列车运行时轴承故障的检测与定位对于列车运行安全与健康维护至关重要.现有的轴承故障报警系统主要是基于单一轴温变量的规则诊断,报警不及时.针对上述问题,本文结合运行于相似环境和速度的同车多轴轴温的相关性及轴温动态性,提出了一种数据驱动的基于多轴轴温动态潜结构的列车轴承故障检测与定位方法.首先,提出基于动态内在典型相关分析（Dynamic-inner canonical correlation analysis, DiCCA）的列车多轴轴温动态潜结构建模方法;其次,利用所建立的模型,提出基于DiCCA综合指标的列车轴承故障检测方法;在此基础上,提出基于DiCCA多向重构的列车轴承故障定位方法.利用某列车实际运行时的轴温数据进行验证,结果表明了所提方法的有效性.
The effective fault detection and diagnosis is necessary for operation safety and maintenance of the trains. The existing bearing alarm system normally applies rule-based method that cannot detect the fault into account before the bearing is heavily damaged. In this paper, taking the correlation and dynamic relation of multi-bearing temperatures, a data-driven dynamic latent structure based train bearing fault detection and diagnosis method is proposed. Firstly, a dynamic-inner canonical correlation analysis (DiCCA) based dynamic latent structure method is applied to extract the cross and auto dynamic relations within multi-dimensional bearing temperatures of the train. Secondly, a DiCCA based combined index is defined for fault detection of dynamic system and applied to detect the operational abnormality of the bearings. Thirdly, a DiCCA based multi-directional reconstruction method is proposed to locate the faulty bearing. Finally, application results using bearing temperature data collected from the practical operation of a train demonstrate the effectiveness of the proposed method.
|Translated title of the contribution||Dynamic Modeling and Reconstruction Based Fault Detection and Location of Train Bearings|
|Original language||Chinese (Simplified)|
|Number of pages||9|
|Journal||Zidonghua Xuebao/Acta Automatica Sinica|
|Publication status||Published - Dec 2019|
Bibliographical note基金资助: 国家自然科学基金 (61490704, 61991401, 61673097, 61573022), 中央高校基本科研业务费 (N180802004, N160801001) <br/>Supported by National Natural Science Foundation of China (61490704, 61991401, 61673097, 61573022) and Fundamental Research Funds for the Central Universities (N180802004, N16080 1001)
- Fault detection, fault location
- fault location
- dynamic latent structrue modeling
- multi-directional reconstruction