基于磷虾群算法的SVR滚动轴承性能衰退预测研究

发布时间：2018-11-23 07:38

【摘要】：滚动轴承作为大型旋转机械设备中的关键零部件之一,对保证机械设备正常运行起着举足轻重的作用。因此对其提前进行预知维修具有非常重要的意义。而振动信号特征提取的好坏直接影响着衰退过程的变化趋势及预测精度。如何建立正确的轴承退化评价指标关系到预测结果的准确性。在滚动轴承发生故障之前采取相应的措施,从而避免其事故的发生,对保证设备能够正常运行至关重要。本文在滚动轴承现有的振动信号特征提取方法的基础上,提出了CEEMD和小波包半软阈值相结合的特征提取方法,此方法区别于传统时域、频域、时频域特征。在保证了原始信号的完整性的基础上,对高频振动信号中的噪声进行滤除,并通过试验验证,与其它时频方法的特征提取效果进行了对比,验证了该方法的有效性。针对上述提出的改进的特征提取方法,本文在得到多个特征参数的基础上对高维特征集进行降维处理,针对PCA、KPCA的缺点提出了LLE与模糊C均值相结合的方法,将LLE聚类后再经模糊C均值二次聚类,通过试验验证,对比了轴承内圈不同衰退程度的聚类效果。针对传统支持向量回归机预测精度低的问题,提出了一种基于磷虾群算法的多变量支持向量回归机预测方法,采用了磷虾群觅食原则,选取支持向量回归机中的最优参数C、?的取值,并对遗传算法和磷虾群算法进行试验验证,进而准确预测滚动轴承内圈衰退趋势。最后采用辛辛那提大学滚动轴承全寿命试验数据,应用本文方法对滚动轴承振动信号进行多种方法特征提取,并对滚动轴承内圈不同衰退过程进行阶段性划分,通过三组不同待预测的输入特征,对滚动轴承的全生命周期退化趋势进行预测。证明了本文方法预测精度高且信息更全面,对滚动轴承的性能衰退预测研究具有重要意义。
[Abstract]:As one of the key parts in large rotating machinery, rolling bearing plays an important role in ensuring the normal operation of machinery and equipment. Therefore, it is very important to carry out predictive maintenance in advance. The feature extraction of vibration signal has a direct impact on the trend of decline and prediction accuracy. How to establish the correct evaluation index of bearing degradation is related to the accuracy of prediction results. It is very important to take the corresponding measures before the rolling bearing failure to avoid the accident. Based on the existing vibration signal feature extraction methods of rolling bearings, a new feature extraction method based on CEEMD and wavelet packet semi-soft threshold is proposed in this paper, which is different from the traditional time domain, frequency domain and time-frequency domain. On the basis of ensuring the integrity of the original signal, the noise in the high frequency vibration signal is filtered, and compared with other time-frequency methods, the effectiveness of the method is verified by experiments. In view of the improved feature extraction method proposed above, this paper deals with the dimensionality reduction of the high Vitert collection on the basis of obtaining several feature parameters, and puts forward the method of combining LLE with fuzzy C-means in view of the shortcomings of PCA,KPCA. After LLE clustering and fuzzy C-means quadratic clustering, the clustering effect of bearing inner ring with different degrees of decline is compared by experiments. Aiming at the problem of low prediction accuracy of traditional support vector regression machine, a multivariable support vector regression method based on krill swarm algorithm is proposed. The feeding principle of krill colony is adopted, and the optimal parameter Con in support vector regression machine is selected. The genetic algorithm and the krill swarm algorithm are tested to predict the decline trend of rolling bearing inner ring accurately. Finally, using the data of the rolling bearing life test at the University of Cincinnati, the vibration signals of the rolling bearing are extracted by the method in this paper, and the different decline process of the inner ring of the rolling bearing is divided into stages. The whole life cycle degradation trend of rolling bearings is predicted by three groups of different input features to be predicted. It is proved that this method has high prediction accuracy and more comprehensive information, which is of great significance to the research of rolling bearing performance decline prediction.
【学位授予单位】：哈尔滨理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TH133.33

【相似文献】