基于自组织映射与受限玻尔兹曼机的滚动轴承健康评估

发布时间：2018-01-02 19:39

  本文关键词：基于自组织映射与受限玻尔兹曼机的滚动轴承健康评估　出处：《机械传动》2017年06期 　论文类型：期刊论文

  更多相关文章： 序列前向算法 自组织映射 受限玻尔兹曼机 健康评估 滚动轴承

【摘要】：为了对滚动轴承进行动态健康评估,准确描述其性能退化的动态过程,采用自组织映射(SOM)与受限玻尔兹曼机(RBM)相结合的方法进行轴承健康评估。考虑轴承健康状态的变化引起响应特征的相应变化,利用SOM的无监督学习特点,通过序列前向排序算法(SFS)筛选时域、频域和时频域特征,进而建立最优特征域,获得特征向量与轴承健康状态间的映射关系。为了避免传统神经网络在处理上述高维特征数据时出现的易陷入局部最优、参数调整困难、训练时间过长问题,将映射后的特征向量与轴承健康状态分别作为RBM的输入与输出,建立健康评估模型。试验数据分析的结果表明,所提方法可准确识别滚动轴承性能退化过程中的不同健康状态,对于滚动轴承健康评估具有较好的工程适用性。
[Abstract]:For the rolling bearing dynamic health assessment, to accurately describe the dynamic process of the performance degradation of the self-organizing map (SOM) and restricted Boltzmann machine (RBM) method combined with the bearing health assessment. Considering the changes caused by the corresponding response characteristics change of the health status of the bearing, unsupervised learning using SOM algorithm to. Through the sorting sequence (SFS) screening in time domain, frequency domain and time-frequency domain features, and then establish the best feature domain, obtain mapping between the feature vector and the bearing health status. In order to avoid easy to fall into local optimum traditional neural network in dealing with the high dimensional feature data, parameter adjustment difficulties, long training time. The feature vector and the bearing health status after mapping were used as input and output of RBM, the establishment of health assessment model. The experimental data analysis results show that the proposed method can accurately It can identify the different health state in the process of the performance degradation of the rolling bearing, and it has good engineering applicability for the health evaluation of rolling bearings.

【作者单位】： 上海大学机械自动化工程系;
【基金】：国家自然科学基金(50876057) 上海高校青年教师培养资助计划(14203) 上海大学(理工类)创新基金(13007)
【分类号】：TH133.33
【正文快照】： 0引言大型风电、盾构机、高速轧钢机等高端装备的工作特点使其实现功能输出的核心部件——转子-轴承系统承受了比以往更为严苛的工作载荷,因零部件损伤而引起的整机失效时有发生[1]。轴承是机械设备的“关节”,由于受到复杂载荷作用,往往成为设备可靠性的薄弱环节[2]。因此,针，

本文编号：1370675