中文说明：
为了精准、稳定地提取滚动轴承故障特征，提出了基于变分模态分解和奇异值分解的特征提取方法，采用标准模糊C均值聚类(fuzzy C means clustering, FCM)进行故障识 别。对同一负荷下的已知故障信号进行变分模态分解，利用 奇异值分解技术进一步提取各模态特征，通过FCM形成标准聚类中心，采用海明贴近度对测试样本进行分类，并通过计算分类系数和“卜均模糊嫡对分类性能进行评价，将该方法 应用于滚动轴承变负荷故障诊断。通过与基于经验模态分解的特征提取方法对比，该方法对标准FCM初始化条件小敏感，在同负荷故障诊断中表现出更好的分类性能 变负荷故障诊断时，除外圈故障特征线发生明显迁移，其他测试样本故障特征线仍在原聚类中心附近，整体故障识别率保持在100 ，因此，该方法能精确、稳定提取故障特征，为实际滚动轴承智能故障诊断提供参考。
English Description：
For precise and steady rolling bearing fault feature extracting, was proposed based on variational mode decomposition and feature extraction method of singular value decomposition, the standard fuzzy c-means clustering (fuzzy C means clustering, FCM) fault _ don't.Of known fault signal under the same load variation mode decomposition, using the singular value decomposition (SVD) _ technology further to extract the modal characteristics of standard are formed by FCM clustering center, classifying test sample using hamming approach degree, and classification by calculation coefficient and "bligh are fuzzy office to evaluate the performance of classification, _ the method was applied to variable load rolling bearings fault diagnosis.With feature extraction method based on empirical mode decomposition, this method is sensitive to small standard FCM initialization conditions, with the load in the fault diagnosis show better classification performance variable load fault diagnosis, with the exception of circle line apparent migration, the fault features, the other test sample near the fault characteristic line is still in the original clustering center, keep the whole fault recognition rate in 100, therefore, this method can precisely and stability to extract fault features, provide reference for the actual rolling bearing intelligent fault diagnosis.