提取方法。该方法首先利用VMD方法对所采集的齿轮振动信号进行分解,然后对分解得到的得到K个模态分量分别求取平均能量,进而得到一个K维的特征向量。利用江苏千鹏公司生产的QPZZ-II故障模拟平台对所提出的方法进行了验证。结果表明,利用本文所提方法提取的特征进行故障诊断时,故障诊断准确率达到100%;2.基于离散度的特征参数选择方法。为了有效地从众多特征参数中挑选出少量最能表征齿轮故障特征的最佳特征参数,本文依据“类间离散度要大,类内离散度要小”的原则,设计了一个基于离散度的最佳特征参数评价指标,旨在为特征参数的选择提供评价依据。为了验证本文提出方法的有效性,首先分别使用EMD分解、小波分解和VMD分解对原始齿轮振动信号进行分解,求取分解后各个分量的时域、信息熵及频域等特征参数;其次,应用本文所提出的评价指标对所提取的各个特征参数计算评价指标;然后,根据所设定的阈值确定了齿轮最佳特征参数;最后,采用欧氏距离最小分类法验证所确定的最佳特征参数的有效性。仿真结果表明,本文所提出的参数评价指标可有效地决策出最佳特征参数。
[Abstract]:In rotating machinery, gear is one of the most important parts with high failure rate. Once gear failure occurs, it may have a serious impact on machinery and equipment, and will bring considerable losses to economy and safety. Therefore, the timely detection and troubleshooting of gear fault is very important for the maintenance of mechanical equipment. In the diagnosis of machinery and equipment, the most important problem is feature extraction, followed by the selection of optimal feature parameters. The quality of feature extraction directly affects the accuracy of fault monitoring and diagnosis. In this paper, gear as the research object, mainly completed the following two aspects of research: 1. Gear fault feature extraction method based on VMD average energy. Compared with the EMD method, the EMD method solves the modal aliasing problem effectively and has high computational efficiency. Therefore, this paper proposes a feature extraction method based on VMD. Firstly, the VMD method is used to decompose the vibration signals of gears, then the average energy of K modal components is obtained, and then a eigenvector of K dimension is obtained. The proposed method is verified by using the QPZZ-II fault simulation platform produced by Jiangsu Qianpeng Company. The results show that the accuracy of fault diagnosis is 100%. The method of feature parameter selection based on dispersion. In order to select a few of the best characteristic parameters which can best characterize the fault characteristics of gears from many characteristic parameters, this paper bases on the principle that the dispersion between classes should be large and the dispersion within classes should be small. An evaluation index of optimal feature parameters based on dispersion is designed in order to provide evaluation basis for the selection of feature parameters. In order to verify the validity of the proposed method, EMD decomposition, wavelet decomposition and VMD decomposition are used to decompose the original gear vibration signal, and the time domain, information entropy and frequency domain characteristic parameters of each component are obtained. Using the evaluation index proposed in this paper to calculate the evaluation index of each characteristic parameter extracted. Then, according to the set threshold value, the optimal characteristic parameters of gear are determined. Finally, Euclidean distance minimum classification method is used to verify the validity of the best characteristic parameters. Simulation results show that the proposed parameter evaluation index can effectively determine the best characteristic parameters.
【学位授予单位】：沈阳航空航天大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TH132.41

【参考文献】