方位(导航)
断层(地质)
稳健性(进化)
振动
计算机科学
状态监测
噪音(视频)
加速度计
工程类
人工智能
模式识别(心理学)
地质学
声学
地震学
基因
操作系统
图像(数学)
电气工程
物理
化学
生物化学
作者
Jin Deng,Han Liu,Hairui Fang,Siyu Shao,Dong Wang,Yimin Hou,Dongsheng Chen,Mingcong Tang
标识
DOI:10.1016/j.ymssp.2023.110253
摘要
With the rapid development of pattern recognition represented by deep learning, the massive excellent bearing fault diagnosis methods have emerged. However, the majority of these reports only focus on the diagnosis of single bearing, while there are few works on fault detection of multi-bearing system. Furthermore, many diagnostic models based on vibration signals need to embed an accelerometer in the base or outer wall of the monitored bearing, which introducing new potential safety hazards, since the original machine structure was destructed. Therefore, with the purpose of not damaging the mechanical structure of the monitored bearing and the goal of promoting the detection efficiency by monitoring multiple bearings, a framework, called MgNet (Multi-granularity Network), based on multi-granularity information fusion was proposed, to complete the fault diagnosis and location of multi-bearing system via the vibration signal of auxiliary bearing. Finally, the effectiveness and superiority of the proposed approach were verified on a fault diagnosis dataset of the actual multi-bearing system, i.e., MgNet with strong robustness can complete the fault diagnosis task of multi-bearing system under the interference of noise signal(Gaussian noise and Laplace noise), and accurately locate the bearing where the fault occurs, which is expected to enrich the application scenarios of fault diagnosis algorithms for rotating machinery and improve the efficiency of fault detection.
科研通智能强力驱动
Strongly Powered by AbleSci AI