计算机与数字工程2025，Vol.53Issue(4)：1208-1213,6.DOI:10.3969/j.issn.1672-9722.2025.04.048
面向轴承故障检测的小样本学习模型算法研究
Research on Few-shot Learning Model Algorithm for Bearing Fault Detection
张子辉 1王勇1
作者信息
- 1. 广东工业大学计算机学院电子信息系 广州 510006
- 折叠
摘要
Abstract
Fault diagnosis for large rotating machinery has received a lot of attention because of its necessity.In this regard,there are a large number of related literature studies based on deep learning,but the fault samples of one or more fault types are diffi-cult to distinguish and collect under the existing working conditions,so most diagnostic models cannot obtain sufficientand accurate-ly classified datasets,which limits the performance of the models.In recent years,there is a bearing fault diagnosis model based on the model agnostic element learning algorithm to train a fault classifier with better performance on limited data,but the algorithm has shortcomings such as instability and high computational cost,so a diagnostic framework based on improved model-agnostic me-ta-learning emerges.It can be seen from the experimental results that the model achieves higher accuracy than the original model in identifying new fault types.Finally,the experiment of artificial bearing fault data to identify real bearing fault data also proves that the generalization ability and robustness of the model are optimized.关键词
轴承故障诊断/有限数据/改进的模型不可知论元学习算法Key words
bearing fault diagnosis/limited data/improved model-agnostic meta-learning algorithm分类
信息技术与安全科学引用本文复制引用
张子辉,王勇..面向轴承故障检测的小样本学习模型算法研究[J].计算机与数字工程,2025,53(4):1208-1213,6.
