论文部分内容阅读
原子发射光谱是分析油液中微小磨损颗粒元素浓度的重要方法。以综合传动全寿命磨损试验不同阶段采集的多个油液样本为研究对象,分别运用基于模糊隶属度的稳健核主成分分析(RKPCA)与传统主成分分析(PCA)对光谱数据进行主成分提取与对比。在剔除光谱数据中的干扰元素后,计算与比较两种方法的主成分数量与贡献率,并利用RKPCA主成分进行综合传动多摩擦副的分类识别;对光谱数据和RKPCA特征值分别进行模糊C均值聚类,对比两种聚类结果应用在磨损状态评价中的效果。研究表明,由于光谱数据离群值与非线性影响,RKPCA较PCA的主成分数量稍小且累积贡献率高,说明前者能更有效地降低变量维数;通过RKPCA主成分与摩擦副组件的相关性分析可以看出,该方法可以精确的实现综合传动多摩擦副、多磨损部位的分类与识别,进而分类评价不同摩擦副的磨损状态;RKPCA特征值的模糊C均值聚类结果与光谱数据直接聚类结果相比,前者能更精确的定位磨损状态转化的临界点,从而准确评价综合传动整体磨损状态。油液光谱RKPCA分析方法的创新在于将特征值变化规律引入整体磨损状态评价,实现整体评价与关键摩擦副的分类评价相结合。这样不仅有助于综合传动大修期的准确判断,还能给出需维修部件建议。该方法也适用于其他复杂机械系统的磨损监测与评价等相关领域。
Atomic emission spectroscopy is an important method for analyzing the concentration of minutely-worn particle elements in the oil. Taking multiple oil samples collected in different stages of the comprehensive life testing of transmission as the research objects, principal component analysis (PCA) and principal components analysis (PCA) Contrast with. After eliminating the interference elements in the spectral data, the principal component quantities and contribution rates of the two methods were calculated and compared. The RKPCA principal components were used to classify the multi-friction pair of the integrated transmission. The spectral data and RKPCA eigenvalues were respectively fuzzy C Mean clustering was used to compare the effect of the two clustering results in evaluating wear status. The results show that due to the outliers and nonlinear effects of spectral data, RKPCA is slightly smaller than PCA and the cumulative contribution rate is high, which shows that the former can reduce the dimensionality of variables more effectively. The correlation between RKPCA principal components and friction pair components It can be seen from the sex analysis that the method can accurately classify and identify the multi-friction pairs and multi-wear parts of the integrated transmission, and then classify and evaluate the wear status of different friction pairs. The fuzzy C-means clustering results of RKPCA eigenvalues are directly correlated with the spectral data Compared with the clustering results, the former can locate the critical point of the wear state transformation more accurately, so as to accurately evaluate the overall wear status of the integrated transmission. The innovation of oil spectrum RKPCA analysis method is to introduce the law of eigenvalues into the overall wear state evaluation, and to achieve the combination of the overall evaluation and the classification and evaluation of the key friction pairs. This not only contributes to the accurate judgment of the comprehensive transmission overhaul period, but also gives advice on the parts to be repaired. The method is also applicable to other complex mechanical systems such as wear monitoring and evaluation of related fields.