He David, Li Ruoyu, Zhu Junda, Zade Mikhail
Intelligent Systems Modeling & Development Laboratory, Department of Mechanical and Industrial Engineering, University of Illinois at Chicago, Chicago, IL 60607, USA.
IEEE Trans Neural Netw. 2011 Dec;22(12):2022-31. doi: 10.1109/TNN.2011.2169087. Epub 2011 Oct 10.
Full ceramic bearings are considered the first step toward full ceramic, oil-free engines in the future. No research on full ceramic bearing fault diagnostics using acoustic emission (AE) sensors has been reported. Unlike their steel counterparts, signal processing methods to extract effective AE fault characteristic features and fault diagnostic systems for full ceramic bearings have not been developed. In this paper, a data mining based full ceramic bearing diagnostic system using AE based condition indicators (CIs) is presented. The system utilizes a new signal processing method based on Hilbert Huang transform to extract AE fault features for the computation of CIs. These CIs are used to build a data mining based fault classifier using a k-nearest neighbor algorithm. Seeded fault tests on full ceramic bearing outer race, inner race, balls, and cage are conducted on a bearing diagnostic test rig and AE burst data are collected. The effectiveness of the developed fault diagnostic system is validated using real full ceramic bearing seeded fault test data.
全陶瓷轴承被认为是未来迈向全陶瓷、无油发动机的第一步。目前尚未有关于使用声发射（AE）传感器进行全陶瓷轴承故障诊断的研究报道。与钢制轴承不同，用于提取全陶瓷轴承有效声发射故障特征的信号处理方法以及故障诊断系统尚未得到开发。本文提出了一种基于数据挖掘的全陶瓷轴承诊断系统，该系统使用基于声发射的状态指标（CI）。该系统利用一种基于希尔伯特-黄变换的新信号处理方法来提取声发射故障特征，用于计算状态指标。这些状态指标被用于使用k近邻算法构建基于数据挖掘的故障分类器。在轴承诊断试验台上对全陶瓷轴承的外圈、内圈、滚珠和保持架进行了植入式故障测试，并收集了声发射突发数据。利用实际的全陶瓷轴承植入式故障测试数据验证了所开发故障诊断系统的有效性。
