期刊
IEEE-ASME TRANSACTIONS ON MECHATRONICS
卷 28, 期 3, 页码 1488-1499出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TMECH.2022.3227960
关键词
Hardware experimental system; industrial time-series; practical application; self-supervised learning; valve stiction detection
This paper proposes a feature learning approach for industrial time-series data based on self-supervised contrastive learning to address the challenge of the lack of labeled data in using neural networks to build a reliable fault detection model. The approach consists of two components: data transformation and representation learning. The data transformation converts the raw time-series into temporal distance matrices with temporal and spatial information. The representation learning component uses a convolution-based encoder to encode the temporal distance matrices into embedding representations.
Using neural networks to build a reliable fault detection model is an attractive topic in industrial processes but remains challenging due to the lack of labeled data. We propose a feature learning approach for industrial time-series data based on self-supervised contrastive learning to tackle this challenge. The proposed approach consists of two components: data transformation and representation learning. The data transformation converts the raw time-series to temporal distance matrices capable of storing temporal and spatial information. The representation learning component uses a convolution-based encoder to encode the temporal distance matrices to embedding representations. The encoder is trained using a new constraint called multitimescale feature consistent constraint. Finally, a fault detection framework for the valve stiction detection task is developed based on the feature learning method. The proposed framework is evaluated not only on an industrial benchmark dataset but also on a hardware experimental system and real industrial environments.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据