JOURNAL ARTICLE

Path Graph Attention Network-based Bearing Remaining Useful Life Prediction Method

Yang ChaoyingJie LiuKaibo Zhou

Year: 2023 Journal:   Journal of Mechanical Engineering Vol: 59 (12)Pages: 195-195

Abstract

摘要: 图数据构建质量直接关系图数据驱动的轴承剩余寿命预测性能。目前传统方法通常利用不同时刻的多传感数据来构建时空图,来表征监测对象性能状态,但如何在单传感监测应用场景下构建表征轴承性能退化状态的图数据并保证其质量仍是一个开放问题。面向单传感监测应用场景,提出一种基于路图注意力网络的轴承剩余寿命预测方法。首先,计算轴承全生命周期时序振动信号的时域统计特征并构造路图,其中,路图中的边用于连接相邻时刻振动信号;在此基础上,设计一种图注意力长短时记忆网络,用于挖掘路图的图特征(节点、边连接)中隐含的时序振动信号特征和时间依赖关系,从而深层次地反映轴承全寿命退化过程。在轴承全寿命公开数据集上开展验证对比试验,结果表明,该路图构造方式明确了边连接的物理意义,并提高了图数据表征性能;所提出的预测方法能有效捕获表征轴承退化状态的图特征以及时间依赖关系,为解决单传感监测应用场景下的轴承性能退化预测问题提供借鉴。

Keywords:
Computer science Path (computing) Graph Bearing (navigation) Artificial intelligence Theoretical computer science Computer network

Metrics

2
Cited By
0.50
FWCI (Field Weighted Citation Impact)
11
Refs
0.64
Citation Normalized Percentile
Is in top 1%
Is in top 10%

Citation History

Topics

Industrial Technology and Control Systems
Physical Sciences →  Engineering →  Control and Systems Engineering
Advanced Decision-Making Techniques
Physical Sciences →  Computer Science →  Information Systems
Evaluation Methods in Various Fields
Physical Sciences →  Environmental Science →  Ecological Modeling

Related Documents

© 2026 ScienceGate Book Chapters — All rights reserved.