关键词: 滚动轴承, 剩余使用寿命预测, 时间卷积自编码器, Informer网络, 深度特征提取

Abstract: Rolling bearing is a key component in a large number of rotating machines, and its remaining useful life (RUL) prediction problem is related to whether the equipment can operate safely and stably. In order to solve the current problem of low accuracy of RUL prediction, this paper proposes a method combining temporal convolutional autoencoder (TCAE) and Informer network (TCAE-Informer) in frequency domain for RUL prediction of rolling bearings. The method designs a temporal convolutional auto-encoder, which is oriented to the frequency-domain signals of different time samples of rolling bearings, and adaptively mines the depth features that better reflect the full-life degradation law of rolling bearings. An Informer network model is built, and with the help of its learning advantage in long sequence information, the mapping relationship between the depth features and the RUL of rolling bearings is effectively fitted, and the RUL prediction function of rolling bearings is realized. Using the comparison and validation of XJTU-SY bearing dataset public dataset with the evaluation indexes of the three RUL prediction results, the proposed method is able to achieve more accurate RUL prediction effect under different working conditions compared with many existing methods, which proves that the proposed method has superior RUL prediction ability and generalization ability. The anti-interference test is carried out for different methods, and the proposed method shows better RUL prediction effect under different noise conditions, which proves that the proposed method has superior anti-interference ability of RUL prediction.

Key words: rolling bearing, remaining useful life prediction, temporal convolutional autoencoder, informer network, deep feature extraction